Merge branch 'master' into data-types-binary-encoding

.gitattributes

@@ -2,3 +2,4 @@ contrib/* linguist-vendored
 *.h linguist-language=C++
 tests/queries/0_stateless/data_json/* binary
 tests/queries/0_stateless/*.reference -crlf
+src/Core/SettingsChangesHistory.cpp merge=union

.yamllint

@@ -13,5 +13,4 @@ rules:
         level: warning
     comments:
         min-spaces-from-content: 1
-    document-start:
+    document-start: disable
-        present: false

CHANGELOG.md

@@ -34,7 +34,7 @@
 * Add `_time` virtual column to file alike storages (s3/file/hdfs/url/azureBlobStorage). [#64947](https://github.com/ClickHouse/ClickHouse/pull/64947) ([Ilya Golshtein](https://github.com/ilejn)).
 * Introduced new functions `base64URLEncode`, `base64URLDecode` and `tryBase64URLDecode`. [#64991](https://github.com/ClickHouse/ClickHouse/pull/64991) ([Mikhail Gorshkov](https://github.com/mgorshkov)).
 * Add new function `editDistanceUTF8`, which calculates the [edit distance](https://en.wikipedia.org/wiki/Edit_distance) between two UTF8 strings. [#65269](https://github.com/ClickHouse/ClickHouse/pull/65269) ([LiuNeng](https://github.com/liuneng1994)).
-* Add `http_response_headers` setting to support custom response headers in custom HTTP handlers. [#63562](https://github.com/ClickHouse/ClickHouse/pull/63562) ([Grigorii](https://github.com/GSokol)).
+* Add `http_response_headers` configuration to support custom response headers in custom HTTP handlers. [#63562](https://github.com/ClickHouse/ClickHouse/pull/63562) ([Grigorii](https://github.com/GSokol)).
 * Added a new table function `loop` to support returning query results in an infinite loop. [#63452](https://github.com/ClickHouse/ClickHouse/pull/63452) ([Sariel](https://github.com/sarielwxm)). This is useful for testing.
 * Introduced two additional columns in the `system.query_log`: `used_privileges` and `missing_privileges`. `used_privileges` is populated with the privileges that were checked during query execution, and `missing_privileges` contains required privileges that are missing. [#64597](https://github.com/ClickHouse/ClickHouse/pull/64597) ([Alexey Katsman](https://github.com/alexkats)).
 * Added a setting `output_format_pretty_display_footer_column_names` which when enabled displays column names at the end of the table for long tables (50 rows by default), with the threshold value for minimum number of rows controlled by `output_format_pretty_display_footer_column_names_min_rows`. [#65144](https://github.com/ClickHouse/ClickHouse/pull/65144) ([Shaun Struwig](https://github.com/Blargian)).

base/base/itoa.cpp

@@ -1,32 +1,3 @@
-// Based on https://github.com/amdn/itoa and combined with our optimizations
-//
-//=== itoa.cpp - Fast integer to ascii conversion                 --*- C++ -*-//
-//
-// The MIT License (MIT)
-// Copyright (c) 2016 Arturo Martin-de-Nicolas
-//
-// Permission is hereby granted, free of charge, to any person obtaining a copy
-// of this software and associated documentation files (the "Software"), to deal
-// in the Software without restriction, including without limitation the rights
-// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-// copies of the Software, and to permit persons to whom the Software is
-// furnished to do so, subject to the following conditions:
-//
-//     The above copyright notice and this permission notice shall be included
-//     in all copies or substantial portions of the Software.
-//
-// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-// SOFTWARE.
-//===----------------------------------------------------------------------===//
-
-#include <cstddef>
-#include <cstdint>
-#include <cstring>
 #include <type_traits>
 #include <base/defines.h>
 #include <base/extended_types.h>
@@ -34,99 +5,15 @@
 
 namespace
 {
-template <typename T>
+ALWAYS_INLINE inline char * outOneDigit(char * p, uint8_t value)
-ALWAYS_INLINE inline constexpr T pow10(size_t x)
-{
-    return x ? 10 * pow10<T>(x - 1) : 1;
-}
-
-// Division by a power of 10 is implemented using a multiplicative inverse.
-// This strength reduction is also done by optimizing compilers, but
-// presently the fastest results are produced by using the values
-// for the multiplication and the shift as given by the algorithm
-// described by Agner Fog in "Optimizing Subroutines in Assembly Language"
-//
-// http://www.agner.org/optimize/optimizing_assembly.pdf
-//
-// "Integer division by a constant (all processors)
-// A floating point number can be divided by a constant by multiplying
-// with the reciprocal. If we want to do the same with integers, we have
-// to scale the reciprocal by 2n and then shift the product to the right
-// by n. There are various algorithms for finding a suitable value of n
-// and compensating for rounding errors. The algorithm described below
-// was invented by Terje Mathisen, Norway, and not published elsewhere."
-
-/// Division by constant is performed by:
-/// 1. Adding 1 if needed;
-/// 2. Multiplying by another constant;
-/// 3. Shifting right by another constant.
-template <typename UInt, bool add_, UInt multiplier_, unsigned shift_>
-struct Division
-{
-    static constexpr bool add{add_};
-    static constexpr UInt multiplier{multiplier_};
-    static constexpr unsigned shift{shift_};
-};
-
-/// Select a type with appropriate number of bytes from the list of types.
-/// First parameter is the number of bytes requested. Then goes a list of types with 1, 2, 4, ... number of bytes.
-/// Example: SelectType<4, uint8_t, uint16_t, uint32_t, uint64_t> will select uint32_t.
-template <size_t N, typename T, typename... Ts>
-struct SelectType
-{
-    using Result = typename SelectType<N / 2, Ts...>::Result;
-};
-
-template <typename T, typename... Ts>
-struct SelectType<1, T, Ts...>
-{
-    using Result = T;
-};
-
-
-/// Division by 10^N where N is the size of the type.
-template <size_t N>
-using DivisionBy10PowN = typename SelectType<
-    N,
-    Division<uint8_t, false, 205U, 11>, /// divide by 10
-    Division<uint16_t, true, 41943U, 22>, /// divide by 100
-    Division<uint32_t, false, 3518437209U, 45>, /// divide by 10000
-    Division<uint64_t, false, 12379400392853802749ULL, 90> /// divide by 100000000
-    >::Result;
-
-template <size_t N>
-using UnsignedOfSize = typename SelectType<N, uint8_t, uint16_t, uint32_t, uint64_t, __uint128_t>::Result;
-
-/// Holds the result of dividing an unsigned N-byte variable by 10^N resulting in
-template <size_t N>
-struct QuotientAndRemainder
-{
-    UnsignedOfSize<N> quotient; // quotient with fewer than 2*N decimal digits
-    UnsignedOfSize<N / 2> remainder; // remainder with at most N decimal digits
-};
-
-template <size_t N>
-QuotientAndRemainder<N> inline split(UnsignedOfSize<N> value)
-{
-    constexpr DivisionBy10PowN<N> division;
-
-    UnsignedOfSize<N> quotient = (division.multiplier * (UnsignedOfSize<2 * N>(value) + division.add)) >> division.shift;
-    UnsignedOfSize<N / 2> remainder = static_cast<UnsignedOfSize<N / 2>>(value - quotient * pow10<UnsignedOfSize<N / 2>>(N));
-
-    return {quotient, remainder};
-}
-
-ALWAYS_INLINE inline char * outDigit(char * p, uint8_t value)
 {
     *p = '0' + value;
-    ++p;
+    return p + 1;
-    return p;
 }
 
 // Using a lookup table to convert binary numbers from 0 to 99
 // into ascii characters as described by Andrei Alexandrescu in
 // https://www.facebook.com/notes/facebook-engineering/three-optimization-tips-for-c/10151361643253920/
-
 const char digits[201] = "00010203040506070809"
                          "10111213141516171819"
                          "20212223242526272829"
@@ -137,7 +24,6 @@ const char digits[201] = "00010203040506070809"
                          "70717273747576777879"
                          "80818283848586878889"
                          "90919293949596979899";
-
 ALWAYS_INLINE inline char * outTwoDigits(char * p, uint8_t value)
 {
     memcpy(p, &digits[value * 2], 2);
@@ -145,153 +31,260 @@ ALWAYS_INLINE inline char * outTwoDigits(char * p, uint8_t value)
     return p;
 }
 
-namespace convert
+namespace jeaiii
 {
-template <typename UInt, size_t N = sizeof(UInt)>
+/*
-char * head(char * p, UInt u);
+    MIT License
-template <typename UInt, size_t N = sizeof(UInt)>
-char * tail(char * p, UInt u);
 
-//===----------------------------------------------------------===//
+    Copyright (c) 2022 James Edward Anhalt III - https://github.com/jeaiii/itoa
-//     head: find most significant digit, skip leading zeros
-//===----------------------------------------------------------===//
 
-// "x" contains quotient and remainder after division by 10^N
+    Permission is hereby granted, free of charge, to any person obtaining a copy
-// quotient is less than 10^N
+    of this software and associated documentation files (the "Software"), to deal
-template <size_t N>
+    in the Software without restriction, including without limitation the rights
-ALWAYS_INLINE inline char * head(char * p, QuotientAndRemainder<N> x)
+    to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+    copies of the Software, and to permit persons to whom the Software is
+    furnished to do so, subject to the following conditions:
+
+    The above copyright notice and this permission notice shall be included in all
+    copies or substantial portions of the Software.
+
+    THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+    IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+    FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+    AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+    LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+    OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+    SOFTWARE.
+*/
+struct pair
 {
-    p = head(p, UnsignedOfSize<N / 2>(x.quotient));
+    char dd[2];
-    p = tail(p, x.remainder);
+    constexpr pair(char c) : dd{c, '\0'} { } /// NOLINT(google-explicit-constructor)
-    return p;
+    constexpr pair(int n) : dd{"0123456789"[n / 10], "0123456789"[n % 10]} { } /// NOLINT(google-explicit-constructor)
-}
+};
 
-// "u" is less than 10^2*N
+constexpr struct
-template <typename UInt, size_t N>
-ALWAYS_INLINE inline char * head(char * p, UInt u)
 {
-    return u < pow10<UnsignedOfSize<N>>(N) ? head(p, UnsignedOfSize<N / 2>(u)) : head<N>(p, split<N>(u));
+    pair dd[100]{
-}
+        0,  1,  2,  3,  4,  5,  6,  7,  8,  9, //
+        10, 11, 12, 13, 14, 15, 16, 17, 18, 19, //
+        20, 21, 22, 23, 24, 25, 26, 27, 28, 29, //
+        30, 31, 32, 33, 34, 35, 36, 37, 38, 39, //
+        40, 41, 42, 43, 44, 45, 46, 47, 48, 49, //
+        50, 51, 52, 53, 54, 55, 56, 57, 58, 59, //
+        60, 61, 62, 63, 64, 65, 66, 67, 68, 69, //
+        70, 71, 72, 73, 74, 75, 76, 77, 78, 79, //
+        80, 81, 82, 83, 84, 85, 86, 87, 88, 89, //
+        90, 91, 92, 93, 94, 95, 96, 97, 98, 99, //
+    };
+    pair fd[100]{
+        '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', //
+        10,  11,  12,  13,  14,  15,  16,  17,  18,  19, //
+        20,  21,  22,  23,  24,  25,  26,  27,  28,  29, //
+        30,  31,  32,  33,  34,  35,  36,  37,  38,  39, //
+        40,  41,  42,  43,  44,  45,  46,  47,  48,  49, //
+        50,  51,  52,  53,  54,  55,  56,  57,  58,  59, //
+        60,  61,  62,  63,  64,  65,  66,  67,  68,  69, //
+        70,  71,  72,  73,  74,  75,  76,  77,  78,  79, //
+        80,  81,  82,  83,  84,  85,  86,  87,  88,  89, //
+        90,  91,  92,  93,  94,  95,  96,  97,  98,  99, //
+    };
+} digits;
 
-// recursion base case, selected when "u" is one byte
+constexpr UInt64 mask24 = (UInt64(1) << 24) - 1;
-template <>
+constexpr UInt64 mask32 = (UInt64(1) << 32) - 1;
-ALWAYS_INLINE inline char * head<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
+constexpr UInt64 mask57 = (UInt64(1) << 57) - 1;
+
+template <bool, class, class F>
+struct _cond
 {
-    return u < 10 ? outDigit(p, u) : outTwoDigits(p, u);
+    using type = F;
-}
+};
-
+template <class T, class F>
-//===----------------------------------------------------------===//
+struct _cond<true, T, F>
-//     tail: produce all digits including leading zeros
-//===----------------------------------------------------------===//
-
-// recursive step, "u" is less than 10^2*N
-template <typename UInt, size_t N>
-ALWAYS_INLINE inline char * tail(char * p, UInt u)
 {
-    QuotientAndRemainder<N> x = split<N>(u);
+    using type = T;
-    p = tail(p, UnsignedOfSize<N / 2>(x.quotient));
+};
-    p = tail(p, x.remainder);
+template <bool B, class T, class F>
-    return p;
+using cond = typename _cond<B, T, F>::type;
-}
 
-// recursion base case, selected when "u" is one byte
+template <class T>
-template <>
+inline ALWAYS_INLINE char * to_text_from_integer(char * b, T i)
-ALWAYS_INLINE inline char * tail<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
 {
-    return outTwoDigits(p, u);
+    constexpr auto q = sizeof(T);
-}
+    using U = cond<q == 1, char8_t, cond<q <= sizeof(UInt16), UInt16, cond<q <= sizeof(UInt32), UInt32, UInt64>>>;
 
-//===----------------------------------------------------------===//
+    // convert bool to int before test with unary + to silence warning if T happens to be bool
-// large values are >= 10^2*N
+    U const n = +i < 0 ? *b++ = '-', U(0) - U(i) : U(i);
-// where x contains quotient and remainder after division by 10^N
-//===----------------------------------------------------------===//
-template <size_t N>
-ALWAYS_INLINE inline char * large(char * p, QuotientAndRemainder<N> x)
-{
-    QuotientAndRemainder<N> y = split<N>(x.quotient);
-    p = head(p, UnsignedOfSize<N / 2>(y.quotient));
-    p = tail(p, y.remainder);
-    p = tail(p, x.remainder);
-    return p;
-}
 
-//===----------------------------------------------------------===//
+    if (n < U(1e2))
-// handle values of "u" that might be >= 10^2*N
+    {
-// where N is the size of "u" in bytes
+        /// This is changed from the original jeaiii implementation
-//===----------------------------------------------------------===//
+        /// For small numbers the extra branch to call outOneDigit() is worth it as it saves some instructions
-template <typename UInt, size_t N = sizeof(UInt)>
+        /// and a memory access (no need to read digits.fd[n])
-ALWAYS_INLINE inline char * uitoa(char * p, UInt u)
+        /// This is not true for pure random numbers, but that's not the common use case of a database
-{
+        /// Original jeaii code
-    if (u < pow10<UnsignedOfSize<N>>(N))
+        //      *reinterpret_cast<pair *>(b) = digits.fd[n];
-        return head(p, UnsignedOfSize<N / 2>(u));
+        //      return n < 10 ? b + 1 : b + 2;
-    QuotientAndRemainder<N> x = split<N>(u);
+        return n < 10 ? outOneDigit(b, n) : outTwoDigits(b, n);
+    }
+    if (n < UInt32(1e6))
+    {
+        if (sizeof(U) == 1 || n < U(1e4))
+        {
+            auto f0 = UInt32(10 * (1 << 24) / 1e3 + 1) * n;
+            *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 24];
+            if constexpr (sizeof(U) == 1)
+                b -= 1;
+            else
+                b -= n < U(1e3);
+            auto f2 = (f0 & mask24) * 100;
+            *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 24];
+            return b + 4;
+        }
+        auto f0 = UInt64(10 * (1ull << 32ull) / 1e5 + 1) * n;
+        *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 32];
+        if constexpr (sizeof(U) == 2)
+            b -= 1;
+        else
+            b -= n < U(1e5);
+        auto f2 = (f0 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
+        auto f4 = (f2 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
+        return b + 6;
+    }
+    if (sizeof(U) == 4 || n < UInt64(1ull << 32ull))
+    {
+        if (n < U(1e8))
+        {
+            auto f0 = UInt64(10 * (1ull << 48ull) / 1e7 + 1) * n >> 16;
+            *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 32];
+            b -= n < U(1e7);
+            auto f2 = (f0 & mask32) * 100;
+            *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
+            auto f4 = (f2 & mask32) * 100;
+            *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
+            auto f6 = (f4 & mask32) * 100;
+            *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 32];
+            return b + 8;
+        }
+        auto f0 = UInt64(10 * (1ull << 57ull) / 1e9 + 1) * n;
+        *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 57];
+        b -= n < UInt32(1e9);
+        auto f2 = (f0 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 57];
+        auto f4 = (f2 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 57];
+        auto f6 = (f4 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 57];
+        auto f8 = (f6 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 8) = digits.dd[f8 >> 57];
+        return b + 10;
+    }
 
-    return u < pow10<UnsignedOfSize<N>>(2 * N) ? head<N>(p, x) : large<N>(p, x);
+    // if we get here U must be UInt64 but some compilers don't know that, so reassign n to a UInt64 to avoid warnings
-}
+    UInt32 z = n % UInt32(1e8);
+    UInt64 u = n / UInt32(1e8);
 
-// selected when "u" is one byte
+    if (u < UInt32(1e2))
-template <>
+    {
-ALWAYS_INLINE inline char * uitoa<UnsignedOfSize<1>, 1>(char * p, UnsignedOfSize<1> u)
+        // u can't be 1 digit (if u < 10 it would have been handled above as a 9 digit 32bit number)
-{
+        *reinterpret_cast<pair *>(b) = digits.dd[u];
-    if (u < 10)
+        b += 2;
-        return outDigit(p, u);
+    }
-    else if (u < 100)
+    else if (u < UInt32(1e6))
-        return outTwoDigits(p, u);
+    {
+        if (u < UInt32(1e4))
+        {
+            auto f0 = UInt32(10 * (1 << 24) / 1e3 + 1) * u;
+            *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 24];
+            b -= u < UInt32(1e3);
+            auto f2 = (f0 & mask24) * 100;
+            *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 24];
+            b += 4;
+        }
+        else
+        {
+            auto f0 = UInt64(10 * (1ull << 32ull) / 1e5 + 1) * u;
+            *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 32];
+            b -= u < UInt32(1e5);
+            auto f2 = (f0 & mask32) * 100;
+            *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
+            auto f4 = (f2 & mask32) * 100;
+            *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
+            b += 6;
+        }
+    }
+    else if (u < UInt32(1e8))
+    {
+        auto f0 = UInt64(10 * (1ull << 48ull) / 1e7 + 1) * u >> 16;
+        *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 32];
+        b -= u < UInt32(1e7);
+        auto f2 = (f0 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
+        auto f4 = (f2 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
+        auto f6 = (f4 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 32];
+        b += 8;
+    }
+    else if (u < UInt64(1ull << 32ull))
+    {
+        auto f0 = UInt64(10 * (1ull << 57ull) / 1e9 + 1) * u;
+        *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 57];
+        b -= u < UInt32(1e9);
+        auto f2 = (f0 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 57];
+        auto f4 = (f2 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 57];
+        auto f6 = (f4 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 57];
+        auto f8 = (f6 & mask57) * 100;
+        *reinterpret_cast<pair *>(b + 8) = digits.dd[f8 >> 57];
+        b += 10;
+    }
     else
     {
-        p = outDigit(p, u / 100);
+        UInt32 y = u % UInt32(1e8);
-        p = outTwoDigits(p, u % 100);
+        u /= UInt32(1e8);
-        return p;
+
+        // u is 2, 3, or 4 digits (if u < 10 it would have been handled above)
+        if (u < UInt32(1e2))
+        {
+            *reinterpret_cast<pair *>(b) = digits.dd[u];
+            b += 2;
+        }
+        else
+        {
+            auto f0 = UInt32(10 * (1 << 24) / 1e3 + 1) * u;
+            *reinterpret_cast<pair *>(b) = digits.fd[f0 >> 24];
+            b -= u < UInt32(1e3);
+            auto f2 = (f0 & mask24) * 100;
+            *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 24];
+            b += 4;
+        }
+        // do 8 digits
+        auto f0 = (UInt64((1ull << 48ull) / 1e6 + 1) * y >> 16) + 1;
+        *reinterpret_cast<pair *>(b) = digits.dd[f0 >> 32];
+        auto f2 = (f0 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
+        auto f4 = (f2 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
+        auto f6 = (f4 & mask32) * 100;
+        *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 32];
+        b += 8;
     }
-}
+    // do 8 digits
-
+    auto f0 = (UInt64((1ull << 48ull) / 1e6 + 1) * z >> 16) + 1;
-//===----------------------------------------------------------===//
+    *reinterpret_cast<pair *>(b) = digits.dd[f0 >> 32];
-//     handle unsigned and signed integral operands
+    auto f2 = (f0 & mask32) * 100;
-//===----------------------------------------------------------===//
+    *reinterpret_cast<pair *>(b + 2) = digits.dd[f2 >> 32];
-
+    auto f4 = (f2 & mask32) * 100;
-// itoa: handle unsigned integral operands (selected by SFINAE)
+    *reinterpret_cast<pair *>(b + 4) = digits.dd[f4 >> 32];
-template <typename U>
+    auto f6 = (f4 & mask32) * 100;
-requires(!std::is_signed_v<U> && std::is_integral_v<U>)
+    *reinterpret_cast<pair *>(b + 6) = digits.dd[f6 >> 32];
-ALWAYS_INLINE inline char * itoa(U u, char * p)
+    return b + 8;
-{
-    return convert::uitoa(p, u);
-}
-
-// itoa: handle signed integral operands (selected by SFINAE)
-template <typename I, size_t N = sizeof(I)>
-requires(std::is_signed_v<I> && std::is_integral_v<I>)
-ALWAYS_INLINE inline char * itoa(I i, char * p)
-{
-    // Need "mask" to be filled with a copy of the sign bit.
-    // If "i" is a negative value, then the result of "operator >>"
-    // is implementation-defined, though usually it is an arithmetic
-    // right shift that replicates the sign bit.
-    // Use a conditional expression to be portable,
-    // a good optimizing compiler generates an arithmetic right shift
-    // and avoids the conditional branch.
-    UnsignedOfSize<N> mask = i < 0 ? ~UnsignedOfSize<N>(0) : 0;
-    // Now get the absolute value of "i" and cast to unsigned type UnsignedOfSize<N>.
-    // Cannot use std::abs() because the result is undefined
-    // in 2's complement systems for the most-negative value.
-    // Want to avoid conditional branch for performance reasons since
-    // CPU branch prediction will be ineffective when negative values
-    // occur randomly.
-    // Let "u" be "i" cast to unsigned type UnsignedOfSize<N>.
-    // Subtract "u" from 2*u if "i" is positive or 0 if "i" is negative.
-    // This yields the absolute value with the desired type without
-    // using a conditional branch and without invoking undefined or
-    // implementation defined behavior:
-    UnsignedOfSize<N> u = ((2 * UnsignedOfSize<N>(i)) & ~mask) - UnsignedOfSize<N>(i);
-    // Unconditionally store a minus sign when producing digits
-    // in a forward direction and increment the pointer only if
-    // the value is in fact negative.
-    // This avoids a conditional branch and is safe because we will
-    // always produce at least one digit and it will overwrite the
-    // minus sign when the value is not negative.
-    *p = '-';
-    p += (mask & 1);
-    p = convert::uitoa(p, u);
-    return p;
 }
 }
 
@@ -303,7 +296,7 @@ ALWAYS_INLINE inline char * writeUIntText(UInt128 _x, char * p)
 {
     /// If we the highest 64bit item is empty, we can print just the lowest item as u64
     if (_x.items[UInt128::_impl::little(1)] == 0)
-        return convert::itoa(_x.items[UInt128::_impl::little(0)], p);
+        return jeaiii::to_text_from_integer(p, _x.items[UInt128::_impl::little(0)]);
 
     /// Doing operations using __int128 is faster and we already rely on this feature
     using T = unsigned __int128;
@@ -334,7 +327,7 @@ ALWAYS_INLINE inline char * writeUIntText(UInt128 _x, char * p)
         current_block += max_multiple_of_hundred_blocks;
     }
 
-    char * highest_part_print = convert::itoa(uint64_t(x), p);
+    char * highest_part_print = jeaiii::to_text_from_integer(p, uint64_t(x));
     for (int i = 0; i < current_block; i++)
     {
         outTwoDigits(highest_part_print, two_values[current_block - 1 - i]);
@@ -450,12 +443,12 @@ ALWAYS_INLINE inline char * writeSIntText(T x, char * pos)
 
 char * itoa(UInt8 i, char * p)
 {
-    return convert::itoa(uint8_t(i), p);
+    return jeaiii::to_text_from_integer(p, uint8_t(i));
 }
 
 char * itoa(Int8 i, char * p)
 {
-    return convert::itoa(int8_t(i), p);
+    return jeaiii::to_text_from_integer(p, int8_t(i));
 }
 
 char * itoa(UInt128 i, char * p)
@@ -481,7 +474,7 @@ char * itoa(Int256 i, char * p)
 #define DEFAULT_ITOA(T) \
     char * itoa(T i, char * p) \
     { \
-        return convert::itoa(i, p); \
+        return jeaiii::to_text_from_integer(p, i); \
     }
 
 #define FOR_MISSING_INTEGER_TYPES(M) \

base/poco/Foundation/include/Poco/Logger.h

@@ -21,6 +21,8 @@
 #include <atomic>
 #include <cstddef>
 #include <map>
+#include <memory>
+#include <unordered_map>
 #include <vector>
 
 #include "Poco/Channel.h"

base/poco/Foundation/include/Poco/Message.h

@@ -19,6 +19,7 @@
 
 
 #include <map>
+#include <vector>
 #include "Poco/Foundation.h"
 #include "Poco/Timestamp.h"
 

base/poco/NetSSL_OpenSSL/include/Poco/Net/SecureSocketImpl.h

@@ -235,8 +235,6 @@ namespace Net
         /// Note that simply closing a socket is not sufficient
         /// to be able to re-use it again.
 
-        Poco::Timespan getMaxTimeout();
-
     private:
         SecureSocketImpl(const SecureSocketImpl &);
         SecureSocketImpl & operator=(const SecureSocketImpl &);
@@ -250,6 +248,9 @@ namespace Net
         Session::Ptr _pSession;
 
         friend class SecureStreamSocketImpl;
+
+        Poco::Timespan getMaxTimeoutOrLimit();
+        //// Return max(send, receive) if non zero, otherwise maximum timeout
     };
 
 

base/poco/NetSSL_OpenSSL/src/SecureSocketImpl.cpp

@@ -199,7 +199,7 @@ void SecureSocketImpl::connectSSL(bool performHandshake)
 		if (performHandshake && _pSocket->getBlocking())
 		{
 			int ret;
-			Poco::Timespan remaining_time = getMaxTimeout();
+			Poco::Timespan remaining_time = getMaxTimeoutOrLimit();
 			do
 			{
 				RemainingTimeCounter counter(remaining_time);
@@ -302,7 +302,7 @@ int SecureSocketImpl::sendBytes(const void* buffer, int length, int flags)
 			return rc;
 	}
 
-	Poco::Timespan remaining_time = getMaxTimeout();
+	Poco::Timespan remaining_time = getMaxTimeoutOrLimit();
 	do
 	{
 		RemainingTimeCounter counter(remaining_time);
@@ -338,7 +338,7 @@ int SecureSocketImpl::receiveBytes(void* buffer, int length, int flags)
 			return rc;
 	}
 
-	Poco::Timespan remaining_time = getMaxTimeout();
+	Poco::Timespan remaining_time = getMaxTimeoutOrLimit();
 	do
 	{
 		/// SSL record may consist of several TCP packets,
@@ -372,7 +372,7 @@ int SecureSocketImpl::completeHandshake()
 	poco_check_ptr (_pSSL);
 
 	int rc;
-	Poco::Timespan remaining_time = getMaxTimeout();
+	Poco::Timespan remaining_time = getMaxTimeoutOrLimit();
 	do
 	{
 		RemainingTimeCounter counter(remaining_time);
@@ -453,18 +453,29 @@ X509* SecureSocketImpl::peerCertificate() const
 		return 0;
 }
 
-Poco::Timespan SecureSocketImpl::getMaxTimeout()
+Poco::Timespan SecureSocketImpl::getMaxTimeoutOrLimit()
 {
 	std::lock_guard<std::recursive_mutex> lock(_mutex);
 	Poco::Timespan remaining_time = _pSocket->getReceiveTimeout();
 	Poco::Timespan send_timeout = _pSocket->getSendTimeout();
 	if (remaining_time < send_timeout)
 		remaining_time = send_timeout;
+	/// zero SO_SNDTIMEO/SO_RCVTIMEO works as no timeout, let's replicate this
+	///
+	/// NOTE: we cannot use INT64_MAX (std::numeric_limits<Poco::Timespan::TimeDiff>::max()),
+	/// since it will be later passed to poll() which accept int timeout, and
+	/// even though poll() accepts milliseconds and Timespan() accepts
+	/// microseconds, let's use smaller maximum value just to avoid some possible
+	/// issues, this should be enough anyway (it is ~24 days).
+	if (remaining_time == 0)
+		remaining_time = Poco::Timespan(std::numeric_limits<int>::max());
 	return remaining_time;
 }
 
 bool SecureSocketImpl::mustRetry(int rc, Poco::Timespan& remaining_time)
 {
+	if (remaining_time == 0)
+		return false;
 	std::lock_guard<std::recursive_mutex> lock(_mutex);
 	if (rc <= 0)
 	{
@@ -475,9 +486,7 @@ bool SecureSocketImpl::mustRetry(int rc, Poco::Timespan& remaining_time)
 		case SSL_ERROR_WANT_READ:
 			if (_pSocket->getBlocking())
 			{
-				/// Level-triggered mode of epoll_wait is used, so if SSL_read don't read all available data from socket,
+				if (_pSocket->pollImpl(remaining_time, Poco::Net::Socket::SELECT_READ))
-				/// epoll_wait returns true without waiting for new data even if remaining_time == 0
-				if (_pSocket->pollImpl(remaining_time, Poco::Net::Socket::SELECT_READ) && remaining_time != 0)
 					return true;
 				else
 					throw Poco::TimeoutException();
@@ -486,13 +495,15 @@ bool SecureSocketImpl::mustRetry(int rc, Poco::Timespan& remaining_time)
 		case SSL_ERROR_WANT_WRITE:
 			if (_pSocket->getBlocking())
 			{
-				/// The same as for SSL_ERROR_WANT_READ
+				if (_pSocket->pollImpl(remaining_time, Poco::Net::Socket::SELECT_WRITE))
-				if (_pSocket->pollImpl(remaining_time, Poco::Net::Socket::SELECT_WRITE) && remaining_time != 0)
 					return true;
 				else
 					throw Poco::TimeoutException();
 			}
 			break;
+		/// NOTE: POCO_EINTR is the same as SSL_ERROR_WANT_READ (at least in
+		/// OpenSSL), so this likely dead code, but let's leave it for
+		/// compatibility with other implementations
 		case SSL_ERROR_SYSCALL:
 			return socketError == POCO_EAGAIN || socketError == POCO_EINTR;
 		default:

contrib/CMakeLists.txt

@@ -228,6 +228,8 @@ add_contrib (ulid-c-cmake ulid-c)
 
 add_contrib (libssh-cmake libssh)
 
+add_contrib (prometheus-protobufs-cmake prometheus-protobufs prometheus-protobufs-gogo)
+
 # Put all targets defined here and in subdirectories under "contrib/<immediate-subdir>" folders in GUI-based IDEs.
 # Some of third-party projects may override CMAKE_FOLDER or FOLDER property of their targets, so they would not appear
 # in "contrib/..." as originally planned, so we workaround this by fixing FOLDER properties of all targets manually,

contrib/azure

@@ -1 +1 @@
-Subproject commit 6262a76ef4c4c330c84e58dd4f6f13f4e6230fcd
+Subproject commit 92c94d7f37a43cc8fc4d466884a95f610c0593bf

contrib/google-protobuf-cmake/protobuf_generate.cmake

@@ -157,15 +157,13 @@ function(protobuf_generate)
 
   set(_generated_srcs_all)
   foreach(_proto ${protobuf_generate_PROTOS})
-    get_filename_component(_abs_file ${_proto} ABSOLUTE)
+    # The protobuf compiler doesn't return paths to the files it generates so we have to calculate those paths here:
-    get_filename_component(_abs_dir ${_abs_file} DIRECTORY)
+    # _abs_file - absolute path to a .proto file,
-    get_filename_component(_basename ${_proto} NAME_WE)
+    # _possible_rel_dir - relative path to the .proto file from some import directory specified in Protobuf_IMPORT_DIRS,
-    file(RELATIVE_PATH _rel_dir ${CMAKE_CURRENT_SOURCE_DIR} ${_abs_dir})
+    # _basename - filename of the .proto file (without path and without extenstion).
-
+    get_proto_absolute_path(_abs_file "${_proto}" ${_protobuf_include_path})
-    set(_possible_rel_dir)
+    get_proto_relative_path(_possible_rel_dir "${_abs_file}" ${_protobuf_include_path})
-    if (NOT protobuf_generate_APPEND_PATH)
+    get_filename_component(_basename "${_abs_file}" NAME_WE)
-        set(_possible_rel_dir ${_rel_dir}/)
-    endif()
 
     set(_generated_srcs)
     foreach(_ext ${protobuf_generate_GENERATE_EXTENSIONS})
@@ -173,7 +171,7 @@ function(protobuf_generate)
     endforeach()
 
     if(protobuf_generate_DESCRIPTORS AND protobuf_generate_LANGUAGE STREQUAL cpp)
-      set(_descriptor_file "${CMAKE_CURRENT_BINARY_DIR}/${_basename}.desc")
+      set(_descriptor_file "${protobuf_generate_PROTOC_OUT_DIR}/${_possible_rel_dir}${_basename}.desc")
       set(_dll_desc_out "--descriptor_set_out=${_descriptor_file}")
       list(APPEND _generated_srcs ${_descriptor_file})
     endif()
@@ -196,3 +194,36 @@ function(protobuf_generate)
     target_sources(${protobuf_generate_TARGET} PRIVATE ${_generated_srcs_all})
   endif()
 endfunction()
+
+# Calculates the absolute path to a .proto file.
+function(get_proto_absolute_path result proto)
+  cmake_path(IS_ABSOLUTE proto _is_abs_path)
+  if(_is_abs_path)
+    set(${result} "${proto}" PARENT_SCOPE)
+    return()
+  endif()
+  foreach(_include_dir ${ARGN})
+    if(EXISTS "${_include_dir}/${proto}")
+      set(${result} "${_include_dir}/${proto}" PARENT_SCOPE)
+      return()
+    endif()
+  endforeach()
+  message(SEND_ERROR "Not found protobuf ${proto} in Protobuf_IMPORT_DIRS: ${ARGN}")
+endfunction()
+
+# Calculates a relative path to a .proto file. The returned path is relative to one of include directories.
+function(get_proto_relative_path result abs_path)
+  set(${result} "" PARENT_SCOPE)
+  get_filename_component(_abs_dir "${abs_path}" DIRECTORY)
+  foreach(_include_dir ${ARGN})
+    cmake_path(IS_PREFIX _include_dir "${_abs_dir}" _is_prefix)
+    if(_is_prefix)
+      file(RELATIVE_PATH _rel_dir "${_include_dir}" "${_abs_dir}")
+      if(NOT _rel_dir STREQUAL "")
+        set(${result} "${_rel_dir}/" PARENT_SCOPE)
+      endif()
+      return()
+    endif()
+  endforeach()
+  message(WARNING "Not found protobuf ${abs_path} in Protobuf_IMPORT_DIRS: ${ARGN}")
+endfunction()

contrib/icu-cmake/CMakeLists.txt

@@ -5,7 +5,7 @@ else ()
 endif ()
 
 if (NOT ENABLE_ICU)
-    message(STATUS "Not using icu")
+    message(STATUS "Not using ICU")
     return()
 endif()
 

contrib/jemalloc-cmake/CMakeLists.txt

@@ -34,7 +34,11 @@ if (OS_LINUX)
     # avoid spurious latencies and additional work associated with
     # MADV_DONTNEED. See
     # https://github.com/ClickHouse/ClickHouse/issues/11121 for motivation.
-    set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:0,dirty_decay_ms:5000,prof:true,prof_active:false,background_thread:true")
+    if (CMAKE_BUILD_TYPE_UC STREQUAL "DEBUG")
+        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:0,dirty_decay_ms:5000")
+    else()
+        set (JEMALLOC_CONFIG_MALLOC_CONF "percpu_arena:percpu,oversize_threshold:0,muzzy_decay_ms:0,dirty_decay_ms:5000,prof:true,prof_active:false,background_thread:true")
+    endif()
 else()
     set (JEMALLOC_CONFIG_MALLOC_CONF "oversize_threshold:0,muzzy_decay_ms:0,dirty_decay_ms:5000")
 endif()

contrib/llvm-project

@@ -1 +1 @@
-Subproject commit d2142eed98046a47ff7112e3cc1e197c8a5cd80f
+Subproject commit 2a8967b60cbe5bc2df253712bac343cc5263c5fc

contrib/openssl

@@ -1 +1 @@
-Subproject commit 5d81fa7068fc8c07f4d0997d5b703f3c541a637c
+Subproject commit ee2bb8513b28bf86b35404dd17a0e29305ca9e08

contrib/prometheus-protobufs-cmake/CMakeLists.txt

@@ -0,0 +1,34 @@
+option(ENABLE_PROMETHEUS_PROTOBUFS "Enable Prometheus Protobufs" ${ENABLE_PROTOBUF})
+
+if(NOT ENABLE_PROMETHEUS_PROTOBUFS)
+  message(STATUS "Not using prometheus-protobufs")
+  return()
+endif()
+
+set(Protobuf_INCLUDE_DIR "${ClickHouse_SOURCE_DIR}/contrib/google-protobuf/src")
+set(Prometheus_INCLUDE_DIR "${ClickHouse_SOURCE_DIR}/contrib/prometheus-protobufs")
+set(GogoProto_INCLUDE_DIR "${ClickHouse_SOURCE_DIR}/contrib/prometheus-protobufs-gogo")
+
+# Protobuf_IMPORT_DIRS specify where the protobuf compiler will look for .proto files.
+set(Old_Protobuf_IMPORT_DIRS ${Protobuf_IMPORT_DIRS})
+list(APPEND Protobuf_IMPORT_DIRS "${Protobuf_INCLUDE_DIR}" "${Prometheus_INCLUDE_DIR}" "${GogoProto_INCLUDE_DIR}")
+
+PROTOBUF_GENERATE_CPP(prometheus_protobufs_sources prometheus_protobufs_headers
+    "prompb/remote.proto"
+    "prompb/types.proto"
+    "gogoproto/gogo.proto"
+)
+
+set(Protobuf_IMPORT_DIRS ${Old_Protobuf_IMPORT_DIRS})
+
+# Ignore warnings while compiling protobuf-generated *.pb.h and *.pb.cpp files.
+set (CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -w")
+
+# Disable clang-tidy for protobuf-generated *.pb.h and *.pb.cpp files.
+set (CMAKE_CXX_CLANG_TIDY "")
+
+add_library(_prometheus_protobufs ${prometheus_protobufs_sources} ${prometheus_protobufs_headers})
+target_include_directories(_prometheus_protobufs SYSTEM PUBLIC "${CMAKE_CURRENT_BINARY_DIR}")
+target_link_libraries (_prometheus_protobufs PUBLIC ch_contrib::protobuf)
+
+add_library (ch_contrib::prometheus_protobufs ALIAS _prometheus_protobufs)

contrib/prometheus-protobufs-gogo/LICENSE

@@ -0,0 +1,35 @@
+Copyright (c) 2022, The Cosmos SDK Authors. All rights reserved.
+Copyright (c) 2013, The GoGo Authors. All rights reserved.
+
+Protocol Buffers for Go with Gadgets
+
+Go support for Protocol Buffers - Google's data interchange format
+
+Copyright 2010 The Go Authors.  All rights reserved.
+https://github.com/golang/protobuf
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+
+    * Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+    * Redistributions in binary form must reproduce the above
+copyright notice, this list of conditions and the following disclaimer
+in the documentation and/or other materials provided with the
+distribution.
+    * Neither the name of Google Inc. nor the names of its
+contributors may be used to endorse or promote products derived from
+this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

contrib/prometheus-protobufs-gogo/README

@@ -0,0 +1,4 @@
+File "gogoproto/gogo.proto" was downloaded from the "Protocol Buffers for Go with Gadgets" project:
+https://github.com/cosmos/gogoproto/blob/main/gogoproto/gogo.proto
+
+File "gogoproto/gogo.proto" is used in ClickHouse to compile prometheus protobufs.

contrib/prometheus-protobufs-gogo/gogoproto/gogo.proto

@@ -0,0 +1,145 @@
+// Protocol Buffers for Go with Gadgets
+//
+// Copyright (c) 2013, The GoGo Authors. All rights reserved.
+// http://github.com/cosmos/gogoproto
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+syntax = "proto2";
+package gogoproto;
+
+import "google/protobuf/descriptor.proto";
+
+option java_package = "com.google.protobuf";
+option java_outer_classname = "GoGoProtos";
+option go_package = "github.com/cosmos/gogoproto/gogoproto";
+
+extend google.protobuf.EnumOptions {
+	optional bool goproto_enum_prefix = 62001;
+	optional bool goproto_enum_stringer = 62021;
+	optional bool enum_stringer = 62022;
+	optional string enum_customname = 62023;
+	optional bool enumdecl = 62024;
+}
+
+extend google.protobuf.EnumValueOptions {
+	optional string enumvalue_customname = 66001;
+}
+
+extend google.protobuf.FileOptions {
+	optional bool goproto_getters_all = 63001;
+	optional bool goproto_enum_prefix_all = 63002;
+	optional bool goproto_stringer_all = 63003;
+	optional bool verbose_equal_all = 63004;
+	optional bool face_all = 63005;
+	optional bool gostring_all = 63006;
+	optional bool populate_all = 63007;
+	optional bool stringer_all = 63008;
+	optional bool onlyone_all = 63009;
+
+	optional bool equal_all = 63013;
+	optional bool description_all = 63014;
+	optional bool testgen_all = 63015;
+	optional bool benchgen_all = 63016;
+	optional bool marshaler_all = 63017;
+	optional bool unmarshaler_all = 63018;
+	optional bool stable_marshaler_all = 63019;
+
+	optional bool sizer_all = 63020;
+
+	optional bool goproto_enum_stringer_all = 63021;
+	optional bool enum_stringer_all = 63022;
+
+	optional bool unsafe_marshaler_all = 63023;
+	optional bool unsafe_unmarshaler_all = 63024;
+
+	optional bool goproto_extensions_map_all = 63025;
+	optional bool goproto_unrecognized_all = 63026;
+	optional bool gogoproto_import = 63027;
+	optional bool protosizer_all = 63028;
+	optional bool compare_all = 63029;
+    optional bool typedecl_all = 63030;
+    optional bool enumdecl_all = 63031;
+
+	optional bool goproto_registration = 63032;
+	optional bool messagename_all = 63033;
+
+	optional bool goproto_sizecache_all = 63034;
+	optional bool goproto_unkeyed_all = 63035;
+}
+
+extend google.protobuf.MessageOptions {
+	optional bool goproto_getters = 64001;
+	optional bool goproto_stringer = 64003;
+	optional bool verbose_equal = 64004;
+	optional bool face = 64005;
+	optional bool gostring = 64006;
+	optional bool populate = 64007;
+	optional bool stringer = 67008;
+	optional bool onlyone = 64009;
+
+	optional bool equal = 64013;
+	optional bool description = 64014;
+	optional bool testgen = 64015;
+	optional bool benchgen = 64016;
+	optional bool marshaler = 64017;
+	optional bool unmarshaler = 64018;
+	optional bool stable_marshaler = 64019;
+
+	optional bool sizer = 64020;
+
+	optional bool unsafe_marshaler = 64023;
+	optional bool unsafe_unmarshaler = 64024;
+
+	optional bool goproto_extensions_map = 64025;
+	optional bool goproto_unrecognized = 64026;
+
+	optional bool protosizer = 64028;
+	optional bool compare = 64029;
+
+	optional bool typedecl = 64030;
+
+	optional bool messagename = 64033;
+
+	optional bool goproto_sizecache = 64034;
+	optional bool goproto_unkeyed = 64035;
+}
+
+extend google.protobuf.FieldOptions {
+	optional bool nullable = 65001;
+	optional bool embed = 65002;
+	optional string customtype = 65003;
+	optional string customname = 65004;
+	optional string jsontag = 65005;
+	optional string moretags = 65006;
+	optional string casttype = 65007;
+	optional string castkey = 65008;
+	optional string castvalue = 65009;
+
+	optional bool stdtime = 65010;
+	optional bool stdduration = 65011;
+	optional bool wktpointer = 65012;
+
+	optional string castrepeated = 65013;
+}

contrib/prometheus-protobufs/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

contrib/prometheus-protobufs/README

@@ -0,0 +1,2 @@
+Files "prompb/remote.proto" and "prompb/types.proto" were downloaded from the Prometheus repository:
+https://github.com/prometheus/prometheus/tree/main/prompb

contrib/prometheus-protobufs/prompb/remote.proto

@@ -0,0 +1,88 @@
+// Copyright 2016 Prometheus Team
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+syntax = "proto3";
+package prometheus;
+
+option go_package = "prompb";
+
+import "prompb/types.proto";
+import "gogoproto/gogo.proto";
+
+message WriteRequest {
+  repeated prometheus.TimeSeries timeseries = 1 [(gogoproto.nullable) = false];
+  // Cortex uses this field to determine the source of the write request.
+  // We reserve it to avoid any compatibility issues.
+  reserved  2;
+  repeated prometheus.MetricMetadata metadata = 3 [(gogoproto.nullable) = false];
+}
+
+// ReadRequest represents a remote read request.
+message ReadRequest {
+  repeated Query queries = 1;
+
+  enum ResponseType {
+    // Server will return a single ReadResponse message with matched series that includes list of raw samples.
+    // It's recommended to use streamed response types instead.
+    //
+    // Response headers:
+    // Content-Type: "application/x-protobuf"
+    // Content-Encoding: "snappy"
+    SAMPLES = 0;
+    // Server will stream a delimited ChunkedReadResponse message that
+    // contains XOR or HISTOGRAM(!) encoded chunks for a single series.
+    // Each message is following varint size and fixed size bigendian
+    // uint32 for CRC32 Castagnoli checksum.
+    //
+    // Response headers:
+    // Content-Type: "application/x-streamed-protobuf; proto=prometheus.ChunkedReadResponse"
+    // Content-Encoding: ""
+    STREAMED_XOR_CHUNKS = 1;
+  }
+
+  // accepted_response_types allows negotiating the content type of the response.
+  //
+  // Response types are taken from the list in the FIFO order. If no response type in `accepted_response_types` is
+  // implemented by server, error is returned.
+  // For request that do not contain `accepted_response_types` field the SAMPLES response type will be used.
+  repeated ResponseType accepted_response_types = 2;
+}
+
+// ReadResponse is a response when response_type equals SAMPLES.
+message ReadResponse {
+  // In same order as the request's queries.
+  repeated QueryResult results = 1;
+}
+
+message Query {
+  int64 start_timestamp_ms = 1;
+  int64 end_timestamp_ms = 2;
+  repeated prometheus.LabelMatcher matchers = 3;
+  prometheus.ReadHints hints = 4;
+}
+
+message QueryResult {
+  // Samples within a time series must be ordered by time.
+  repeated prometheus.TimeSeries timeseries = 1;
+}
+
+// ChunkedReadResponse is a response when response_type equals STREAMED_XOR_CHUNKS.
+// We strictly stream full series after series, optionally split by time. This means that a single frame can contain
+// partition of the single series, but once a new series is started to be streamed it means that no more chunks will
+// be sent for previous one. Series are returned sorted in the same way TSDB block are internally.
+message ChunkedReadResponse {
+  repeated prometheus.ChunkedSeries chunked_series = 1;
+
+  // query_index represents an index of the query from ReadRequest.queries these chunks relates to.
+  int64 query_index = 2;
+}

contrib/prometheus-protobufs/prompb/types.proto

@@ -0,0 +1,187 @@
+// Copyright 2017 Prometheus Team
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+// http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+syntax = "proto3";
+package prometheus;
+
+option go_package = "prompb";
+
+import "gogoproto/gogo.proto";
+
+message MetricMetadata {
+  enum MetricType {
+    UNKNOWN        = 0;
+    COUNTER        = 1;
+    GAUGE          = 2;
+    HISTOGRAM      = 3;
+    GAUGEHISTOGRAM = 4;
+    SUMMARY        = 5;
+    INFO           = 6;
+    STATESET       = 7;
+  }
+
+  // Represents the metric type, these match the set from Prometheus.
+  // Refer to github.com/prometheus/common/model/metadata.go for details.
+  MetricType type = 1;
+  string metric_family_name = 2;
+  string help = 4;
+  string unit = 5;
+}
+
+message Sample {
+  double value    = 1;
+  // timestamp is in ms format, see model/timestamp/timestamp.go for
+  // conversion from time.Time to Prometheus timestamp.
+  int64 timestamp = 2;
+}
+
+message Exemplar {
+  // Optional, can be empty.
+  repeated Label labels = 1 [(gogoproto.nullable) = false];
+  double value = 2;
+  // timestamp is in ms format, see model/timestamp/timestamp.go for
+  // conversion from time.Time to Prometheus timestamp.
+  int64 timestamp = 3;
+}
+
+// A native histogram, also known as a sparse histogram.
+// Original design doc:
+// https://docs.google.com/document/d/1cLNv3aufPZb3fNfaJgdaRBZsInZKKIHo9E6HinJVbpM/edit
+// The appendix of this design doc also explains the concept of float
+// histograms. This Histogram message can represent both, the usual
+// integer histogram as well as a float histogram.
+message Histogram {
+  enum ResetHint {
+    UNKNOWN = 0; // Need to test for a counter reset explicitly.
+    YES     = 1; // This is the 1st histogram after a counter reset.
+    NO      = 2; // There was no counter reset between this and the previous Histogram.
+    GAUGE   = 3; // This is a gauge histogram where counter resets don't happen.
+  }
+
+  oneof count { // Count of observations in the histogram.
+    uint64 count_int   = 1;
+    double count_float = 2;
+  }
+  double sum = 3; // Sum of observations in the histogram.
+  // The schema defines the bucket schema. Currently, valid numbers
+  // are -4 <= n <= 8. They are all for base-2 bucket schemas, where 1
+  // is a bucket boundary in each case, and then each power of two is
+  // divided into 2^n logarithmic buckets. Or in other words, each
+  // bucket boundary is the previous boundary times 2^(2^-n). In the
+  // future, more bucket schemas may be added using numbers < -4 or >
+  // 8.
+  sint32 schema             = 4;
+  double zero_threshold     = 5; // Breadth of the zero bucket.
+  oneof zero_count { // Count in zero bucket.
+    uint64 zero_count_int     = 6;
+    double zero_count_float   = 7;
+  }
+
+  // Negative Buckets.
+  repeated BucketSpan negative_spans =  8 [(gogoproto.nullable) = false];
+  // Use either "negative_deltas" or "negative_counts", the former for
+  // regular histograms with integer counts, the latter for float
+  // histograms.
+  repeated sint64 negative_deltas    =  9; // Count delta of each bucket compared to previous one (or to zero for 1st bucket).
+  repeated double negative_counts    = 10; // Absolute count of each bucket.
+
+  // Positive Buckets.
+  repeated BucketSpan positive_spans = 11 [(gogoproto.nullable) = false];
+  // Use either "positive_deltas" or "positive_counts", the former for
+  // regular histograms with integer counts, the latter for float
+  // histograms.
+  repeated sint64 positive_deltas    = 12; // Count delta of each bucket compared to previous one (or to zero for 1st bucket).
+  repeated double positive_counts    = 13; // Absolute count of each bucket.
+
+  ResetHint reset_hint               = 14;
+  // timestamp is in ms format, see model/timestamp/timestamp.go for
+  // conversion from time.Time to Prometheus timestamp.
+  int64 timestamp = 15;
+}
+
+// A BucketSpan defines a number of consecutive buckets with their
+// offset. Logically, it would be more straightforward to include the
+// bucket counts in the Span. However, the protobuf representation is
+// more compact in the way the data is structured here (with all the
+// buckets in a single array separate from the Spans).
+message BucketSpan {
+  sint32 offset = 1; // Gap to previous span, or starting point for 1st span (which can be negative).
+  uint32 length = 2; // Length of consecutive buckets.
+}
+
+// TimeSeries represents samples and labels for a single time series.
+message TimeSeries {
+  // For a timeseries to be valid, and for the samples and exemplars
+  // to be ingested by the remote system properly, the labels field is required.
+  repeated Label labels         = 1 [(gogoproto.nullable) = false];
+  repeated Sample samples       = 2 [(gogoproto.nullable) = false];
+  repeated Exemplar exemplars   = 3 [(gogoproto.nullable) = false];
+  repeated Histogram histograms = 4 [(gogoproto.nullable) = false];
+}
+
+message Label {
+  string name  = 1;
+  string value = 2;
+}
+
+message Labels {
+  repeated Label labels = 1 [(gogoproto.nullable) = false];
+}
+
+// Matcher specifies a rule, which can match or set of labels or not.
+message LabelMatcher {
+  enum Type {
+    EQ  = 0;
+    NEQ = 1;
+    RE  = 2;
+    NRE = 3;
+  }
+  Type type    = 1;
+  string name  = 2;
+  string value = 3;
+}
+
+message ReadHints {
+  int64 step_ms = 1;  // Query step size in milliseconds.
+  string func = 2;    // String representation of surrounding function or aggregation.
+  int64 start_ms = 3; // Start time in milliseconds.
+  int64 end_ms = 4;   // End time in milliseconds.
+  repeated string grouping = 5; // List of label names used in aggregation.
+  bool by = 6; // Indicate whether it is without or by.
+  int64 range_ms = 7; // Range vector selector range in milliseconds.
+}
+
+// Chunk represents a TSDB chunk.
+// Time range [min, max] is inclusive.
+message Chunk {
+  int64 min_time_ms = 1;
+  int64 max_time_ms = 2;
+
+  // We require this to match chunkenc.Encoding.
+  enum Encoding {
+    UNKNOWN         = 0;
+    XOR             = 1;
+    HISTOGRAM       = 2;
+    FLOAT_HISTOGRAM = 3;
+  }
+  Encoding type  = 3;
+  bytes data     = 4;
+}
+
+// ChunkedSeries represents single, encoded time series.
+message ChunkedSeries {
+  // Labels should be sorted.
+  repeated Label labels = 1 [(gogoproto.nullable) = false];
+  // Chunks will be in start time order and may overlap.
+  repeated Chunk chunks = 2 [(gogoproto.nullable) = false];
+}

contrib/s2geometry

@@ -1 +1 @@
-Subproject commit 0547c38371777a1c1c8be263a6f05c3bf71bb05b
+Subproject commit 0146e2d1355828f8f633cb050948250ad7406c57

contrib/s2geometry-cmake/CMakeLists.txt

@@ -1,7 +1,8 @@
-option(ENABLE_S2_GEOMETRY "Enable S2 geometry library" ${ENABLE_LIBRARIES})
+option(ENABLE_S2_GEOMETRY "Enable S2 Geometry" ${ENABLE_LIBRARIES})
 
-if (NOT ENABLE_S2_GEOMETRY)
+# ARCH_S390X broke upstream, it can be re-enabled once https://github.com/google/s2geometry/pull/372 is merged
-    message(STATUS "Not using S2 geometry")
+if (NOT ENABLE_S2_GEOMETRY OR ARCH_S390X)
+    message(STATUS "Not using S2 Geometry")
     return()
 endif()
 
@@ -38,6 +39,7 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2cell_index.cc"
     "${S2_SOURCE_DIR}/s2/s2cell_union.cc"
     "${S2_SOURCE_DIR}/s2/s2centroids.cc"
+    "${S2_SOURCE_DIR}/s2/s2chain_interpolation_query.cc"
     "${S2_SOURCE_DIR}/s2/s2closest_cell_query.cc"
     "${S2_SOURCE_DIR}/s2/s2closest_edge_query.cc"
     "${S2_SOURCE_DIR}/s2/s2closest_point_query.cc"
@@ -46,6 +48,7 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2coords.cc"
     "${S2_SOURCE_DIR}/s2/s2crossing_edge_query.cc"
     "${S2_SOURCE_DIR}/s2/s2debug.cc"
+    "${S2_SOURCE_DIR}/s2/s2density_tree.cc"
     "${S2_SOURCE_DIR}/s2/s2earth.cc"
     "${S2_SOURCE_DIR}/s2/s2edge_clipping.cc"
     "${S2_SOURCE_DIR}/s2/s2edge_crosser.cc"
@@ -53,8 +56,10 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2edge_distances.cc"
     "${S2_SOURCE_DIR}/s2/s2edge_tessellator.cc"
     "${S2_SOURCE_DIR}/s2/s2error.cc"
+    "${S2_SOURCE_DIR}/s2/s2fractal.cc"
     "${S2_SOURCE_DIR}/s2/s2furthest_edge_query.cc"
     "${S2_SOURCE_DIR}/s2/s2hausdorff_distance_query.cc"
+    "${S2_SOURCE_DIR}/s2/s2index_cell_data.cc"
     "${S2_SOURCE_DIR}/s2/s2latlng.cc"
     "${S2_SOURCE_DIR}/s2/s2latlng_rect.cc"
     "${S2_SOURCE_DIR}/s2/s2latlng_rect_bounder.cc"
@@ -63,10 +68,10 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2lax_polyline_shape.cc"
     "${S2_SOURCE_DIR}/s2/s2loop.cc"
     "${S2_SOURCE_DIR}/s2/s2loop_measures.cc"
+    "${S2_SOURCE_DIR}/s2/s2max_distance_targets.cc"
     "${S2_SOURCE_DIR}/s2/s2measures.cc"
     "${S2_SOURCE_DIR}/s2/s2memory_tracker.cc"
     "${S2_SOURCE_DIR}/s2/s2metrics.cc"
-    "${S2_SOURCE_DIR}/s2/s2max_distance_targets.cc"
     "${S2_SOURCE_DIR}/s2/s2min_distance_targets.cc"
     "${S2_SOURCE_DIR}/s2/s2padded_cell.cc"
     "${S2_SOURCE_DIR}/s2/s2point_compression.cc"
@@ -80,10 +85,11 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2predicates.cc"
     "${S2_SOURCE_DIR}/s2/s2projections.cc"
     "${S2_SOURCE_DIR}/s2/s2r2rect.cc"
-    "${S2_SOURCE_DIR}/s2/s2region.cc"
+    "${S2_SOURCE_DIR}/s2/s2random.cc"
-    "${S2_SOURCE_DIR}/s2/s2region_term_indexer.cc"
     "${S2_SOURCE_DIR}/s2/s2region_coverer.cc"
     "${S2_SOURCE_DIR}/s2/s2region_intersection.cc"
+    "${S2_SOURCE_DIR}/s2/s2region_sharder.cc"
+    "${S2_SOURCE_DIR}/s2/s2region_term_indexer.cc"
     "${S2_SOURCE_DIR}/s2/s2region_union.cc"
     "${S2_SOURCE_DIR}/s2/s2shape_index.cc"
     "${S2_SOURCE_DIR}/s2/s2shape_index_buffered_region.cc"
@@ -94,9 +100,12 @@ set(S2_SRCS
     "${S2_SOURCE_DIR}/s2/s2shapeutil_coding.cc"
     "${S2_SOURCE_DIR}/s2/s2shapeutil_contains_brute_force.cc"
     "${S2_SOURCE_DIR}/s2/s2shapeutil_conversion.cc"
+    "${S2_SOURCE_DIR}/s2/s2shapeutil_count_vertices.cc"
     "${S2_SOURCE_DIR}/s2/s2shapeutil_edge_iterator.cc"
+    "${S2_SOURCE_DIR}/s2/s2shapeutil_edge_wrap.cc"
     "${S2_SOURCE_DIR}/s2/s2shapeutil_get_reference_point.cc"
     "${S2_SOURCE_DIR}/s2/s2shapeutil_visit_crossing_edge_pairs.cc"
+    "${S2_SOURCE_DIR}/s2/s2testing.cc"
     "${S2_SOURCE_DIR}/s2/s2text_format.cc"
     "${S2_SOURCE_DIR}/s2/s2wedge_relations.cc"
     "${S2_SOURCE_DIR}/s2/s2winding_operation.cc"
@@ -140,6 +149,7 @@ target_link_libraries(_s2 PRIVATE
     absl::strings
     absl::type_traits
     absl::utility
+    absl::vlog_is_on
 )
 
 target_include_directories(_s2 SYSTEM BEFORE PUBLIC "${S2_SOURCE_DIR}/")

contrib/sysroot

@@ -1 +1 @@
-Subproject commit 39c4713334f9f156dbf508f548d510d9129a657c
+Subproject commit cc385041b226d1fc28ead14dbab5d40a5f821dd8

contrib/vectorscan

@@ -1 +1 @@
-Subproject commit 38431d111781843741a781a57a6381a527d900a4
+Subproject commit d29730e1cb9daaa66bda63426cdce83505d2c809

contrib/vectorscan-cmake/CMakeLists.txt

@@ -1,11 +1,8 @@
-# We use vectorscan, a portable and API/ABI-compatible drop-in replacement for hyperscan.
+# Vectorscan is drop-in replacement for Hyperscan.
-
 if ((ARCH_AMD64 AND NOT NO_SSE3_OR_HIGHER) OR ARCH_AARCH64)
-    option (ENABLE_VECTORSCAN "Enable vectorscan library" ${ENABLE_LIBRARIES})
+    option (ENABLE_VECTORSCAN "Enable vectorscan" ${ENABLE_LIBRARIES})
 endif()
 
-# TODO PPC should generally work but needs manual generation of ppc/config.h file on a PPC machine
-
 if (NOT ENABLE_VECTORSCAN)
     message (STATUS "Not using vectorscan")
     return()
@@ -272,34 +269,24 @@ if (ARCH_AARCH64)
     )
 endif()
 
-# TODO
-# if (ARCH_PPC64LE)
-#     list(APPEND SRCS
-#         "${LIBRARY_DIR}/src/util/supervector/arch/ppc64el/impl.cpp"
-#     )
-# endif()
-
 add_library (_vectorscan ${SRCS})
 
-target_compile_options (_vectorscan PRIVATE
-    -fno-sanitize=undefined # assume the library takes care of itself
-    -O2 -fno-strict-aliasing -fno-omit-frame-pointer -fvisibility=hidden # options from original build system
-)
 # library has too much debug information
 if (OMIT_HEAVY_DEBUG_SYMBOLS)
     target_compile_options (_vectorscan PRIVATE -g0)
 endif()
 
-# Include version header manually generated by running the original build system
+target_include_directories (_vectorscan SYSTEM PUBLIC "${LIBRARY_DIR}/src")
-target_include_directories (_vectorscan SYSTEM PRIVATE common)
+
+# Makes the version header visible. It was generated by running the native build system manually.
+# Please update whenever you update vectorscan.
+target_include_directories (_vectorscan SYSTEM PUBLIC common)
 
 # vectorscan inherited some patched in-source versions of boost headers to fix a bug in
 # boost 1.69. This bug has been solved long ago but vectorscan's source code still
 # points to the patched versions, so include it here.
 target_include_directories (_vectorscan SYSTEM PRIVATE "${LIBRARY_DIR}/include")
 
-target_include_directories (_vectorscan SYSTEM PUBLIC "${LIBRARY_DIR}/src")
-
 # Include platform-specific config header generated by manually running the original build system
 # Please regenerate these files if you update vectorscan.
 

contrib/vectorscan-cmake/common/hs_version.h

@@ -32,8 +32,12 @@
 /**
  * A version string to identify this release of Hyperscan.
  */
-#define HS_VERSION_STRING "5.4.7 2022-06-20"
+#define HS_VERSION_STRING "5.4.11 2024-07-04"
 
 #define HS_VERSION_32BIT ((5 << 24) | (1 << 16) | (7 << 8) | 0)
 
+#define HS_MAJOR      5
+#define HS_MINOR      4
+#define HS_PATCH      11
+
 #endif /* HS_VERSION_H_C6428FAF8E3713 */

docker/packager/binary-builder/build.sh

@@ -111,6 +111,7 @@ fi
 mv ./programs/clickhouse* /output || mv ./programs/*_fuzzer /output
 [ -x ./programs/self-extracting/clickhouse ] && mv ./programs/self-extracting/clickhouse /output
 [ -x ./programs/self-extracting/clickhouse-stripped ] && mv ./programs/self-extracting/clickhouse-stripped /output
+[ -x ./programs/self-extracting/clickhouse-keeper ] && mv ./programs/self-extracting/clickhouse-keeper /output
 mv ./src/unit_tests_dbms /output ||: # may not exist for some binary builds
 mv ./programs/*.dict ./programs/*.options ./programs/*_seed_corpus.zip /output ||: # libFuzzer oss-fuzz compatible infrastructure
 

docker/packager/packager

@@ -276,10 +276,7 @@ def parse_env_variables(
         if is_release_build(debug_build, package_type, sanitizer, coverage):
             cmake_flags.append("-DSPLIT_DEBUG_SYMBOLS=ON")
             result.append("WITH_PERFORMANCE=1")
-            if is_cross_arm:
+            cmake_flags.append("-DBUILD_STANDALONE_KEEPER=1")
-                cmake_flags.append("-DBUILD_STANDALONE_KEEPER=1")
-            else:
-                result.append("BUILD_MUSL_KEEPER=1")
     elif package_type == "fuzzers":
         cmake_flags.append("-DENABLE_FUZZING=1")
         cmake_flags.append("-DENABLE_PROTOBUF=1")

docker/test/stateful/run.sh

@@ -213,6 +213,10 @@ function run_tests()
         ADDITIONAL_OPTIONS+=('--s3-storage')
     fi
 
+    if [[ -n "$USE_AZURE_STORAGE_FOR_MERGE_TREE" ]] && [[ "$USE_AZURE_STORAGE_FOR_MERGE_TREE" -eq 1 ]]; then
+        ADDITIONAL_OPTIONS+=('--azure-blob-storage')
+    fi
+
     if [[ -n "$USE_DATABASE_ORDINARY" ]] && [[ "$USE_DATABASE_ORDINARY" -eq 1 ]]; then
         ADDITIONAL_OPTIONS+=('--db-engine=Ordinary')
     fi

docker/test/stateless/run.sh

@@ -207,7 +207,7 @@ function run_tests()
 
     if [[ -n "$USE_AZURE_STORAGE_FOR_MERGE_TREE" ]] && [[ "$USE_AZURE_STORAGE_FOR_MERGE_TREE"  -eq 1 ]]; then
         # to disable the same tests
-        ADDITIONAL_OPTIONS+=('--s3-storage')
+        ADDITIONAL_OPTIONS+=('--azure-blob-storage')
         # azurite is slow, but with these two settings it can be super slow
         ADDITIONAL_OPTIONS+=('--no-random-settings')
         ADDITIONAL_OPTIONS+=('--no-random-merge-tree-settings')
@@ -253,7 +253,7 @@ function run_tests()
     try_run_with_retry 10 clickhouse-client -q "insert into system.zookeeper (name, path, value) values ('auxiliary_zookeeper2', '/test/chroot/', '')"
 
     set +e
-    clickhouse-test --testname --shard --zookeeper --check-zookeeper-session --hung-check --print-time \
+    timeout -s TERM --preserve-status 120m clickhouse-test --testname --shard --zookeeper --check-zookeeper-session --hung-check --print-time \
          --no-drop-if-fail --test-runs "$NUM_TRIES" "${ADDITIONAL_OPTIONS[@]}" 2>&1 \
     | ts '%Y-%m-%d %H:%M:%S' \
     | tee -a test_output/test_result.txt

docker/test/stress/run.sh

@@ -110,6 +110,15 @@ start_server
 clickhouse-client --query "SHOW TABLES FROM datasets"
 clickhouse-client --query "SHOW TABLES FROM test"
 
+if [[ "$USE_S3_STORAGE_FOR_MERGE_TREE" == "1" ]]; then
+    TEMP_POLICY="s3_cache"
+elif [[ "$USE_AZURE_STORAGE_FOR_MERGE_TREE" == "1" ]]; then
+    TEMP_POLICY="azure_cache"
+else
+    TEMP_POLICY="default"
+fi
+
+
 clickhouse-client --query "CREATE TABLE test.hits_s3 (WatchID UInt64,  JavaEnable UInt8,  Title String,  GoodEvent Int16,
     EventTime DateTime,  EventDate Date,  CounterID UInt32,  ClientIP UInt32,  ClientIP6 FixedString(16),  RegionID UInt32,
     UserID UInt64,  CounterClass Int8,  OS UInt8,  UserAgent UInt8,  URL String,  Referer String,  URLDomain String,  RefererDomain String,
@@ -135,7 +144,7 @@ clickhouse-client --query "CREATE TABLE test.hits_s3 (WatchID UInt64,  JavaEnabl
     URLHash UInt64,  CLID UInt32,  YCLID UInt64,  ShareService String,  ShareURL String,  ShareTitle String,
     ParsedParams Nested(Key1 String,  Key2 String, Key3 String, Key4 String, Key5 String,  ValueDouble Float64),
     IslandID FixedString(16),  RequestNum UInt32,  RequestTry UInt8) ENGINE = MergeTree() PARTITION BY toYYYYMM(EventDate)
-    ORDER BY (CounterID, EventDate, intHash32(UserID)) SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='s3_cache'"
+    ORDER BY (CounterID, EventDate, intHash32(UserID)) SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='$TEMP_POLICY'"
 clickhouse-client --query "CREATE TABLE test.hits (WatchID UInt64,  JavaEnable UInt8,  Title String,  GoodEvent Int16,
     EventTime DateTime,  EventDate Date,  CounterID UInt32,  ClientIP UInt32,  ClientIP6 FixedString(16),  RegionID UInt32,
     UserID UInt64,  CounterClass Int8,  OS UInt8,  UserAgent UInt8,  URL String,  Referer String,  URLDomain String,
@@ -161,7 +170,7 @@ clickhouse-client --query "CREATE TABLE test.hits (WatchID UInt64,  JavaEnable U
     URLHash UInt64,  CLID UInt32,  YCLID UInt64,  ShareService String,  ShareURL String,  ShareTitle String,
     ParsedParams Nested(Key1 String,  Key2 String, Key3 String, Key4 String, Key5 String,  ValueDouble Float64),
     IslandID FixedString(16),  RequestNum UInt32,  RequestTry UInt8) ENGINE = MergeTree() PARTITION BY toYYYYMM(EventDate)
-    ORDER BY (CounterID, EventDate, intHash32(UserID)) SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='s3_cache'"
+    ORDER BY (CounterID, EventDate, intHash32(UserID)) SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='$TEMP_POLICY'"
 clickhouse-client --query "CREATE TABLE test.visits (CounterID UInt32,  StartDate Date,  Sign Int8,  IsNew UInt8,
     VisitID UInt64,  UserID UInt64,  StartTime DateTime,  Duration UInt32,  UTCStartTime DateTime,  PageViews Int32,
     Hits Int32,  IsBounce UInt8,  Referer String,  StartURL String,  RefererDomain String,  StartURLDomain String,
@@ -195,7 +204,7 @@ clickhouse-client --query "CREATE TABLE test.visits (CounterID UInt32,  StartDat
     Market Nested(Type UInt8, GoalID UInt32, OrderID String,  OrderPrice Int64,  PP UInt32,  DirectPlaceID UInt32,  DirectOrderID  UInt32,
     DirectBannerID UInt32,  GoodID String, GoodName String, GoodQuantity Int32,  GoodPrice Int64),  IslandID FixedString(16))
     ENGINE = CollapsingMergeTree(Sign) PARTITION BY toYYYYMM(StartDate) ORDER BY (CounterID, StartDate, intHash32(UserID), VisitID)
-    SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='s3_cache'"
+    SAMPLE BY intHash32(UserID) SETTINGS index_granularity = 8192, storage_policy='$TEMP_POLICY'"
 
 clickhouse-client --query "INSERT INTO test.hits_s3 SELECT * FROM datasets.hits_v1 SETTINGS enable_filesystem_cache_on_write_operations=0"
 clickhouse-client --query "INSERT INTO test.hits SELECT * FROM datasets.hits_v1 SETTINGS enable_filesystem_cache_on_write_operations=0"
@@ -211,19 +220,29 @@ clickhouse-client --query "SYSTEM STOP THREAD FUZZER"
 stop_server
 
 # Let's enable S3 storage by default
-export USE_S3_STORAGE_FOR_MERGE_TREE=1
 export RANDOMIZE_OBJECT_KEY_TYPE=1
 export ZOOKEEPER_FAULT_INJECTION=1
 export THREAD_POOL_FAULT_INJECTION=1
 configure
 
-# But we still need default disk because some tables loaded only into it
+if [[ "$USE_S3_STORAGE_FOR_MERGE_TREE" == "1" ]]; then
-sudo cat /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml \
+    # But we still need default disk because some tables loaded only into it
-  | sed "s|<main><disk>s3</disk></main>|<main><disk>s3</disk></main><default><disk>default</disk></default>|" \
+    sudo cat /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml \
-  > /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp
+      | sed "s|<main><disk>s3</disk></main>|<main><disk>s3</disk></main><default><disk>default</disk></default>|" \
-mv /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
+      > /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp
-sudo chown clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
+    mv /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml.tmp /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
-sudo chgrp clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
+    sudo chown clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
+    sudo chgrp clickhouse /etc/clickhouse-server/config.d/s3_storage_policy_by_default.xml
+elif [[ "$USE_AZURE_STORAGE_FOR_MERGE_TREE" == "1" ]]; then
+    # But we still need default disk because some tables loaded only into it
+    sudo cat /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml \
+      | sed "s|<main><disk>azure</disk></main>|<main><disk>azure</disk></main><default><disk>default</disk></default>|" \
+      > /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml.tmp
+    mv /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml.tmp /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml
+    sudo chown clickhouse /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml
+    sudo chgrp clickhouse /etc/clickhouse-server/config.d/azure_storage_policy_by_default.xml
+fi
+
 
 sudo cat /etc/clickhouse-server/config.d/logger_trace.xml \
    | sed "s|<level>trace</level>|<level>test</level>|" \

docs/en/engines/table-engines/mergetree-family/mergetree.md

@@ -993,11 +993,11 @@ They can be used for prewhere optimization only if we enable `set allow_statisti
 
 - `TDigest`
 
-    Stores distribution of values from numeric columns in [TDigest](https://github.com/tdunning/t-digest) sketch.
+    [TDigest](https://github.com/tdunning/t-digest) sketches which allow to compute approximate percentiles (e.g. the 90th percentile) for numeric columns.
 
 - `Uniq`
 
-    Estimate the number of distinct values of a column by HyperLogLog.
+    [HyperLogLog](https://en.wikipedia.org/wiki/HyperLogLog) sketches which provide an estimation how many distinct values a column contains.
 
 ## Column-level Settings {#column-level-settings}
 

docs/en/operations/external-authenticators/ssl-x509.md

@@ -6,23 +6,30 @@ import SelfManaged from '@site/docs/en/_snippets/_self_managed_only_no_roadmap.m
 
 <SelfManaged />
 
-[SSL 'strict' option](../server-configuration-parameters/settings.md#server_configuration_parameters-openssl) enables mandatory certificate validation for the incoming connections. In this case, only connections with trusted certificates can be established. Connections with untrusted certificates will be rejected. Thus, certificate validation allows to uniquely authenticate an incoming connection. `Common Name` field of the certificate is used to identify connected user. This allows to associate multiple certificates with the same user. Additionally, reissuing and revoking of the certificates does not affect the ClickHouse configuration.
+[SSL 'strict' option](../server-configuration-parameters/settings.md#server_configuration_parameters-openssl) enables mandatory certificate validation for the incoming connections. In this case, only connections with trusted certificates can be established. Connections with untrusted certificates will be rejected. Thus, certificate validation allows to uniquely authenticate an incoming connection. `Common Name` or `subjectAltName extension` field of the certificate is used to identify the connected user. This allows to associate multiple certificates with the same user. Additionally, reissuing and revoking of the certificates does not affect the ClickHouse configuration.
 
-To enable SSL certificate authentication, a list of `Common Name`'s for each ClickHouse user must be specified in the settings file `users.xml `:
+To enable SSL certificate authentication, a list of `Common Name`'s or `Subject Alt Name`'s for each ClickHouse user must be specified in the settings file `users.xml `:
 
 **Example**
 ```xml
 <clickhouse>
     <!- ... -->
     <users>
-        <user_name>
+        <user_name_1>
             <ssl_certificates>
                 <common_name>host.domain.com:example_user</common_name>
                 <common_name>host.domain.com:example_user_dev</common_name>
                 <!-- More names -->
             </ssl_certificates>
             <!-- Other settings -->
-        </user_name>
+        </user_name_1>
+        <user_name_2>
+            <ssl_certificates>
+                <subject_alt_name>DNS:host.domain.com</subject_alt_name>
+                <!-- More names -->
+            </ssl_certificates>
+            <!-- Other settings -->
+        </user_name_2>
     </users>
 </clickhouse>
 ```

docs/en/operations/settings/settings.md

@@ -2536,7 +2536,7 @@ Possible values:
 - 0 — Optimization disabled.
 - 1 — Optimization enabled.
 
-Default value: `0`.
+Default value: `1`.
 
 ## optimize_trivial_count_query {#optimize-trivial-count-query}
 

docs/en/sql-reference/data-types/json.md

@@ -1,24 +1,20 @@
 ---
-slug: /en/sql-reference/data-types/json
+slug: /en/sql-reference/data-types/object-data-type
 sidebar_position: 26
-sidebar_label: JSON
+sidebar_label: Object Data Type
+keywords: [object, data type]
 ---
 
-# JSON
+# Object Data Type
 
 :::note
-This feature is experimental and is not production-ready. If you need to work with JSON documents, consider using [this guide](/docs/en/integrations/data-ingestion/data-formats/json.md) instead.
+This feature is not production-ready and is now deprecated. If you need to work with JSON documents, consider using [this guide](/docs/en/integrations/data-ingestion/data-formats/json) instead. A new implementation to support JSON object is in progress and can be tracked [here](https://github.com/ClickHouse/ClickHouse/issues/54864)
 :::
 
 Stores JavaScript Object Notation (JSON) documents in a single column.
 
 `JSON` is an alias for `Object('json')`.
 
-:::note
-The JSON data type is an obsolete feature. Do not use it.
-If you want to use it, set `allow_experimental_object_type = 1`.
-:::
-
 ## Example
 
 **Example 1**
@@ -49,7 +45,7 @@ SELECT o.a, o.b.c, o.b.d[3] FROM json
 
 **Example 2**
 
-To be able to create an ordered `MergeTree` family table the sorting key has to be extracted into its column. For example, to insert a file of compressed HTTP access logs in JSON format:
+To be able to create an ordered `MergeTree` family table, the sorting key has to be extracted into its column. For example, to insert a file of compressed HTTP access logs in JSON format:
 
 ```sql
 CREATE TABLE logs
@@ -69,7 +65,7 @@ FROM file('access.json.gz', JSONAsString)
 
 ## Displaying JSON columns
 
-When displaying a `JSON` column ClickHouse only shows the field values by default (because internally, it is represented as a tuple). You can display the field names as well by setting `output_format_json_named_tuples_as_objects = 1`:
+When displaying a `JSON` column, ClickHouse only shows the field values by default (because internally, it is represented as a tuple). You can also display the field names by setting `output_format_json_named_tuples_as_objects = 1`:
 
 ```sql
 SET output_format_json_named_tuples_as_objects = 1
@@ -83,4 +79,5 @@ SELECT * FROM json FORMAT JSONEachRow
 
 ## Related Content
 
+- [Using JSON in ClickHouse](/docs/en/integrations/data-formats/json)
 - [Getting Data Into ClickHouse - Part 2 - A JSON detour](https://clickhouse.com/blog/getting-data-into-clickhouse-part-2-json)

docs/en/sql-reference/functions/bit-functions.md

@@ -173,7 +173,7 @@ See function [substring](string-functions.md#substring).
 
 ## bitTest
 
-Takes any integer and converts it into [binary form](https://en.wikipedia.org/wiki/Binary_number), returns the value of a bit at specified position. The countdown starts from 0 from the right to the left.
+Takes any integer and converts it into [binary form](https://en.wikipedia.org/wiki/Binary_number), returns the value of a bit at specified position. Counting is right-to-left, starting at 0.
 
 **Syntax**
 
@@ -226,7 +226,7 @@ Result:
 
 ## bitTestAll
 
-Returns result of [logical conjuction](https://en.wikipedia.org/wiki/Logical_conjunction) (AND operator) of all bits at given positions. The countdown starts from 0 from the right to the left.
+Returns result of [logical conjuction](https://en.wikipedia.org/wiki/Logical_conjunction) (AND operator) of all bits at given positions. Counting is right-to-left, starting at 0.
 
 The conjuction for bit-wise operations:
 
@@ -289,7 +289,7 @@ Result:
 
 ## bitTestAny
 
-Returns result of [logical disjunction](https://en.wikipedia.org/wiki/Logical_disjunction) (OR operator) of all bits at given positions. The countdown starts from 0 from the right to the left.
+Returns result of [logical disjunction](https://en.wikipedia.org/wiki/Logical_disjunction) (OR operator) of all bits at given positions. Counting is right-to-left, starting at 0.
 
 The disjunction for bit-wise operations:
 

docs/en/sql-reference/functions/other-functions.md

@@ -3860,3 +3860,138 @@ Result:
 └───────────────┘
 ```
 
+## transactionID
+
+Returns the ID of a [transaction](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback).
+
+:::note
+This function is part of an experimental feature set. Enable experimental transaction support by adding this setting to your configuration:
+
+```
+<clickhouse>
+  <allow_experimental_transactions>1</allow_experimental_transactions>
+</clickhouse>
+```
+
+For more information see the page [Transactional (ACID) support](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback).
+:::
+
+**Syntax**
+
+```sql
+transactionID()
+```
+
+**Returned value**
+
+- Returns a tuple consisting of `start_csn`, `local_tid` and `host_id`. [Tuple](../data-types/tuple.md).
+
+- `start_csn`: Global sequential number, the newest commit timestamp that was seen when this transaction began. [UInt64](../data-types/int-uint.md).
+- `local_tid`: Local sequential number that is unique for each transaction started by this host within a specific start_csn. [UInt64](../data-types/int-uint.md).
+- `host_id`: UUID of the host that has started this transaction. [UUID](../data-types/uuid.md).
+
+**Example**
+
+Query:
+
+```sql
+BEGIN TRANSACTION;
+SELECT transactionID();
+ROLLBACK;
+```
+
+Result:
+
+```response
+┌─transactionID()────────────────────────────────┐
+│ (32,34,'0ee8b069-f2bb-4748-9eae-069c85b5252b') │
+└────────────────────────────────────────────────┘
+```
+
+## transactionLatestSnapshot
+
+Returns the newest snapshot (Commit Sequence Number) of a [transaction](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback) that is available for reading.
+
+:::note
+This function is part of an experimental feature set. Enable experimental transaction support by adding this setting to your configuration:
+
+```
+<clickhouse>
+  <allow_experimental_transactions>1</allow_experimental_transactions>
+</clickhouse>
+```
+
+For more information see the page [Transactional (ACID) support](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback).
+:::
+
+**Syntax**
+
+```sql
+transactionLatestSnapshot()
+```
+
+**Returned value**
+
+- Returns the latest snapshot (CSN) of a transaction. [UInt64](../data-types/int-uint.md)
+
+**Example**
+
+Query:
+
+```sql
+BEGIN TRANSACTION;
+SELECT transactionLatestSnapshot();
+ROLLBACK;
+```
+
+Result:
+
+```response
+┌─transactionLatestSnapshot()─┐
+│                          32 │
+└─────────────────────────────┘
+```
+
+## transactionOldestSnapshot
+
+Returns the oldest snapshot (Commit Sequence Number) that is visible for some running [transaction](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback).
+
+:::note
+This function is part of an experimental feature set. Enable experimental transaction support by adding this setting to your configuration:
+
+```
+<clickhouse>
+  <allow_experimental_transactions>1</allow_experimental_transactions>
+</clickhouse>
+```
+
+For more information see the page [Transactional (ACID) support](https://clickhouse.com/docs/en/guides/developer/transactional#transactions-commit-and-rollback).
+:::
+
+**Syntax**
+
+```sql
+transactionOldestSnapshot()
+```
+
+**Returned value**
+
+- Returns the oldest snapshot (CSN) of a transaction. [UInt64](../data-types/int-uint.md)
+
+**Example**
+
+Query:
+
+```sql
+BEGIN TRANSACTION;
+SELECT transactionLatestSnapshot();
+ROLLBACK;
+```
+
+Result:
+
+```response
+┌─transactionOldestSnapshot()─┐
+│                          32 │
+└─────────────────────────────┘
+```

docs/en/sql-reference/statements/alter/user.md

@@ -12,7 +12,7 @@ Syntax:
 ``` sql
 ALTER USER [IF EXISTS] name1 [ON CLUSTER cluster_name1] [RENAME TO new_name1]
         [, name2 [ON CLUSTER cluster_name2] [RENAME TO new_name2] ...]
-    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name'}]
+    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name' | SAN 'TYPE:subject_alt_name'}]
     [[ADD | DROP] HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
     [VALID UNTIL datetime]
     [DEFAULT ROLE role [,...] | ALL | ALL EXCEPT role [,...] ]

docs/en/sql-reference/statements/create/table.md

@@ -152,7 +152,7 @@ SELECT * FROM test;
 
 `MATERIALIZED expr`
 
-Materialized expression. Values of such columns are always calculated, they cannot be specified in INSERT queries.
+Materialized expression. Values of such columns are automatically calculated according to the specified materialized expression when rows are inserted. Values cannot be explicitly specified during `INSERT`s.
 
 Also, default value columns of this type are not included in the result of `SELECT *`. This is to preserve the invariant that the result of a `SELECT *` can always be inserted back into the table using `INSERT`. This behavior can be disabled with setting `asterisk_include_materialized_columns`.
 

docs/en/sql-reference/statements/create/user.md

@@ -12,7 +12,7 @@ Syntax:
 ``` sql
 CREATE USER [IF NOT EXISTS | OR REPLACE] name1 [ON CLUSTER cluster_name1]
         [, name2 [ON CLUSTER cluster_name2] ...]
-    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name'} | {WITH ssh_key BY KEY 'public_key' TYPE 'ssh-rsa|...'} | {WITH http SERVER 'server_name' [SCHEME 'Basic']}]
+    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name' | SAN 'TYPE:subject_alt_name'} | {WITH ssh_key BY KEY 'public_key' TYPE 'ssh-rsa|...'} | {WITH http SERVER 'server_name' [SCHEME 'Basic']}]
     [HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
     [VALID UNTIL datetime]
     [IN access_storage_type]

docs/en/sql-reference/table-functions/fuzzQuery.md

@@ -0,0 +1,36 @@
+---
+slug: /en/sql-reference/table-functions/fuzzQuery
+sidebar_position: 75
+sidebar_label: fuzzQuery
+---
+
+# fuzzQuery
+
+Perturbs the given query string with random variations.
+
+``` sql
+fuzzQuery(query[, max_query_length[, random_seed]])
+```
+
+**Arguments**
+
+- `query` (String) - The source query to perform the fuzzing on.
+- `max_query_length` (UInt64) - A maximum length the query can get during the fuzzing process.
+- `random_seed` (UInt64) - A random seed for producing stable results.
+
+**Returned Value**
+
+A table object with a single column containing perturbed query strings.
+
+## Usage Example
+
+``` sql
+SELECT * FROM fuzzQuery('SELECT materialize(\'a\' AS key) GROUP BY key') LIMIT 2;
+```
+
+```
+   ┌─query──────────────────────────────────────────────────────────┐
+1. │ SELECT 'a' AS key GROUP BY key                                 │
+2. │ EXPLAIN PIPELINE compact = true SELECT 'a' AS key GROUP BY key │
+   └────────────────────────────────────────────────────────────────┘
+```

docs/en/sql-reference/table-functions/s3.md

@@ -269,9 +269,9 @@ FROM s3(
 
 ## Virtual Columns {#virtual-columns}
 
-- `_path` — Path to the file. Type: `LowCardinalty(String)`.
+- `_path` — Path to the file. Type: `LowCardinalty(String)`. In case of archive, shows path in a format: "{path_to_archive}::{path_to_file_inside_archive}"
-- `_file` — Name of the file. Type: `LowCardinalty(String)`.
+- `_file` — Name of the file. Type: `LowCardinalty(String)`. In case of archive shows name of the file inside the archive.
-- `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the file size is unknown, the value is `NULL`.
+- `_size` — Size of the file in bytes. Type: `Nullable(UInt64)`. If the file size is unknown, the value is `NULL`. In case of archive shows uncompressed file size of the file inside the archive. 
 - `_time` — Last modified time of the file. Type: `Nullable(DateTime)`. If the time is unknown, the value is `NULL`.
 
 ## Storage Settings {#storage-settings}

docs/en/sql-reference/window-functions/index.md

@@ -80,8 +80,8 @@ These functions can be used only as a window function.
 - `nth_value(x, offset)` - Return the first non-NULL value evaluated against the nth row (offset) in its ordered frame.
 - `rank()` - Rank the current row within its partition with gaps.
 - `dense_rank()` - Rank the current row within its partition without gaps.
-- `lagInFrame(x)` - Return a value evaluated at the row that is at a specified physical offset row before the current row within the ordered frame.
+- `lagInFrame(x[, offset[, default]])` - Return a value evaluated at the row that is at a specified physical offset row before the current row within the ordered frame. The offset parameter, if not specified, defaults to 1, meaning it will fetch the value from the next row. If the calculated row exceeds the boundaries of the window frame, the specified default value is returned.
-- `leadInFrame(x)` - Return a value evaluated at the row that is offset rows after the current row within the ordered frame.
+- `leadInFrame(x[, offset[, default]])` - Return a value evaluated at the row that is offset rows after the current row within the ordered frame. If offset is not provided, it defaults to 1. If the offset leads to a position outside the window frame, the specified default value is used.
 
 ## Examples
 

docs/ru/operations/external-authenticators/ssl-x509.md

@@ -3,23 +3,30 @@ slug: /ru/operations/external-authenticators/ssl-x509
 ---
 # Аутентификация по сертификату SSL X.509  {#ssl-external-authentication}
 
-[Опция 'strict'](../server-configuration-parameters/settings.md#server_configuration_parameters-openssl) включает обязательную проверку сертификатов входящих соединений в библиотеке `SSL`. В этом случае могут быть установлены только соединения, представившие действительный сертификат. Соединения с недоверенными сертификатами будут отвергнуты. Таким образом, проверка сертификата позволяет однозначно аутентифицировать входящее соединение. Идентификация пользователя осуществляется по полю `Common Name` сертификата. Это позволяет ассоциировать несколько сертификатов с одним и тем же пользователем. Дополнительно, перевыпуск и отзыв сертификата не требуют изменения конфигурации ClickHouse.
+[Опция 'strict'](../server-configuration-parameters/settings.md#server_configuration_parameters-openssl) включает обязательную проверку сертификатов входящих соединений в библиотеке `SSL`. В этом случае могут быть установлены только соединения, представившие действительный сертификат. Соединения с недоверенными сертификатами будут отвергнуты. Таким образом, проверка сертификата позволяет однозначно аутентифицировать входящее соединение. Идентификация пользователя осуществляется по полю `Common Name` или `subjectAltName` сертификата. Это позволяет ассоциировать несколько сертификатов с одним и тем же пользователем. Дополнительно, перевыпуск и отзыв сертификата не требуют изменения конфигурации ClickHouse.
 
-Для включения аутентификации по SSL сертификату, необходимо указать список `Common Name` для каждого пользователя ClickHouse в файле настройки `config.xml`:
+Для включения аутентификации по SSL сертификату, необходимо указать список `Common Name` или `subjectAltName` для каждого пользователя ClickHouse в файле настройки `config.xml`:
 
 **Example**
 ```xml
 <clickhouse>
     <!- ... -->
     <users>
-        <user_name>
+        <user_name_1>
-            <certificates>
+            <ssl_certificates>
                 <common_name>host.domain.com:example_user</common_name>
                 <common_name>host.domain.com:example_user_dev</common_name>
                 <!-- More names -->
-            </certificates>
+            </ssl_certificates>
             <!-- Other settings -->
-        </user_name>
+        </user_name_1>
+        <user_name_2>
+            <ssl_certificates>
+                <subject_alt_name>DNS:host.domain.com</subject_alt_name>
+                <!-- More names -->
+            </ssl_certificates>
+            <!-- Other settings -->
+        </user_name_2>
     </users>
 </clickhouse>
 ```

docs/ru/operations/settings/settings.md

@@ -2077,7 +2077,7 @@ SELECT * FROM test_table
 -   0 — оптимизация отключена.
 -   1 — оптимизация включена.
 
-Значение по умолчанию: `0`.
+Значение по умолчанию: `1`.
 
 ## optimize_trivial_count_query {#optimize-trivial-count-query}
 

docs/ru/sql-reference/statements/create/user.md

@@ -13,7 +13,7 @@ sidebar_label: "Пользователь"
 ``` sql
 CREATE USER [IF NOT EXISTS | OR REPLACE] name1 [ON CLUSTER cluster_name1]
         [, name2 [ON CLUSTER cluster_name2] ...]
-    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name'} | {WITH ssh_key BY KEY 'public_key' TYPE 'ssh-rsa|...'}]
+    [NOT IDENTIFIED | IDENTIFIED {[WITH {no_password | plaintext_password | sha256_password | sha256_hash | double_sha1_password | double_sha1_hash}] BY {'password' | 'hash'}} | {WITH ldap SERVER 'server_name'} | {WITH kerberos [REALM 'realm']} | {WITH ssl_certificate CN 'common_name' | SAN 'TYPE:subject_alt_name'} | {WITH ssh_key BY KEY 'public_key' TYPE 'ssh-rsa|...'}]
     [HOST {LOCAL | NAME 'name' | REGEXP 'name_regexp' | IP 'address' | LIKE 'pattern'} [,...] | ANY | NONE]
     [DEFAULT ROLE role [,...]]
     [DEFAULT DATABASE database | NONE]

programs/CMakeLists.txt

@@ -66,18 +66,18 @@ else()
     message(STATUS "Library bridge mode: OFF")
 endif()
 
-if (ENABLE_CLICKHOUSE_KEEPER)
-    message(STATUS "ClickHouse keeper mode: ON")
-else()
-    message(STATUS "ClickHouse keeper mode: OFF")
-endif()
-
 if (ENABLE_CLICKHOUSE_KEEPER_CONVERTER)
     message(STATUS "ClickHouse keeper-converter mode: ON")
 else()
     message(STATUS "ClickHouse keeper-converter mode: OFF")
 endif()
 
+if (ENABLE_CLICKHOUSE_KEEPER)
+    message(STATUS "ClickHouse Keeper: ON")
+else()
+    message(STATUS "ClickHouse Keeper: OFF")
+endif()
+
 if (ENABLE_CLICKHOUSE_KEEPER_CLIENT)
     message(STATUS "ClickHouse keeper-client mode: ON")
 else()
@@ -131,10 +131,6 @@ add_subdirectory (static-files-disk-uploader)
 add_subdirectory (su)
 add_subdirectory (disks)
 
-if (ENABLE_CLICKHOUSE_KEEPER)
-    add_subdirectory (keeper)
-endif()
-
 if (ENABLE_CLICKHOUSE_KEEPER_CONVERTER)
     add_subdirectory (keeper-converter)
 endif()
@@ -143,6 +139,10 @@ if (ENABLE_CLICKHOUSE_KEEPER_CLIENT)
     add_subdirectory (keeper-client)
 endif()
 
+if (ENABLE_CLICKHOUSE_KEEPER)
+    add_subdirectory (keeper)
+endif()
+
 if (ENABLE_CLICKHOUSE_ODBC_BRIDGE)
     add_subdirectory (odbc-bridge)
 endif ()

programs/client/Client.h

@@ -9,7 +9,10 @@ namespace DB
 class Client : public ClientBase
 {
 public:
-    Client() = default;
+    Client()
+    {
+        fuzzer = QueryFuzzer(randomSeed(), &std::cout, &std::cerr);
+    }
 
     void initialize(Poco::Util::Application & self) override;
 

programs/keeper/CMakeLists.txt

@@ -1,4 +1,5 @@
 set(CLICKHOUSE_KEEPER_SOURCES
+    keeper_main.cpp
     Keeper.cpp
 )
 
@@ -8,6 +9,9 @@ set (CLICKHOUSE_KEEPER_LINK
         clickhouse_common_io
         clickhouse_common_zookeeper
         daemon
+        clickhouse-keeper-converter-lib
+        clickhouse-keeper-client-lib
+        clickhouse_functions
         dbms
 )
 
@@ -17,199 +21,11 @@ install(FILES keeper_config.xml DESTINATION "${CLICKHOUSE_ETC_DIR}/clickhouse-ke
 
 if (BUILD_STANDALONE_KEEPER)
     # Straight list of all required sources
-    set(CLICKHOUSE_KEEPER_STANDALONE_SOURCES
+    clickhouse_add_executable(clickhouse-keeper ${CLICKHOUSE_KEEPER_SOURCES})
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperReconfiguration.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/RaftServerConfig.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/ACLMap.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/Changelog.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/CoordinationSettings.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/FourLetterCommand.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/InMemoryLogStore.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperConnectionStats.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperDispatcher.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperLogStore.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperServer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperContext.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperFeatureFlags.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperSnapshotManager.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperSnapshotManagerS3.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperStateMachine.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperContext.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperStateManager.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperConstants.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperAsynchronousMetrics.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/KeeperCommon.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/SessionExpiryQueue.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/SummingStateMachine.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/WriteBufferFromNuraftBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/ZooKeeperDataReader.cpp
 
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/SettingsFields.cpp
+    target_link_libraries(clickhouse-keeper PUBLIC ${CLICKHOUSE_KEEPER_LINK})
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/BaseSettings.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/ServerSettings.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/Field.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/SettingsEnums.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/ServerUUID.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/UUID.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Core/BackgroundSchedulePool.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/IO/ReadBuffer.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTPPathHints.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/KeeperTCPHandler.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/TCPServer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/NotFoundHandler.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/ProtocolServerAdapter.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/CertificateReloader.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/PrometheusRequestHandler.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/PrometheusMetricsWriter.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/waitServersToFinish.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/ServerType.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTPRequestHandlerFactoryMain.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/KeeperReadinessHandler.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/CloudPlacementInfo.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/HTTPServer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/ReadHeaders.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/HTTPServerConnection.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/HTTPServerRequest.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/HTTPServerResponse.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/HTTPServerConnectionFactory.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Server/HTTP/WriteBufferFromHTTPServerResponse.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CachedCompressedReadBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CheckingCompressedReadBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressedReadBufferBase.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressedReadBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressedReadBufferFromFile.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressedWriteBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionCodecEncrypted.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionCodecLZ4.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionCodecMultiple.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionCodecNone.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionCodecZSTD.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/CompressionFactory.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/ICompressionCodec.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Compression/LZ4_decompress_faster.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/CurrentThread.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/NamedCollections/NamedCollections.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/NamedCollections/NamedCollectionConfiguration.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/Jemalloc.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/IKeeper.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/TestKeeper.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperCommon.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperConstants.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeper.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperImpl.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperIO.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperLock.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Common/ZooKeeper/ZooKeeperNodeCache.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/registerDisks.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IDisk.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/DiskFactory.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/DiskSelector.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/DiskLocal.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/DiskLocalCheckThread.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/LocalDirectorySyncGuard.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/TemporaryFileOnDisk.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/loadLocalDiskConfig.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/DiskType.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/IObjectStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataOperationsHolder.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorageOperations.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFromPlainObjectStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFromPlainRewritableObjectStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFromDisk.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageTransactionState.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/DiskObjectStorageMetadata.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFromDiskTransactionOperations.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/DiskObjectStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/DiskObjectStorageTransaction.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/DiskObjectStorageRemoteMetadataRestoreHelper.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/ObjectStorageIteratorAsync.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/ObjectStorageIterator.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/StoredObject.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/S3/S3ObjectStorage.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/S3/S3Capabilities.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/S3/diskSettings.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/S3/DiskS3Utils.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/CommonPathPrefixKeyGenerator.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/ObjectStorageFactory.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/MetadataStorageFactory.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/ObjectStorages/RegisterDiskObjectStorage.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/createReadBufferFromFileBase.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/ReadBufferFromRemoteFSGather.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/IOUringReader.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/getIOUringReader.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/WriteBufferFromTemporaryFile.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/WriteBufferWithFinalizeCallback.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/AsynchronousBoundedReadBuffer.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/getThreadPoolReader.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/ThreadPoolRemoteFSReader.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Disks/IO/ThreadPoolReader.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Daemon/BaseDaemon.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Daemon/SentryWriter.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Daemon/GraphiteWriter.cpp
-        ${CMAKE_CURRENT_BINARY_DIR}/../../src/Daemon/GitHash.generated.cpp
-
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/Standalone/Context.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/Standalone/Settings.cpp
-        ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Coordination/Standalone/ThreadStatusExt.cpp
-
-        Keeper.cpp
-        clickhouse-keeper.cpp
-    )
-
-    # List of resources for clickhouse-keeper client
-    if (ENABLE_CLICKHOUSE_KEEPER_CLIENT)
-        list(APPEND CLICKHOUSE_KEEPER_STANDALONE_SOURCES
-            ${CMAKE_CURRENT_SOURCE_DIR}/../../programs/keeper-client/KeeperClient.cpp
-            ${CMAKE_CURRENT_SOURCE_DIR}/../../programs/keeper-client/Commands.cpp
-            ${CMAKE_CURRENT_SOURCE_DIR}/../../programs/keeper-client/Parser.cpp
-
-            ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Client/LineReader.cpp
-            ${CMAKE_CURRENT_SOURCE_DIR}/../../src/Client/ReplxxLineReader.cpp
-        )
-    endif()
-
-    clickhouse_add_executable(clickhouse-keeper ${CLICKHOUSE_KEEPER_STANDALONE_SOURCES})
-
-    # Remove some redundant dependencies
-    target_compile_definitions (clickhouse-keeper PRIVATE -DCLICKHOUSE_KEEPER_STANDALONE_BUILD)
-    target_compile_definitions (clickhouse-keeper PUBLIC -DWITHOUT_TEXT_LOG)
-
-    if (ENABLE_CLICKHOUSE_KEEPER_CLIENT AND TARGET ch_rust::skim)
-        target_link_libraries(clickhouse-keeper PRIVATE ch_rust::skim)
-    endif()
-
-    target_link_libraries(clickhouse-keeper
-        PRIVATE
-            ch_contrib::abseil_swiss_tables
-            ch_contrib::nuraft
-            ch_contrib::lz4
-            ch_contrib::zstd
-            ch_contrib::cityhash
-            ch_contrib::jemalloc
-            common ch_contrib::double_conversion
-            ch_contrib::dragonbox_to_chars
-            pcg_random
-            ch_contrib::pdqsort
-            ch_contrib::miniselect
-            clickhouse_common_config_no_zookeeper_log
-            loggers_no_text_log
-            clickhouse_common_io
-            clickhouse_parsers # Otherwise compression will not built. FIXME.
-    )
 
     set_target_properties(clickhouse-keeper PROPERTIES RUNTIME_OUTPUT_DIRECTORY ../)
-
     if (SPLIT_DEBUG_SYMBOLS)
         clickhouse_split_debug_symbols(TARGET clickhouse-keeper DESTINATION_DIR ${CMAKE_CURRENT_BINARY_DIR}/../${SPLITTED_DEBUG_SYMBOLS_DIR} BINARY_PATH ../clickhouse-keeper)
     else()

programs/keeper/Keeper.cpp

@@ -27,6 +27,8 @@
 #include <sys/stat.h>
 #include <pwd.h>
 
+#include <Common/Jemalloc.h>
+
 #include <Interpreters/Context.h>
 
 #include <Coordination/FourLetterCommand.h>
@@ -75,16 +77,6 @@ int mainEntryClickHouseKeeper(int argc, char ** argv)
     }
 }
 
-#ifdef CLICKHOUSE_KEEPER_STANDALONE_BUILD
-
-// Weak symbols don't work correctly on Darwin
-// so we have a stub implementation to avoid linker errors
-void collectCrashLog(
-    Int32, UInt64, const String &, const StackTrace &)
-{}
-
-#endif
-
 namespace DB
 {
 
@@ -277,6 +269,9 @@ HTTPContextPtr httpContext()
 int Keeper::main(const std::vector<std::string> & /*args*/)
 try
 {
+#if USE_JEMALLOC
+    setJemallocBackgroundThreads(true);
+#endif
     Poco::Logger * log = &logger();
 
     UseSSL use_ssl;

programs/keeper/clickhouse-keeper.cpp

@@ -1,30 +0,0 @@
-#include <Common/StringUtils.h>
-#include "config_tools.h"
-
-
-int mainEntryClickHouseKeeper(int argc, char ** argv);
-
-#if ENABLE_CLICKHOUSE_KEEPER_CLIENT
-int mainEntryClickHouseKeeperClient(int argc, char ** argv);
-#endif
-
-int main(int argc_, char ** argv_)
-{
-#if ENABLE_CLICKHOUSE_KEEPER_CLIENT
-
-    if (argc_ >= 2)
-    {
-        /// 'clickhouse-keeper --client ...' and 'clickhouse-keeper client ...' are OK
-        if (strcmp(argv_[1], "--client") == 0 || strcmp(argv_[1], "client") == 0)
-        {
-            argv_[1] = argv_[0];
-            return mainEntryClickHouseKeeperClient(--argc_, argv_ + 1);
-        }
-    }
-
-    if (argc_ > 0 && (strcmp(argv_[0], "clickhouse-keeper-client") == 0 || endsWith(argv_[0], "/clickhouse-keeper-client")))
-        return mainEntryClickHouseKeeperClient(argc_, argv_);
-#endif
-
-    return mainEntryClickHouseKeeper(argc_, argv_);
-}

programs/keeper/keeper_main.cpp

@@ -0,0 +1,189 @@
+#include <unistd.h>
+#include <fcntl.h>
+
+#include <new>
+#include <iostream>
+#include <vector>
+#include <string_view>
+#include <utility> /// pair
+
+#include <fmt/format.h>
+
+#include "config.h"
+#include "config_tools.h"
+
+#include <Common/EnvironmentChecks.h>
+#include <Common/Coverage.h>
+
+#include <Common/StringUtils.h>
+#include <Common/getHashOfLoadedBinary.h>
+#include <Common/IO.h>
+
+#include <base/phdr_cache.h>
+#include <base/coverage.h>
+
+
+int mainEntryClickHouseKeeper(int argc, char ** argv);
+#if ENABLE_CLICKHOUSE_KEEPER_CONVERTER
+int mainEntryClickHouseKeeperConverter(int argc, char ** argv);
+#endif
+#if ENABLE_CLICKHOUSE_KEEPER_CLIENT
+int mainEntryClickHouseKeeperClient(int argc, char ** argv);
+#endif
+
+namespace
+{
+
+using MainFunc = int (*)(int, char**);
+
+/// Add an item here to register new application
+std::pair<std::string_view, MainFunc> clickhouse_applications[] =
+{
+    // keeper
+    {"keeper", mainEntryClickHouseKeeper},
+#if ENABLE_CLICKHOUSE_KEEPER_CONVERTER
+    {"converter", mainEntryClickHouseKeeperConverter},
+    {"keeper-converter", mainEntryClickHouseKeeperConverter},
+#endif
+#if ENABLE_CLICKHOUSE_KEEPER_CLIENT
+    {"client", mainEntryClickHouseKeeperClient},
+    {"keeper-client", mainEntryClickHouseKeeperClient},
+#endif
+
+};
+
+int printHelp(int, char **)
+{
+    std::cerr << "Use one of the following commands:" << std::endl;
+    for (auto & application : clickhouse_applications)
+        std::cerr << "clickhouse " << application.first << " [args] " << std::endl;
+    return -1;
+}
+
+}
+
+
+bool isClickhouseApp(std::string_view app_suffix, std::vector<char *> & argv)
+{
+    /// Use app if the first arg 'app' is passed (the arg should be quietly removed)
+    if (argv.size() >= 2)
+    {
+        auto first_arg = argv.begin() + 1;
+
+        /// 'clickhouse --client ...' and 'clickhouse client ...' are Ok
+        if (*first_arg == app_suffix
+            || (std::string_view(*first_arg).starts_with("--") && std::string_view(*first_arg).substr(2) == app_suffix))
+        {
+            argv.erase(first_arg);
+            return true;
+        }
+    }
+
+    /// keeper suffix is default which will be used if no other app is detected
+    if (app_suffix == "keeper")
+        return false;
+
+    /// Use app if clickhouse binary is run through symbolic link with name clickhouse-app
+    std::string app_name = "clickhouse-" + std::string(app_suffix);
+    return !argv.empty() && (app_name == argv[0] || endsWith(argv[0], "/" + app_name));
+}
+
+/// Don't allow dlopen in the main ClickHouse binary, because it is harmful and insecure.
+/// We don't use it. But it can be used by some libraries for implementation of "plugins".
+/// We absolutely discourage the ancient technique of loading
+/// 3rd-party uncontrolled dangerous libraries into the process address space,
+/// because it is insane.
+
+#if !defined(USE_MUSL)
+extern "C"
+{
+    void * dlopen(const char *, int)
+    {
+        return nullptr;
+    }
+
+    void * dlmopen(long, const char *, int) // NOLINT
+    {
+        return nullptr;
+    }
+
+    int dlclose(void *)
+    {
+        return 0;
+    }
+
+    const char * dlerror()
+    {
+        return "ClickHouse does not allow dynamic library loading";
+    }
+}
+#endif
+
+/// Prevent messages from JeMalloc in the release build.
+/// Some of these messages are non-actionable for the users, such as:
+/// <jemalloc>: Number of CPUs detected is not deterministic. Per-CPU arena disabled.
+#if USE_JEMALLOC && defined(NDEBUG) && !defined(SANITIZER)
+extern "C" void (*malloc_message)(void *, const char *s);
+__attribute__((constructor(0))) void init_je_malloc_message() { malloc_message = [](void *, const char *){}; }
+#endif
+
+/// This allows to implement assert to forbid initialization of a class in static constructors.
+/// Usage:
+///
+/// extern bool inside_main;
+/// class C { C() { assert(inside_main); } };
+bool inside_main = false;
+
+int main(int argc_, char ** argv_)
+{
+    inside_main = true;
+    SCOPE_EXIT({ inside_main = false; });
+
+    /// PHDR cache is required for query profiler to work reliably
+    /// It also speed up exception handling, but exceptions from dynamically loaded libraries (dlopen)
+    ///  will work only after additional call of this function.
+    /// Note: we forbid dlopen in our code.
+    updatePHDRCache();
+
+#if !defined(USE_MUSL)
+    checkHarmfulEnvironmentVariables(argv_);
+#endif
+
+    /// This is used for testing. For example,
+    /// clickhouse-local should be able to run a simple query without throw/catch.
+    if (getenv("CLICKHOUSE_TERMINATE_ON_ANY_EXCEPTION")) // NOLINT(concurrency-mt-unsafe)
+        DB::terminate_on_any_exception = true;
+
+    /// Reset new handler to default (that throws std::bad_alloc)
+    /// It is needed because LLVM library clobbers it.
+    std::set_new_handler(nullptr);
+
+    std::vector<char *> argv(argv_, argv_ + argc_);
+
+    /// Print a basic help if nothing was matched
+    MainFunc main_func = mainEntryClickHouseKeeper;
+
+    if (isClickhouseApp("help", argv))
+    {
+        main_func = printHelp;
+    }
+    else
+    {
+        for (auto & application : clickhouse_applications)
+        {
+            if (isClickhouseApp(application.first, argv))
+            {
+                main_func = application.second;
+                break;
+            }
+        }
+    }
+
+    int exit_code = main_func(static_cast<int>(argv.size()), argv.data());
+
+#if defined(SANITIZE_COVERAGE)
+    dumpCoverage();
+#endif
+
+    return exit_code;
+}

programs/main.cpp

@@ -1,5 +1,3 @@
-#include <csignal>
-#include <csetjmp>
 #include <unistd.h>
 #include <fcntl.h>
 
@@ -7,7 +5,6 @@
 #include <iostream>
 #include <vector>
 #include <string>
-#include <tuple>
 #include <string_view>
 #include <utility> /// pair
 
@@ -16,6 +13,8 @@
 #include "config.h"
 #include "config_tools.h"
 
+#include <Common/EnvironmentChecks.h>
+#include <Common/Coverage.h>
 #include <Common/StringUtils.h>
 #include <Common/getHashOfLoadedBinary.h>
 #include <Common/IO.h>
@@ -119,268 +118,6 @@ std::pair<std::string_view, std::string_view> clickhouse_short_names[] =
     {"chc", "client"},
 };
 
-
-enum class InstructionFail : uint8_t
-{
-    NONE = 0,
-    SSE3 = 1,
-    SSSE3 = 2,
-    SSE4_1 = 3,
-    SSE4_2 = 4,
-    POPCNT = 5,
-    AVX = 6,
-    AVX2 = 7,
-    AVX512 = 8
-};
-
-auto instructionFailToString(InstructionFail fail)
-{
-    switch (fail)
-    {
-#define ret(x) return std::make_tuple(STDERR_FILENO, x, sizeof(x) - 1)
-        case InstructionFail::NONE:
-            ret("NONE");
-        case InstructionFail::SSE3:
-            ret("SSE3");
-        case InstructionFail::SSSE3:
-            ret("SSSE3");
-        case InstructionFail::SSE4_1:
-            ret("SSE4.1");
-        case InstructionFail::SSE4_2:
-            ret("SSE4.2");
-        case InstructionFail::POPCNT:
-            ret("POPCNT");
-        case InstructionFail::AVX:
-            ret("AVX");
-        case InstructionFail::AVX2:
-            ret("AVX2");
-        case InstructionFail::AVX512:
-            ret("AVX512");
-#undef ret
-    }
-}
-
-
-sigjmp_buf jmpbuf;
-
-[[noreturn]] void sigIllCheckHandler(int, siginfo_t *, void *)
-{
-    siglongjmp(jmpbuf, 1);
-}
-
-/// Check if necessary SSE extensions are available by trying to execute some sse instructions.
-/// If instruction is unavailable, SIGILL will be sent by kernel.
-void checkRequiredInstructionsImpl(volatile InstructionFail & fail)
-{
-#if defined(__SSE3__)
-    fail = InstructionFail::SSE3;
-    __asm__ volatile ("addsubpd %%xmm0, %%xmm0" : : : "xmm0");
-#endif
-
-#if defined(__SSSE3__)
-    fail = InstructionFail::SSSE3;
-    __asm__ volatile ("pabsw %%xmm0, %%xmm0" : : : "xmm0");
-
-#endif
-
-#if defined(__SSE4_1__)
-    fail = InstructionFail::SSE4_1;
-    __asm__ volatile ("pmaxud %%xmm0, %%xmm0" : : : "xmm0");
-#endif
-
-#if defined(__SSE4_2__)
-    fail = InstructionFail::SSE4_2;
-    __asm__ volatile ("pcmpgtq %%xmm0, %%xmm0" : : : "xmm0");
-#endif
-
-    /// Defined by -msse4.2
-#if defined(__POPCNT__)
-    fail = InstructionFail::POPCNT;
-    {
-        uint64_t a = 0;
-        uint64_t b = 0;
-        __asm__ volatile ("popcnt %1, %0" : "=r"(a) :"r"(b) :);
-    }
-#endif
-
-#if defined(__AVX__)
-    fail = InstructionFail::AVX;
-    __asm__ volatile ("vaddpd %%ymm0, %%ymm0, %%ymm0" : : : "ymm0");
-#endif
-
-#if defined(__AVX2__)
-    fail = InstructionFail::AVX2;
-    __asm__ volatile ("vpabsw %%ymm0, %%ymm0" : : : "ymm0");
-#endif
-
-#if defined(__AVX512__)
-    fail = InstructionFail::AVX512;
-    __asm__ volatile ("vpabsw %%zmm0, %%zmm0" : : : "zmm0");
-#endif
-
-    fail = InstructionFail::NONE;
-}
-
-/// Macros to avoid using strlen(), since it may fail if SSE is not supported.
-#define writeError(data) do \
-    { \
-        static_assert(__builtin_constant_p(data)); \
-        if (!writeRetry(STDERR_FILENO, data, sizeof(data) - 1)) \
-            _Exit(1); \
-    } while (false)
-
-/// Check SSE and others instructions availability. Calls exit on fail.
-/// This function must be called as early as possible, even before main, because static initializers may use unavailable instructions.
-void checkRequiredInstructions()
-{
-    struct sigaction sa{};
-    struct sigaction sa_old{};
-    sa.sa_sigaction = sigIllCheckHandler;
-    sa.sa_flags = SA_SIGINFO;
-    auto signal = SIGILL;
-    if (sigemptyset(&sa.sa_mask) != 0
-        || sigaddset(&sa.sa_mask, signal) != 0
-        || sigaction(signal, &sa, &sa_old) != 0)
-    {
-        /// You may wonder about strlen.
-        /// Typical implementation of strlen is using SSE4.2 or AVX2.
-        /// But this is not the case because it's compiler builtin and is executed at compile time.
-
-        writeError("Can not set signal handler\n");
-        _Exit(1);
-    }
-
-    volatile InstructionFail fail = InstructionFail::NONE;
-
-    if (sigsetjmp(jmpbuf, 1))
-    {
-        writeError("Instruction check fail. The CPU does not support ");
-        if (!std::apply(writeRetry, instructionFailToString(fail)))
-            _Exit(1);
-        writeError(" instruction set.\n");
-        _Exit(1);
-    }
-
-    checkRequiredInstructionsImpl(fail);
-
-    if (sigaction(signal, &sa_old, nullptr))
-    {
-        writeError("Can not set signal handler\n");
-        _Exit(1);
-    }
-}
-
-struct Checker
-{
-    Checker()
-    {
-        checkRequiredInstructions();
-    }
-} checker
-#ifndef OS_DARWIN
-    __attribute__((init_priority(101)))    /// Run before other static initializers.
-#endif
-;
-
-
-#if !defined(USE_MUSL)
-/// NOTE: We will migrate to full static linking or our own dynamic loader to make this code obsolete.
-void checkHarmfulEnvironmentVariables(char ** argv)
-{
-    std::initializer_list<const char *> harmful_env_variables = {
-        /// The list is a selection from "man ld-linux".
-        "LD_PRELOAD",
-        "LD_LIBRARY_PATH",
-        "LD_ORIGIN_PATH",
-        "LD_AUDIT",
-        "LD_DYNAMIC_WEAK",
-        /// The list is a selection from "man dyld" (osx).
-        "DYLD_LIBRARY_PATH",
-        "DYLD_FALLBACK_LIBRARY_PATH",
-        "DYLD_VERSIONED_LIBRARY_PATH",
-        "DYLD_INSERT_LIBRARIES",
-    };
-
-    bool require_reexec = false;
-    for (const auto * var : harmful_env_variables)
-    {
-        if (const char * value = getenv(var); value && value[0]) // NOLINT(concurrency-mt-unsafe)
-        {
-            /// NOTE: setenv() is used over unsetenv() since unsetenv() marked as harmful
-            if (setenv(var, "", true)) // NOLINT(concurrency-mt-unsafe) // this is safe if not called concurrently
-            {
-                fmt::print(stderr, "Cannot override {} environment variable", var);
-                _exit(1);
-            }
-            require_reexec = true;
-        }
-    }
-
-    if (require_reexec)
-    {
-        /// Use execvp() over execv() to search in PATH.
-        ///
-        /// This should be safe, since:
-        /// - if argv[0] is relative path - it is OK
-        /// - if argv[0] has only basename, the it will search in PATH, like shell will do.
-        ///
-        /// Also note, that this (search in PATH) because there is no easy and
-        /// portable way to get absolute path of argv[0].
-        /// - on linux there is /proc/self/exec and AT_EXECFN
-        /// - but on other OSes there is no such thing (especially on OSX).
-        ///
-        /// And since static linking will be done someday anyway,
-        /// let's not pollute the code base with special cases.
-        int error = execvp(argv[0], argv);
-        _exit(error);
-    }
-}
-#endif
-
-
-#if defined(SANITIZE_COVERAGE)
-__attribute__((no_sanitize("coverage")))
-void dumpCoverage()
-{
-    /// A user can request to dump the coverage information into files at exit.
-    /// This is useful for non-server applications such as clickhouse-format or clickhouse-client,
-    /// that cannot introspect it with SQL functions at runtime.
-
-    /// The CLICKHOUSE_WRITE_COVERAGE environment variable defines a prefix for a filename 'prefix.pid'
-    /// containing the list of addresses of covered .
-
-    /// The format is even simpler than Clang's "sancov": an array of 64-bit addresses, native byte order, no header.
-
-    if (const char * coverage_filename_prefix = getenv("CLICKHOUSE_WRITE_COVERAGE")) // NOLINT(concurrency-mt-unsafe)
-    {
-        auto dump = [](const std::string & name, auto span)
-        {
-            /// Write only non-zeros.
-            std::vector<uintptr_t> data;
-            data.reserve(span.size());
-            for (auto addr : span)
-                if (addr)
-                    data.push_back(addr);
-
-            int fd = ::open(name.c_str(), O_WRONLY | O_CREAT | O_TRUNC | O_CLOEXEC, 0400);
-            if (-1 == fd)
-            {
-                writeError("Cannot open a file to write the coverage data\n");
-            }
-            else
-            {
-                if (!writeRetry(fd, reinterpret_cast<const char *>(data.data()), data.size() * sizeof(data[0])))
-                    writeError("Cannot write the coverage data to a file\n");
-                if (0 != ::close(fd))
-                    writeError("Cannot close the file with coverage data\n");
-            }
-        };
-
-        dump(fmt::format("{}.{}", coverage_filename_prefix, getpid()), getCumulativeCoverage());
-    }
-}
-#endif
-
 }
 
 bool isClickhouseApp(std::string_view app_suffix, std::vector<char *> & argv)

programs/odbc-bridge/ODBCSource.cpp

@@ -3,6 +3,7 @@
 #include <IO/ReadBufferFromString.h>
 #include <DataTypes/DataTypeNullable.h>
 #include <DataTypes/DataTypeDateTime64.h>
+#include <Columns/ColumnNullable.h>
 #include <Common/assert_cast.h>
 #include <IO/ReadHelpers.h>
 
@@ -47,9 +48,17 @@ Chunk ODBCSource::generate()
         for (int idx = 0; idx < result.columns(); ++idx)
         {
             const auto & sample = description.sample_block.getByPosition(idx);
-
             if (!result.is_null(idx))
-                insertValue(*columns[idx], removeNullable(sample.type), description.types[idx].first, result, idx);
+            {
+                if (columns[idx]->isNullable())
+                {
+                    ColumnNullable & column_nullable = assert_cast<ColumnNullable &>(*columns[idx]);
+                    insertValue(column_nullable.getNestedColumn(), removeNullable(sample.type), description.types[idx].first, result, idx);
+                    column_nullable.getNullMapData().emplace_back(0);
+                }
+                else
+                     insertValue(*columns[idx], removeNullable(sample.type), description.types[idx].first, result, idx);
+            }
             else
                 insertDefaultValue(*columns[idx], *sample.column);
         }

programs/self-extracting/CMakeLists.txt

@@ -10,9 +10,24 @@ else ()
     set (COMPRESSOR "${PROJECT_BINARY_DIR}/utils/self-extracting-executable/compressor")
 endif ()
 
-add_custom_target (self-extracting ALL
+add_custom_target (self-extracting-server ALL
     ${CMAKE_COMMAND} -E remove clickhouse clickhouse-stripped
     COMMAND ${COMPRESSOR} ${DECOMPRESSOR} clickhouse ../clickhouse
     COMMAND ${COMPRESSOR} ${DECOMPRESSOR} clickhouse-stripped ../clickhouse-stripped
     DEPENDS clickhouse clickhouse-stripped compressor
 )
+
+set(self_extracting_deps "self-extracting-server")
+
+if (BUILD_STANDALONE_KEEPER)
+    add_custom_target (self-extracting-keeper ALL
+        ${CMAKE_COMMAND} -E remove clickhouse-keeper
+        COMMAND ${COMPRESSOR} ${DECOMPRESSOR} clickhouse-keeper ../clickhouse-keeper
+        DEPENDS compressor clickhouse-keeper
+    )
+    list(APPEND self_extracting_deps "self-extracting-keeper")
+endif()
+
+add_custom_target (self-extracting ALL
+    DEPENDS ${self_extracting_deps}
+)

programs/server/Server.cpp

@@ -11,6 +11,7 @@
 #include <Poco/Util/HelpFormatter.h>
 #include <Poco/Environment.h>
 #include <Poco/Config.h>
+#include <Common/Jemalloc.h>
 #include <Common/scope_guard_safe.h>
 #include <Common/logger_useful.h>
 #include <base/phdr_cache.h>
@@ -133,10 +134,6 @@
 #    include <Server/KeeperTCPHandlerFactory.h>
 #endif
 
-#if USE_JEMALLOC
-#    include <jemalloc/jemalloc.h>
-#endif
-
 #if USE_AZURE_BLOB_STORAGE
 #   include <azure/storage/common/internal/xml_wrapper.hpp>
 #   include <azure/core/diagnostics/logger.hpp>
@@ -176,34 +173,10 @@ namespace ProfileEvents
 
 namespace fs = std::filesystem;
 
-#if USE_JEMALLOC
-static bool jemallocOptionEnabled(const char *name)
-{
-    bool value;
-    size_t size = sizeof(value);
-
-    if (mallctl(name, reinterpret_cast<void *>(&value), &size, /* newp= */ nullptr, /* newlen= */ 0))
-        throw Poco::SystemException("mallctl() failed");
-
-    return value;
-}
-#else
-static bool jemallocOptionEnabled(const char *) { return false; }
-#endif
-
 int mainEntryClickHouseServer(int argc, char ** argv)
 {
     DB::Server app;
 
-    if (jemallocOptionEnabled("opt.background_thread"))
-    {
-        LOG_ERROR(&app.logger(),
-            "jemalloc.background_thread was requested, "
-            "however ClickHouse uses percpu_arena and background_thread most likely will not give any benefits, "
-            "and also background_thread is not compatible with ClickHouse watchdog "
-            "(that can be disabled with CLICKHOUSE_WATCHDOG_ENABLE=0)");
-    }
-
     /// Do not fork separate process from watchdog if we attached to terminal.
     /// Otherwise it breaks gdb usage.
     /// Can be overridden by environment variable (cannot use server config at this moment).
@@ -656,6 +629,10 @@ static void initializeAzureSDKLogger(
 int Server::main(const std::vector<std::string> & /*args*/)
 try
 {
+#if USE_JEMALLOC
+    setJemallocBackgroundThreads(true);
+#endif
+
     Stopwatch startup_watch;
 
     Poco::Logger * log = &logger();

programs/server/dashboard.html

@@ -506,6 +506,14 @@ let user = 'default';
 let password = '';
 let add_http_cors_header = (location.protocol != 'file:');
 
+const current_url = new URL(window.location);
+/// Substitute user name if it's specified in the query string
+const user_from_url = current_url.searchParams.get('user');
+if (user_from_url) {
+    user = user_from_url;
+}
+
+
 const errorCodeMessageMap = {
     516: 'Error authenticating with database. Please check your connection params and try again.'
 }

src/Access/Authentication.cpp

@@ -8,6 +8,7 @@
 #include <Common/Exception.h>
 #include <Common/SSHWrapper.h>
 #include <Common/typeid_cast.h>
+#include <Access/Common/SSLCertificateSubjects.h>
 
 #include "config.h"
 
@@ -238,7 +239,15 @@ bool Authentication::areCredentialsValid(
                 throw Authentication::Require<GSSAcceptorContext>(auth_data.getKerberosRealm());
 
             case AuthenticationType::SSL_CERTIFICATE:
-                return auth_data.getSSLCertificateCommonNames().contains(ssl_certificate_credentials->getCommonName());
+                for (SSLCertificateSubjects::Type type : {SSLCertificateSubjects::Type::CN, SSLCertificateSubjects::Type::SAN})
+                {
+                    for (const auto & subject : auth_data.getSSLCertificateSubjects().at(type))
+                    {
+                        if (ssl_certificate_credentials->getSSLCertificateSubjects().at(type).contains(subject))
+                            return true;
+                    }
+                }
+                return false;
 
             case AuthenticationType::SSH_KEY:
 #if USE_SSH

src/Access/AuthenticationData.cpp

@@ -15,6 +15,7 @@
 #include <boost/algorithm/hex.hpp>
 #include <boost/algorithm/string/case_conv.hpp>
 
+#include <Access/Common/SSLCertificateSubjects.h>
 #include "config.h"
 
 #if USE_SSL
@@ -107,7 +108,7 @@ bool operator ==(const AuthenticationData & lhs, const AuthenticationData & rhs)
 {
     return (lhs.type == rhs.type) && (lhs.password_hash == rhs.password_hash)
         && (lhs.ldap_server_name == rhs.ldap_server_name) && (lhs.kerberos_realm == rhs.kerberos_realm)
-        && (lhs.ssl_certificate_common_names == rhs.ssl_certificate_common_names)
+        && (lhs.ssl_certificate_subjects == rhs.ssl_certificate_subjects)
 #if USE_SSH
         && (lhs.ssh_keys == rhs.ssh_keys)
 #endif
@@ -277,11 +278,16 @@ String AuthenticationData::getSalt() const
     return salt;
 }
 
-void AuthenticationData::setSSLCertificateCommonNames(boost::container::flat_set<String> common_names_)
+void AuthenticationData::setSSLCertificateSubjects(SSLCertificateSubjects && ssl_certificate_subjects_)
 {
-    if (common_names_.empty())
+    if (ssl_certificate_subjects_.empty())
-        throw Exception(ErrorCodes::BAD_ARGUMENTS, "The 'SSL CERTIFICATE' authentication type requires a non-empty list of common names.");
+        throw Exception(ErrorCodes::BAD_ARGUMENTS, "The 'SSL CERTIFICATE' authentication type requires a non-empty list of subjects.");
-    ssl_certificate_common_names = std::move(common_names_);
+    ssl_certificate_subjects = std::move(ssl_certificate_subjects_);
+}
+
+void AuthenticationData::addSSLCertificateSubject(SSLCertificateSubjects::Type type_, String && subject_)
+{
+    ssl_certificate_subjects.insert(type_, std::move(subject_));
 }
 
 std::shared_ptr<ASTAuthenticationData> AuthenticationData::toAST() const
@@ -339,7 +345,14 @@ std::shared_ptr<ASTAuthenticationData> AuthenticationData::toAST() const
         }
         case AuthenticationType::SSL_CERTIFICATE:
         {
-            for (const auto & name : getSSLCertificateCommonNames())
+            using SSLCertificateSubjects::Type::CN;
+            using SSLCertificateSubjects::Type::SAN;
+
+            const auto &subjects = getSSLCertificateSubjects();
+            SSLCertificateSubjects::Type cert_subject_type = !subjects.at(SAN).empty() ? SAN : CN;
+
+            node->ssl_cert_subject_type = toString(cert_subject_type);
+            for (const auto & name : getSSLCertificateSubjects().at(cert_subject_type))
                 node->children.push_back(std::make_shared<ASTLiteral>(name));
 
             break;
@@ -513,11 +526,9 @@ AuthenticationData AuthenticationData::fromAST(const ASTAuthenticationData & que
     }
     else if (query.type == AuthenticationType::SSL_CERTIFICATE)
     {
-        boost::container::flat_set<String> common_names;
+        auto ssl_cert_subject_type = parseSSLCertificateSubjectType(*query.ssl_cert_subject_type);
         for (const auto & arg : args)
-            common_names.insert(checkAndGetLiteralArgument<String>(arg, "common_name"));
+            auth_data.addSSLCertificateSubject(ssl_cert_subject_type, checkAndGetLiteralArgument<String>(arg, "ssl_certificate_subject"));
-
-        auth_data.setSSLCertificateCommonNames(std::move(common_names));
     }
     else if (query.type == AuthenticationType::HTTP)
     {

src/Access/AuthenticationData.h

@@ -2,13 +2,14 @@
 
 #include <Access/Common/AuthenticationType.h>
 #include <Access/Common/HTTPAuthenticationScheme.h>
+#include <Access/Common/SSLCertificateSubjects.h>
 #include <Common/SSHWrapper.h>
 #include <Interpreters/Context_fwd.h>
 #include <Parsers/Access/ASTAuthenticationData.h>
 
 #include <vector>
 #include <base/types.h>
-#include <boost/container/flat_set.hpp>
+
 
 #include "config.h"
 
@@ -58,8 +59,9 @@ public:
     const String & getKerberosRealm() const { return kerberos_realm; }
     void setKerberosRealm(const String & realm) { kerberos_realm = realm; }
 
-    const boost::container::flat_set<String> & getSSLCertificateCommonNames() const { return ssl_certificate_common_names; }
+    const SSLCertificateSubjects & getSSLCertificateSubjects() const { return ssl_certificate_subjects; }
-    void setSSLCertificateCommonNames(boost::container::flat_set<String> common_names_);
+    void setSSLCertificateSubjects(SSLCertificateSubjects && ssl_certificate_subjects_);
+    void addSSLCertificateSubject(SSLCertificateSubjects::Type type_, String && subject_);
 
 #if USE_SSH
     const std::vector<SSHKey> & getSSHKeys() const { return ssh_keys; }
@@ -96,7 +98,7 @@ private:
     Digest password_hash;
     String ldap_server_name;
     String kerberos_realm;
-    boost::container::flat_set<String> ssl_certificate_common_names;
+    SSLCertificateSubjects ssl_certificate_subjects;
     String salt;
 #if USE_SSH
     std::vector<SSHKey> ssh_keys;

src/Access/Common/SSLCertificateSubjects.cpp

@@ -0,0 +1,95 @@
+#include <Access/Common/SSLCertificateSubjects.h>
+#include <Common/Exception.h>
+
+#if USE_SSL
+#include <openssl/x509v3.h>
+#endif
+
+namespace DB
+{
+namespace ErrorCodes
+{
+    extern const int BAD_ARGUMENTS;
+}
+
+#if USE_SSL
+SSLCertificateSubjects extractSSLCertificateSubjects(const Poco::Net::X509Certificate & certificate)
+{
+
+    SSLCertificateSubjects subjects;
+    if (!certificate.commonName().empty())
+    {
+        subjects.insert(SSLCertificateSubjects::Type::CN, certificate.commonName());
+    }
+
+#pragma clang diagnostic push
+#pragma clang diagnostic ignored "-Wused-but-marked-unused"
+    auto stackof_general_name_deleter = [](void * ptr) { GENERAL_NAMES_free(static_cast<STACK_OF(GENERAL_NAME) *>(ptr)); };
+    std::unique_ptr<void, decltype(stackof_general_name_deleter)> cert_names(
+        X509_get_ext_d2i(const_cast<X509 *>(certificate.certificate()), NID_subject_alt_name, nullptr, nullptr),
+        stackof_general_name_deleter);
+
+    if (STACK_OF(GENERAL_NAME) * names = static_cast<STACK_OF(GENERAL_NAME) *>(cert_names.get()))
+    {
+        for (int i = 0; i < sk_GENERAL_NAME_num(names); ++i)
+        {
+            const GENERAL_NAME * name = sk_GENERAL_NAME_value(names, i);
+            if (name->type == GEN_DNS || name->type == GEN_URI)
+            {
+                const char * data = reinterpret_cast<const char *>(ASN1_STRING_get0_data(name->d.ia5));
+                std::size_t len = ASN1_STRING_length(name->d.ia5);
+                std::string subject = (name->type == GEN_DNS ? "DNS:" : "URI:") + std::string(data, len);
+                subjects.insert(SSLCertificateSubjects::Type::SAN, std::move(subject));
+            }
+        }
+    }
+
+#pragma clang diagnostic pop
+    return subjects;
+}
+#endif
+
+
+void SSLCertificateSubjects::insert(const String & subject_type_, String && subject)
+{
+    insert(parseSSLCertificateSubjectType(subject_type_), std::move(subject));
+}
+
+void SSLCertificateSubjects::insert(Type subject_type_, String && subject)
+{
+    subjects[static_cast<size_t>(subject_type_)].insert(std::move(subject));
+}
+
+SSLCertificateSubjects::Type parseSSLCertificateSubjectType(const String & type_)
+{
+    if (type_ == "CN")
+        return SSLCertificateSubjects::Type::CN;
+    if (type_ == "SAN")
+        return SSLCertificateSubjects::Type::SAN;
+
+    throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unknown SSL Certificate Subject Type: {}", type_);
+}
+
+String toString(SSLCertificateSubjects::Type type_)
+{
+    switch (type_)
+    {
+        case SSLCertificateSubjects::Type::CN:
+            return "CN";
+        case SSLCertificateSubjects::Type::SAN:
+            return "SAN";
+    }
+}
+
+bool operator==(const SSLCertificateSubjects & lhs, const SSLCertificateSubjects & rhs)
+{
+    for (SSLCertificateSubjects::Type type : {SSLCertificateSubjects::Type::CN, SSLCertificateSubjects::Type::SAN})
+    {
+        if (lhs.at(type) != rhs.at(type))
+            return false;
+    }
+    return true;
+}
+
+}
+

src/Access/Common/SSLCertificateSubjects.h

@@ -0,0 +1,48 @@
+#pragma once
+
+#include "config.h"
+#include <base/types.h>
+#include <boost/container/flat_set.hpp>
+
+#if USE_SSL
+#    include <Poco/Net/X509Certificate.h>
+#endif
+
+namespace DB
+{
+class SSLCertificateSubjects
+{
+public:
+    using container = boost::container::flat_set<String>;
+    enum class Type
+    {
+        CN,
+        SAN
+    };
+
+private:
+    std::array<container, size_t(Type::SAN) + 1> subjects;
+
+public:
+    inline const container & at(Type type_) const { return subjects[static_cast<size_t>(type_)]; }
+    inline bool empty()
+    {
+        for (auto & subject_list : subjects)
+        {
+            if (!subject_list.empty())
+                return false;
+        }
+        return true;
+    }
+    void insert(const String & subject_type_, String && subject);
+    void insert(Type type_, String && subject);
+    friend bool operator==(const SSLCertificateSubjects & lhs, const SSLCertificateSubjects & rhs);
+};
+
+String toString(SSLCertificateSubjects::Type type_);
+SSLCertificateSubjects::Type parseSSLCertificateSubjectType(const String & type_);
+
+#if USE_SSL
+SSLCertificateSubjects extractSSLCertificateSubjects(const Poco::Net::X509Certificate & certificate);
+#endif
+}

src/Access/Credentials.cpp

@@ -1,7 +1,7 @@
 #include <Access/Credentials.h>
+#include <Access/Common/SSLCertificateSubjects.h>
 #include <Common/Exception.h>
 
-
 namespace DB
 {
 
@@ -48,18 +48,18 @@ void AlwaysAllowCredentials::setUserName(const String & user_name_)
     user_name = user_name_;
 }
 
-SSLCertificateCredentials::SSLCertificateCredentials(const String & user_name_, const String & common_name_)
+SSLCertificateCredentials::SSLCertificateCredentials(const String & user_name_, SSLCertificateSubjects && subjects_)
     : Credentials(user_name_)
-    , common_name(common_name_)
+    , certificate_subjects(subjects_)
 {
     is_ready = true;
 }
 
-const String & SSLCertificateCredentials::getCommonName() const
+const SSLCertificateSubjects & SSLCertificateCredentials::getSSLCertificateSubjects() const
 {
     if (!isReady())
         throwNotReady();
-    return common_name;
+    return certificate_subjects;
 }
 
 BasicCredentials::BasicCredentials()

src/Access/Credentials.h

@@ -1,6 +1,8 @@
 #pragma once
 
 #include <base/types.h>
+#include <boost/container/flat_set.hpp>
+#include <Access/Common/SSLCertificateSubjects.h>
 #include <memory>
 
 #include "config.h"
@@ -42,11 +44,11 @@ class SSLCertificateCredentials
     : public Credentials
 {
 public:
-    explicit SSLCertificateCredentials(const String & user_name_, const String & common_name_);
+    explicit SSLCertificateCredentials(const String & user_name_, SSLCertificateSubjects && subjects_);
-    const String & getCommonName() const;
+    const SSLCertificateSubjects & getSSLCertificateSubjects() const;
 
 private:
-    String common_name;
+    SSLCertificateSubjects certificate_subjects;
 };
 
 class BasicCredentials

src/Access/UsersConfigAccessStorage.cpp

@@ -1,4 +1,5 @@
 #include <Access/UsersConfigAccessStorage.h>
+#include <Access/Common/SSLCertificateSubjects.h>
 #include <Access/Quota.h>
 #include <Access/RowPolicy.h>
 #include <Access/User.h>
@@ -194,18 +195,23 @@ namespace
             /// Fill list of allowed certificates.
             Poco::Util::AbstractConfiguration::Keys keys;
             config.keys(certificates_config, keys);
-            boost::container::flat_set<String> common_names;
             for (const String & key : keys)
             {
                 if (key.starts_with("common_name"))
                 {
                     String value = config.getString(certificates_config + "." + key);
-                    common_names.insert(std::move(value));
+                    user->auth_data.addSSLCertificateSubject(SSLCertificateSubjects::Type::CN, std::move(value));
+                }
+                else if (key.starts_with("subject_alt_name"))
+                {
+                    String value = config.getString(certificates_config + "." + key);
+                    if (value.empty())
+                        throw Exception(ErrorCodes::BAD_ARGUMENTS, "Expected ssl_certificates.subject_alt_name to not be empty");
+                    user->auth_data.addSSLCertificateSubject(SSLCertificateSubjects::Type::SAN, std::move(value));
                 }
                 else
                     throw Exception(ErrorCodes::BAD_ARGUMENTS, "Unknown certificate pattern type: {}", key);
             }
-            user->auth_data.setSSLCertificateCommonNames(std::move(common_names));
         }
         else if (has_ssh_keys)
         {

src/Analyzer/Passes/AggregateFunctionsArithmericOperationsPass.cpp

@@ -9,6 +9,7 @@
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Analyzer/ConstantNode.h>
 #include <Analyzer/FunctionNode.h>
+#include <Analyzer/Utils.h>
 
 namespace DB
 {
@@ -164,32 +165,15 @@ private:
 
         auto aggregate_function_clone = aggregate_function->clone();
         auto & aggregate_function_clone_typed = aggregate_function_clone->as<FunctionNode &>();
+
         aggregate_function_clone_typed.getArguments().getNodes() = { arithmetic_function_clone_argument };
-        resolveAggregateFunctionNode(aggregate_function_clone_typed, arithmetic_function_clone_argument, result_aggregate_function_name);
+        resolveAggregateFunctionNodeByName(aggregate_function_clone_typed, result_aggregate_function_name);
 
         arithmetic_function_clone_arguments_nodes[arithmetic_function_argument_index] = std::move(aggregate_function_clone);
-        resolveOrdinaryFunctionNode(arithmetic_function_clone_typed, arithmetic_function_clone_typed.getFunctionName());
+        resolveOrdinaryFunctionNodeByName(arithmetic_function_clone_typed, arithmetic_function_clone_typed.getFunctionName(), getContext());
 
         return arithmetic_function_clone;
     }
-
-    void resolveOrdinaryFunctionNode(FunctionNode & function_node, const String & function_name) const
-    {
-        auto function = FunctionFactory::instance().get(function_name, getContext());
-        function_node.resolveAsFunction(function->build(function_node.getArgumentColumns()));
-    }
-
-    static void resolveAggregateFunctionNode(FunctionNode & function_node, const QueryTreeNodePtr & argument, const String & aggregate_function_name)
-    {
-        auto function_aggregate_function = function_node.getAggregateFunction();
-
-        AggregateFunctionProperties properties;
-        auto action = NullsAction::EMPTY;
-        auto aggregate_function = AggregateFunctionFactory::instance().get(
-            aggregate_function_name, action, {argument->getResultType()}, function_aggregate_function->getParameters(), properties);
-
-        function_node.resolveAsAggregateFunction(std::move(aggregate_function));
-    }
 };
 
 }

src/Analyzer/Passes/ComparisonTupleEliminationPass.cpp

@@ -11,6 +11,7 @@
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Analyzer/ConstantNode.h>
 #include <Analyzer/FunctionNode.h>
+#include <Analyzer/Utils.h>
 
 namespace DB
 {
@@ -18,19 +19,18 @@ namespace DB
 namespace
 {
 
-class ComparisonTupleEliminationPassVisitor : public InDepthQueryTreeVisitor<ComparisonTupleEliminationPassVisitor>
+class ComparisonTupleEliminationPassVisitor : public InDepthQueryTreeVisitorWithContext<ComparisonTupleEliminationPassVisitor>
 {
 public:
-    explicit ComparisonTupleEliminationPassVisitor(ContextPtr context_)
+    using Base = InDepthQueryTreeVisitorWithContext<ComparisonTupleEliminationPassVisitor>;
-        : context(std::move(context_))
+    using Base::Base;
-    {}
 
     static bool needChildVisit(QueryTreeNodePtr &, QueryTreeNodePtr & child)
     {
         return child->getNodeType() != QueryTreeNodeType::TABLE_FUNCTION;
     }
 
-    void visitImpl(QueryTreeNodePtr & node) const
+    void enterImpl(QueryTreeNodePtr & node) const
     {
         auto * function_node = node->as<FunctionNode>();
         if (!function_node)
@@ -171,13 +171,13 @@ private:
     {
         auto result_function = std::make_shared<FunctionNode>("and");
         result_function->getArguments().getNodes() = std::move(tuple_arguments_equals_functions);
-        resolveOrdinaryFunctionNode(*result_function, result_function->getFunctionName());
+        resolveOrdinaryFunctionNodeByName(*result_function, result_function->getFunctionName(), getContext());
 
         if (comparison_function_name == "notEquals")
         {
             auto not_function = std::make_shared<FunctionNode>("not");
             not_function->getArguments().getNodes().push_back(std::move(result_function));
-            resolveOrdinaryFunctionNode(*not_function, not_function->getFunctionName());
+            resolveOrdinaryFunctionNodeByName(*not_function, not_function->getFunctionName(), getContext());
             result_function = std::move(not_function);
         }
 
@@ -197,18 +197,10 @@ private:
         comparison_function->getArguments().getNodes().push_back(std::move(lhs_argument));
         comparison_function->getArguments().getNodes().push_back(std::move(rhs_argument));
 
-        resolveOrdinaryFunctionNode(*comparison_function, comparison_function->getFunctionName());
+        resolveOrdinaryFunctionNodeByName(*comparison_function, comparison_function->getFunctionName(), getContext());
 
         return comparison_function;
     }
-
-    void resolveOrdinaryFunctionNode(FunctionNode & function_node, const String & function_name) const
-    {
-        auto function = FunctionFactory::instance().get(function_name, context);
-        function_node.resolveAsFunction(function->build(function_node.getArgumentColumns()));
-    }
-
-    ContextPtr context;
 };
 
 }

src/Analyzer/Passes/CountDistinctPass.cpp

@@ -9,6 +9,7 @@
 #include <Analyzer/ColumnNode.h>
 #include <Analyzer/FunctionNode.h>
 #include <Analyzer/QueryNode.h>
+#include <Analyzer/Utils.h>
 
 namespace DB
 {
@@ -77,11 +78,9 @@ public:
 
         /// Replace `countDistinct` of initial query into `count`
         auto result_type = function_node->getResultType();
-        AggregateFunctionProperties properties;
+
-        auto action = NullsAction::EMPTY;
-        auto aggregate_function = AggregateFunctionFactory::instance().get("count", action, {}, {}, properties);
-        function_node->resolveAsAggregateFunction(std::move(aggregate_function));
         function_node->getArguments().getNodes().clear();
+        resolveAggregateFunctionNodeByName(*function_node, "count");
     }
 };
 

src/Analyzer/Passes/FunctionToSubcolumnsPass.cpp

@@ -4,6 +4,7 @@
 #include <DataTypes/DataTypeTuple.h>
 #include <DataTypes/DataTypeArray.h>
 #include <DataTypes/DataTypeMap.h>
+#include <DataTypes/DataTypeVariant.h>
 
 #include <Storages/IStorage.h>
 
@@ -16,6 +17,9 @@
 #include <Analyzer/ColumnNode.h>
 #include <Analyzer/FunctionNode.h>
 #include <Analyzer/TableNode.h>
+#include <Analyzer/TableFunctionNode.h>
+#include <Analyzer/Utils.h>
+#include <Analyzer/JoinNode.h>
 
 namespace DB
 {
@@ -23,202 +27,410 @@ namespace DB
 namespace
 {
 
-class FunctionToSubcolumnsVisitor : public InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitor>
+struct ColumnContext
+{
+    NameAndTypePair column;
+    QueryTreeNodePtr column_source;
+    ContextPtr context;
+};
+
+using NodeToSubcolumnTransformer = std::function<void(QueryTreeNodePtr &, FunctionNode &, ColumnContext &)>;
+
+void optimizeFunctionLength(QueryTreeNodePtr & node, FunctionNode &, ColumnContext & ctx)
+{
+    /// Replace `length(argument)` with `argument.size0`
+    /// `argument` may be Array or Map.
+
+    NameAndTypePair column{ctx.column.name + ".size0", std::make_shared<DataTypeUInt64>()};
+    node = std::make_shared<ColumnNode>(column, ctx.column_source);
+}
+
+template <bool positive>
+void optimizeFunctionEmpty(QueryTreeNodePtr &, FunctionNode & function_node, ColumnContext & ctx)
+{
+    /// Replace `empty(argument)` with `equals(argument.size0, 0)` if positive
+    /// Replace `notEmpty(argument)` with `notEquals(argument.size0, 0)` if not positive
+    /// `argument` may be Array or Map.
+
+    NameAndTypePair column{ctx.column.name + ".size0", std::make_shared<DataTypeUInt64>()};
+    auto & function_arguments_nodes = function_node.getArguments().getNodes();
+
+    function_arguments_nodes.clear();
+    function_arguments_nodes.push_back(std::make_shared<ColumnNode>(column, ctx.column_source));
+    function_arguments_nodes.push_back(std::make_shared<ConstantNode>(static_cast<UInt64>(0)));
+
+    const auto * function_name = positive ? "equals" : "notEquals";
+    resolveOrdinaryFunctionNodeByName(function_node, function_name, ctx.context);
+}
+
+String getSubcolumnNameForElement(const Field & value, const DataTypeTuple & data_type_tuple)
+{
+    if (value.getType() == Field::Types::String)
+        return value.get<const String &>();
+
+    if (value.getType() == Field::Types::UInt64)
+        return data_type_tuple.getNameByPosition(value.get<UInt64>());
+
+    return "";
+}
+
+String getSubcolumnNameForElement(const Field & value, const DataTypeVariant &)
+{
+    if (value.getType() == Field::Types::String)
+        return value.get<const String &>();
+
+    return "";
+}
+
+template <typename DataType>
+void optimizeTupleOrVariantElement(QueryTreeNodePtr & node, FunctionNode & function_node, ColumnContext & ctx)
+{
+    /// Replace `tupleElement(tuple_argument, string_literal)`, `tupleElement(tuple_argument, integer_literal)` with `tuple_argument.column_name`.
+    /// Replace `variantElement(variant_argument, string_literal)` with `variant_argument.column_name`.
+
+    auto & function_arguments_nodes = function_node.getArguments().getNodes();
+    if (function_arguments_nodes.size() != 2)
+        return;
+
+    const auto * second_argument_constant_node = function_arguments_nodes[1]->as<ConstantNode>();
+    if (!second_argument_constant_node)
+        return;
+
+    const auto & data_type_concrete = assert_cast<const DataType &>(*ctx.column.type);
+    auto subcolumn_name = getSubcolumnNameForElement(second_argument_constant_node->getValue(), data_type_concrete);
+
+    if (subcolumn_name.empty())
+        return;
+
+    NameAndTypePair column{ctx.column.name + "." + subcolumn_name, function_node.getResultType()};
+    node = std::make_shared<ColumnNode>(column, ctx.column_source);
+}
+
+std::map<std::pair<TypeIndex, String>, NodeToSubcolumnTransformer> node_transformers =
+{
+    {
+        {TypeIndex::Array, "length"}, optimizeFunctionLength,
+    },
+    {
+        {TypeIndex::Array, "empty"}, optimizeFunctionEmpty<true>,
+    },
+    {
+        {TypeIndex::Array, "notEmpty"}, optimizeFunctionEmpty<false>,
+    },
+    {
+        {TypeIndex::Map, "length"}, optimizeFunctionLength,
+    },
+    {
+        {TypeIndex::Map, "empty"}, optimizeFunctionEmpty<true>,
+    },
+    {
+        {TypeIndex::Map, "notEmpty"}, optimizeFunctionEmpty<false>,
+    },
+    {
+        {TypeIndex::Map, "mapKeys"},
+        [](QueryTreeNodePtr & node, FunctionNode & function_node, ColumnContext & ctx)
+        {
+            /// Replace `mapKeys(map_argument)` with `map_argument.keys`
+            NameAndTypePair column{ctx.column.name + ".keys", function_node.getResultType()};
+            node = std::make_shared<ColumnNode>(column, ctx.column_source);
+        },
+    },
+    {
+        {TypeIndex::Map, "mapValues"},
+        [](QueryTreeNodePtr & node, FunctionNode & function_node, ColumnContext & ctx)
+        {
+            /// Replace `mapValues(map_argument)` with `map_argument.values`
+            NameAndTypePair column{ctx.column.name + ".values", function_node.getResultType()};
+            node = std::make_shared<ColumnNode>(column, ctx.column_source);
+        },
+    },
+    {
+        {TypeIndex::Map, "mapContains"},
+        [](QueryTreeNodePtr &, FunctionNode & function_node, ColumnContext & ctx)
+        {
+            /// Replace `mapContains(map_argument, argument)` with `has(map_argument.keys, argument)`
+            const auto & data_type_map = assert_cast<const DataTypeMap &>(*ctx.column.type);
+
+            NameAndTypePair column{ctx.column.name + ".keys", std::make_shared<DataTypeArray>(data_type_map.getKeyType())};
+            auto & function_arguments_nodes = function_node.getArguments().getNodes();
+
+            auto has_function_argument = std::make_shared<ColumnNode>(column, ctx.column_source);
+            function_arguments_nodes[0] = std::move(has_function_argument);
+
+            resolveOrdinaryFunctionNodeByName(function_node, "has", ctx.context);
+        },
+    },
+    {
+        {TypeIndex::Nullable, "count"},
+        [](QueryTreeNodePtr &, FunctionNode & function_node, ColumnContext & ctx)
+        {
+            /// Replace `count(nullable_argument)` with `sum(not(nullable_argument.null))`
+            NameAndTypePair column{ctx.column.name + ".null", std::make_shared<DataTypeUInt8>()};
+            auto & function_arguments_nodes = function_node.getArguments().getNodes();
+
+            auto new_column_node = std::make_shared<ColumnNode>(column, ctx.column_source);
+            auto function_node_not = std::make_shared<FunctionNode>("not");
+
+            function_node_not->getArguments().getNodes().push_back(std::move(new_column_node));
+            resolveOrdinaryFunctionNodeByName(*function_node_not, "not", ctx.context);
+
+            function_arguments_nodes = {std::move(function_node_not)};
+            resolveAggregateFunctionNodeByName(function_node, "sum");
+        },
+    },
+    {
+        {TypeIndex::Nullable, "isNull"},
+        [](QueryTreeNodePtr & node, FunctionNode &, ColumnContext & ctx)
+        {
+            /// Replace `isNull(nullable_argument)` with `nullable_argument.null`
+            NameAndTypePair column{ctx.column.name + ".null", std::make_shared<DataTypeUInt8>()};
+            node = std::make_shared<ColumnNode>(column, ctx.column_source);
+        },
+    },
+    {
+        {TypeIndex::Nullable, "isNotNull"},
+        [](QueryTreeNodePtr &, FunctionNode & function_node, ColumnContext & ctx)
+        {
+            /// Replace `isNotNull(nullable_argument)` with `not(nullable_argument.null)`
+            NameAndTypePair column{ctx.column.name + ".null", std::make_shared<DataTypeUInt8>()};
+            auto & function_arguments_nodes = function_node.getArguments().getNodes();
+
+            function_arguments_nodes = {std::make_shared<ColumnNode>(column, ctx.column_source)};
+            resolveOrdinaryFunctionNodeByName(function_node, "not", ctx.context);
+        },
+    },
+    {
+        {TypeIndex::Tuple, "tupleElement"}, optimizeTupleOrVariantElement<DataTypeTuple>,
+    },
+    {
+        {TypeIndex::Variant, "variantElement"}, optimizeTupleOrVariantElement<DataTypeVariant>,
+    },
+};
+
+std::tuple<FunctionNode *, ColumnNode *, TableNode *> getTypedNodesForOptimization(const QueryTreeNodePtr & node)
+{
+    auto * function_node = node->as<FunctionNode>();
+    if (!function_node)
+        return {};
+
+    auto & function_arguments_nodes = function_node->getArguments().getNodes();
+    if (function_arguments_nodes.empty() || function_arguments_nodes.size() > 2)
+        return {};
+
+    auto * first_argument_column_node = function_arguments_nodes.front()->as<ColumnNode>();
+    if (!first_argument_column_node || first_argument_column_node->getColumnName() == "__grouping_set")
+        return {};
+
+    auto column_source = first_argument_column_node->getColumnSource();
+    auto * table_node = column_source->as<TableNode>();
+    if (!table_node)
+        return {};
+
+    const auto & storage = table_node->getStorage();
+    const auto & storage_snapshot = table_node->getStorageSnapshot();
+    auto column = first_argument_column_node->getColumn();
+
+    if (!storage->supportsOptimizationToSubcolumns() || storage->isVirtualColumn(column.name, storage_snapshot->metadata))
+        return {};
+
+    auto column_in_table = storage_snapshot->tryGetColumn(GetColumnsOptions::All, column.name);
+    if (!column_in_table || !column_in_table->type->equals(*column.type))
+        return {};
+
+    return std::make_tuple(function_node, first_argument_column_node, table_node);
+}
+
+/// First pass collects info about identifiers to determine which identifiers are allowed to optimize.
+class FunctionToSubcolumnsVisitorFirstPass : public InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitorFirstPass>
 {
 public:
-    using Base = InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitor>;
+    using Base = InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitorFirstPass>;
     using Base::Base;
 
+    void enterImpl(const QueryTreeNodePtr & node)
+    {
+        if (!getSettings().optimize_functions_to_subcolumns)
+            return;
+
+        if (auto * table_node = node->as<TableNode>())
+        {
+            enterImpl(*table_node);
+            return;
+        }
+
+        if (auto * column_node = node->as<ColumnNode>())
+        {
+            enterImpl(*column_node);
+            return;
+        }
+
+        auto [function_node, first_argument_node, table_node] = getTypedNodesForOptimization(node);
+        if (function_node && first_argument_node && table_node)
+        {
+            enterImpl(*function_node, *first_argument_node, *table_node);
+            return;
+        }
+
+        if (const auto * join_node = node->as<JoinNode>())
+        {
+            can_wrap_result_columns_with_nullable |= getContext()->getSettingsRef().join_use_nulls;
+            return;
+        }
+
+        if (const auto * query_node = node->as<QueryNode>())
+        {
+            if (query_node->isGroupByWithCube() || query_node->isGroupByWithRollup() || query_node->isGroupByWithGroupingSets())
+                can_wrap_result_columns_with_nullable |= getContext()->getSettingsRef().group_by_use_nulls;
+            return;
+        }
+    }
+
+    std::unordered_set<Identifier> getIdentifiersToOptimize() const
+    {
+        if (can_wrap_result_columns_with_nullable)
+        {
+            /// Do not optimize if we have JOIN with setting join_use_null.
+            /// Do not optimize if we have GROUP BY WITH ROLLUP/CUBE/GROUPING SETS with setting group_by_use_nulls.
+            /// It may change the behaviour if subcolumn can be converted
+            /// to Nullable while the original column cannot (e.g. for Array type).
+            return {};
+        }
+
+        /// Do not optimize if full column is requested in other context.
+        /// It doesn't make sense because it doesn't reduce amount of read data
+        /// and optimized functions are not computation heavy. But introducing
+        /// new identifier complicates query analysis and may break it.
+        ///
+        /// E.g. query:
+        ///     SELECT n FROM table GROUP BY n HAVING isNotNull(n)
+        /// may be optimized to incorrect query:
+        ///     SELECT n FROM table GROUP BY n HAVING not(n.null)
+        /// Will produce: `n.null` is not under aggregate function and not in GROUP BY keys)
+        ///
+        /// Do not optimize index columns (primary, min-max, secondary),
+        /// because otherwise analysis of indexes may be broken.
+        /// TODO: handle subcolumns in index analysis.
+
+        std::unordered_set<Identifier> identifiers_to_optimize;
+        for (const auto & [identifier, count] : optimized_identifiers_count)
+        {
+            if (all_key_columns.contains(identifier))
+                continue;
+
+            auto it = identifiers_count.find(identifier);
+            if (it != identifiers_count.end() && it->second == count)
+                identifiers_to_optimize.insert(identifier);
+        }
+
+        return identifiers_to_optimize;
+    }
+
+private:
+    std::unordered_set<Identifier> all_key_columns;
+    std::unordered_map<Identifier, UInt64> identifiers_count;
+    std::unordered_map<Identifier, UInt64> optimized_identifiers_count;
+
+    NameSet processed_tables;
+    bool can_wrap_result_columns_with_nullable = false;
+
+    void enterImpl(const TableNode & table_node)
+    {
+        auto table_name = table_node.getStorage()->getStorageID().getFullTableName();
+        if (processed_tables.emplace(table_name).second)
+            return;
+
+        auto add_key_columns = [&](const auto & key_columns)
+        {
+            for (const auto & column_name : key_columns)
+            {
+                Identifier identifier({table_name, column_name});
+                all_key_columns.insert(identifier);
+            }
+        };
+
+        const auto & metadata_snapshot = table_node.getStorageSnapshot()->metadata;
+        const auto & primary_key_columns = metadata_snapshot->getColumnsRequiredForPrimaryKey();
+        const auto & partition_key_columns = metadata_snapshot->getColumnsRequiredForPartitionKey();
+
+        add_key_columns(primary_key_columns);
+        add_key_columns(partition_key_columns);
+
+        for (const auto & index : metadata_snapshot->getSecondaryIndices())
+        {
+            const auto & index_columns = index.expression->getRequiredColumns();
+            add_key_columns(index_columns);
+        }
+    }
+
+    void enterImpl(const ColumnNode & column_node)
+    {
+        if (column_node.getColumnName() == "__grouping_set")
+            return;
+
+        auto column_source = column_node.getColumnSource();
+        auto * table_node = column_source->as<TableNode>();
+        if (!table_node)
+            return;
+
+        auto table_name = table_node->getStorage()->getStorageID().getFullTableName();
+        Identifier qualified_name({table_name, column_node.getColumnName()});
+
+        ++identifiers_count[qualified_name];
+    }
+
+    void enterImpl(const FunctionNode & function_node, const ColumnNode & first_argument_column_node, const TableNode & table_node)
+    {
+        /// For queries with FINAL converting function to subcolumn may alter
+        /// special merging algorithms and produce wrong result of query.
+        if (table_node.hasTableExpressionModifiers() && table_node.getTableExpressionModifiers()->hasFinal())
+            return;
+
+        const auto & column = first_argument_column_node.getColumn();
+        auto table_name = table_node.getStorage()->getStorageID().getFullTableName();
+        Identifier qualified_name({table_name, column.name});
+
+        if (node_transformers.contains({column.type->getTypeId(), function_node.getFunctionName()}))
+            ++optimized_identifiers_count[qualified_name];
+    }
+};
+
+/// Second pass optimizes functions to subcolumns for allowed identifiers.
+class FunctionToSubcolumnsVisitorSecondPass : public InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitorSecondPass>
+{
+private:
+    std::unordered_set<Identifier> identifiers_to_optimize;
+
+public:
+    using Base = InDepthQueryTreeVisitorWithContext<FunctionToSubcolumnsVisitorSecondPass>;
+    using Base::Base;
+
+    FunctionToSubcolumnsVisitorSecondPass(ContextPtr context_, std::unordered_set<Identifier> identifiers_to_optimize_)
+        : Base(std::move(context_)), identifiers_to_optimize(std::move(identifiers_to_optimize_))
+    {
+    }
+
     void enterImpl(QueryTreeNodePtr & node) const
     {
         if (!getSettings().optimize_functions_to_subcolumns)
             return;
 
-        auto * function_node = node->as<FunctionNode>();
+        auto [function_node, first_argument_column_node, table_node] = getTypedNodesForOptimization(node);
-        if (!function_node)
+        if (!function_node || !first_argument_column_node || !table_node)
-            return;
-
-        auto & function_arguments_nodes = function_node->getArguments().getNodes();
-        size_t function_arguments_nodes_size = function_arguments_nodes.size();
-
-        if (function_arguments_nodes.empty() || function_arguments_nodes_size > 2)
-            return;
-
-        auto * first_argument_column_node = function_arguments_nodes.front()->as<ColumnNode>();
-
-        if (!first_argument_column_node)
-            return;
-
-        if (first_argument_column_node->getColumnName() == "__grouping_set")
-            return;
-
-        auto column_source = first_argument_column_node->getColumnSource();
-        auto * table_node = column_source->as<TableNode>();
-
-        if (!table_node)
-            return;
-
-        const auto & storage = table_node->getStorage();
-        if (!storage->supportsSubcolumns())
             return;
 
         auto column = first_argument_column_node->getColumn();
-        WhichDataType column_type(column.type);
+        auto table_name = table_node->getStorage()->getStorageID().getFullTableName();
 
-        const auto & function_name = function_node->getFunctionName();
+        Identifier qualified_name({table_name, column.name});
+        if (!identifiers_to_optimize.contains(qualified_name))
+            return;
 
-        if (function_arguments_nodes_size == 1)
+        auto transformer_it = node_transformers.find({column.type->getTypeId(), function_node->getFunctionName()});
+        if (transformer_it != node_transformers.end())
         {
-            if (column_type.isArray())
+            ColumnContext ctx{std::move(column), first_argument_column_node->getColumnSource(), getContext()};
-            {
+            transformer_it->second(node, *function_node, ctx);
-                if (function_name == "length")
-                {
-                    /// Replace `length(array_argument)` with `array_argument.size0`
-                    column.name += ".size0";
-                    column.type = std::make_shared<DataTypeUInt64>();
-
-                    node = std::make_shared<ColumnNode>(column, column_source);
-                }
-                else if (function_name == "empty")
-                {
-                    /// Replace `empty(array_argument)` with `equals(array_argument.size0, 0)`
-                    column.name += ".size0";
-                    column.type = std::make_shared<DataTypeUInt64>();
-
-                    function_arguments_nodes.clear();
-                    function_arguments_nodes.push_back(std::make_shared<ColumnNode>(column, column_source));
-                    function_arguments_nodes.push_back(std::make_shared<ConstantNode>(static_cast<UInt64>(0)));
-
-                    resolveOrdinaryFunctionNode(*function_node, "equals");
-                }
-                else if (function_name == "notEmpty")
-                {
-                    /// Replace `notEmpty(array_argument)` with `notEquals(array_argument.size0, 0)`
-                    column.name += ".size0";
-                    column.type = std::make_shared<DataTypeUInt64>();
-
-                    function_arguments_nodes.clear();
-                    function_arguments_nodes.push_back(std::make_shared<ColumnNode>(column, column_source));
-                    function_arguments_nodes.push_back(std::make_shared<ConstantNode>(static_cast<UInt64>(0)));
-
-                    resolveOrdinaryFunctionNode(*function_node, "notEquals");
-                }
-            }
-            else if (column_type.isNullable())
-            {
-                if (function_name == "isNull")
-                {
-                    /// Replace `isNull(nullable_argument)` with `nullable_argument.null`
-                    column.name += ".null";
-                    column.type = std::make_shared<DataTypeUInt8>();
-
-                    node = std::make_shared<ColumnNode>(column, column_source);
-                }
-                else if (function_name == "isNotNull")
-                {
-                    /// Replace `isNotNull(nullable_argument)` with `not(nullable_argument.null)`
-                    column.name += ".null";
-                    column.type = std::make_shared<DataTypeUInt8>();
-
-                    function_arguments_nodes = {std::make_shared<ColumnNode>(column, column_source)};
-
-                    resolveOrdinaryFunctionNode(*function_node, "not");
-                }
-            }
-            else if (column_type.isMap())
-            {
-                if (function_name == "mapKeys")
-                {
-                    /// Replace `mapKeys(map_argument)` with `map_argument.keys`
-                    column.name += ".keys";
-                    column.type = function_node->getResultType();
-
-                    node = std::make_shared<ColumnNode>(column, column_source);
-                }
-                else if (function_name == "mapValues")
-                {
-                    /// Replace `mapValues(map_argument)` with `map_argument.values`
-                    column.name += ".values";
-                    column.type = function_node->getResultType();
-
-                    node = std::make_shared<ColumnNode>(column, column_source);
-                }
-            }
         }
-        else
-        {
-            const auto * second_argument_constant_node = function_arguments_nodes[1]->as<ConstantNode>();
-
-            if (function_name == "tupleElement" && column_type.isTuple() && second_argument_constant_node)
-            {
-                /** Replace `tupleElement(tuple_argument, string_literal)`, `tupleElement(tuple_argument, integer_literal)`
-                  * with `tuple_argument.column_name`.
-                  */
-                const auto & tuple_element_constant_value = second_argument_constant_node->getValue();
-                const auto & tuple_element_constant_value_type = tuple_element_constant_value.getType();
-
-                const auto & data_type_tuple = assert_cast<const DataTypeTuple &>(*column.type);
-
-                String subcolumn_name;
-
-                if (tuple_element_constant_value_type == Field::Types::String)
-                {
-                    subcolumn_name = tuple_element_constant_value.get<const String &>();
-                }
-                else if (tuple_element_constant_value_type == Field::Types::UInt64)
-                {
-                    auto tuple_column_index = tuple_element_constant_value.get<UInt64>();
-                    subcolumn_name = data_type_tuple.getNameByPosition(tuple_column_index);
-                }
-                else
-                {
-                    return;
-                }
-
-                column.name += '.';
-                column.name += subcolumn_name;
-                column.type = function_node->getResultType();
-
-                node = std::make_shared<ColumnNode>(column, column_source);
-            }
-            else if (function_name == "variantElement" && isVariant(column_type) && second_argument_constant_node)
-            {
-                /// Replace `variantElement(variant_argument, type_name)` with `variant_argument.type_name`.
-                const auto & variant_element_constant_value = second_argument_constant_node->getValue();
-                String subcolumn_name;
-
-                if (variant_element_constant_value.getType() != Field::Types::String)
-                    return;
-
-                subcolumn_name = variant_element_constant_value.get<const String &>();
-
-                column.name += '.';
-                column.name += subcolumn_name;
-                column.type = function_node->getResultType();
-
-                node = std::make_shared<ColumnNode>(column, column_source);
-            }
-            else if (function_name == "mapContains" && column_type.isMap())
-            {
-                const auto & data_type_map = assert_cast<const DataTypeMap &>(*column.type);
-
-                /// Replace `mapContains(map_argument, argument)` with `has(map_argument.keys, argument)`
-                column.name += ".keys";
-                column.type = std::make_shared<DataTypeArray>(data_type_map.getKeyType());
-
-                auto has_function_argument = std::make_shared<ColumnNode>(column, column_source);
-                function_arguments_nodes[0] = std::move(has_function_argument);
-
-                resolveOrdinaryFunctionNode(*function_node, "has");
-            }
-        }
-    }
-
-private:
-    void resolveOrdinaryFunctionNode(FunctionNode & function_node, const String & function_name) const
-    {
-        auto function = FunctionFactory::instance().get(function_name, getContext());
-        function_node.resolveAsFunction(function->build(function_node.getArgumentColumns()));
     }
 };
 
@@ -226,8 +438,15 @@ private:
 
 void FunctionToSubcolumnsPass::run(QueryTreeNodePtr & query_tree_node, ContextPtr context)
 {
-    FunctionToSubcolumnsVisitor visitor(context);
+    FunctionToSubcolumnsVisitorFirstPass first_visitor(context);
-    visitor.visit(query_tree_node);
+    first_visitor.visit(query_tree_node);
+    auto identifiers_to_optimize = first_visitor.getIdentifiersToOptimize();
+
+    if (identifiers_to_optimize.empty())
+        return;
+
+    FunctionToSubcolumnsVisitorSecondPass second_visitor(std::move(context), std::move(identifiers_to_optimize));
+    second_visitor.visit(query_tree_node);
 }
 
 }

src/Analyzer/Passes/NormalizeCountVariantsPass.cpp

@@ -6,6 +6,7 @@
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Analyzer/ConstantNode.h>
 #include <Analyzer/FunctionNode.h>
+#include <Analyzer/Utils.h>
 #include <Interpreters/Context.h>
 #include <DataTypes/DataTypesNumber.h>
 
@@ -47,25 +48,17 @@ public:
 
         if (function_node->getFunctionName() == "count" && !first_argument_constant_literal.isNull())
         {
-            resolveAsCountAggregateFunction(*function_node);
             function_node->getArguments().getNodes().clear();
+            resolveAggregateFunctionNodeByName(*function_node, "count");
         }
         else if (function_node->getFunctionName() == "sum" &&
             first_argument_constant_literal.getType() == Field::Types::UInt64 &&
             first_argument_constant_literal.get<UInt64>() == 1)
         {
-            resolveAsCountAggregateFunction(*function_node);
             function_node->getArguments().getNodes().clear();
+            resolveAggregateFunctionNodeByName(*function_node, "count");
         }
     }
-private:
-    static void resolveAsCountAggregateFunction(FunctionNode & function_node)
-    {
-        AggregateFunctionProperties properties;
-        auto aggregate_function = AggregateFunctionFactory::instance().get("count", NullsAction::EMPTY, {}, {}, properties);
-
-        function_node.resolveAsAggregateFunction(std::move(aggregate_function));
-    }
 };
 
 }

src/Analyzer/Passes/OptimizeDateOrDateTimeConverterWithPreimagePass.cpp

@@ -5,6 +5,7 @@
 #include <Analyzer/ColumnNode.h>
 #include <Analyzer/ConstantNode.h>
 #include <Analyzer/FunctionNode.h>
+#include <Analyzer/Utils.h>
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Common/DateLUT.h>
 #include <Common/DateLUTImpl.h>

src/Analyzer/Passes/RewriteAggregateFunctionWithIfPass.cpp

@@ -74,8 +74,7 @@ public:
 
                 new_arguments[1] = std::move(if_arguments_nodes[0]);
                 function_arguments_nodes = std::move(new_arguments);
-                resolveAsAggregateFunctionWithIf(
+                resolveAggregateFunctionNodeByName(*function_node, function_node->getFunctionName() + "If");
-                    *function_node, {function_arguments_nodes[0]->getResultType(), function_arguments_nodes[1]->getResultType()});
             }
         }
         else if (first_const_node)
@@ -104,27 +103,10 @@ public:
                 new_arguments[1] = std::move(not_function);
 
                 function_arguments_nodes = std::move(new_arguments);
-                resolveAsAggregateFunctionWithIf(
+                resolveAggregateFunctionNodeByName(*function_node, function_node->getFunctionName() + "If");
-                    *function_node, {function_arguments_nodes[0]->getResultType(), function_arguments_nodes[1]->getResultType()});
             }
         }
     }
-
-private:
-    static void resolveAsAggregateFunctionWithIf(FunctionNode & function_node, const DataTypes & argument_types)
-    {
-        auto result_type = function_node.getResultType();
-
-        AggregateFunctionProperties properties;
-        auto aggregate_function = AggregateFunctionFactory::instance().get(
-            function_node.getFunctionName() + "If",
-            function_node.getNullsAction(),
-            argument_types,
-            function_node.getAggregateFunction()->getParameters(),
-            properties);
-
-        function_node.resolveAsAggregateFunction(std::move(aggregate_function));
-    }
 };
 
 }

src/Analyzer/Passes/RewriteSumFunctionWithSumAndCountPass.cpp

@@ -73,23 +73,24 @@ public:
 
         const auto lhs = std::make_shared<FunctionNode>("sum");
         lhs->getArguments().getNodes().push_back(func_plus_minus_nodes[column_id]);
-        resolveAsAggregateFunctionNode(*lhs, column_type);
+        resolveAggregateFunctionNodeByName(*lhs, lhs->getFunctionName());
 
         const auto rhs_count = std::make_shared<FunctionNode>("count");
         rhs_count->getArguments().getNodes().push_back(func_plus_minus_nodes[column_id]);
-        resolveAsAggregateFunctionNode(*rhs_count, column_type);
+        resolveAggregateFunctionNodeByName(*rhs_count, rhs_count->getFunctionName());
 
         const auto rhs = std::make_shared<FunctionNode>("multiply");
         rhs->getArguments().getNodes().push_back(func_plus_minus_nodes[literal_id]);
         rhs->getArguments().getNodes().push_back(rhs_count);
-        resolveOrdinaryFunctionNode(*rhs, rhs->getFunctionName());
+        resolveOrdinaryFunctionNodeByName(*rhs, rhs->getFunctionName(), getContext());
 
         auto new_node = std::make_shared<FunctionNode>(Poco::toLower(func_plus_minus_node->getFunctionName()));
         if (column_id == 0)
             new_node->getArguments().getNodes() = {lhs, rhs};
         else if (column_id == 1)
             new_node->getArguments().getNodes() = {rhs, lhs};
-        resolveOrdinaryFunctionNode(*new_node, new_node->getFunctionName());
+
+        resolveOrdinaryFunctionNodeByName(*new_node, new_node->getFunctionName(), getContext());
 
         if (!new_node)
             return;
@@ -100,28 +101,7 @@ public:
             res = createCastFunction(res, function_node->getResultType(), getContext());
 
         node = std::move(res);
-
     }
-
-private:
-    void resolveOrdinaryFunctionNode(FunctionNode & function_node, const String & function_name) const
-    {
-        const auto function = FunctionFactory::instance().get(function_name, getContext());
-        function_node.resolveAsFunction(function->build(function_node.getArgumentColumns()));
-    }
-
-    static void resolveAsAggregateFunctionNode(FunctionNode & function_node, const DataTypePtr & argument_type)
-    {
-        AggregateFunctionProperties properties;
-        const auto aggregate_function = AggregateFunctionFactory::instance().get(function_node.getFunctionName(),
-            NullsAction::EMPTY,
-            {argument_type},
-            {},
-            properties);
-
-        function_node.resolveAsAggregateFunction(aggregate_function);
-    }
-
 };
 
 }

src/Analyzer/Passes/SumIfToCountIfPass.cpp

@@ -5,6 +5,7 @@
 
 #include <AggregateFunctions/AggregateFunctionFactory.h>
 #include <AggregateFunctions/IAggregateFunction.h>
+#include <Analyzer/Utils.h>
 
 #include <Functions/FunctionFactory.h>
 
@@ -65,7 +66,8 @@ public:
             auto multiplier_node = function_node_arguments_nodes[0];
             function_node_arguments_nodes[0] = std::move(function_node_arguments_nodes[1]);
             function_node_arguments_nodes.resize(1);
-            resolveAsCountIfAggregateFunction(*function_node, function_node_arguments_nodes[0]->getResultType());
+
+            resolveAggregateFunctionNodeByName(*function_node, "countIf");
 
             if (constant_value_literal.get<UInt64>() != 1)
             {
@@ -115,7 +117,7 @@ public:
             function_node_arguments_nodes[0] = nested_if_function_arguments_nodes[0];
             function_node_arguments_nodes.resize(1);
 
-            resolveAsCountIfAggregateFunction(*function_node, function_node_arguments_nodes[0]->getResultType());
+            resolveAggregateFunctionNodeByName(*function_node, "countIf");
 
             if (if_true_condition_value != 1)
             {
@@ -144,7 +146,7 @@ public:
             function_node_arguments_nodes[0] = std::move(not_function);
             function_node_arguments_nodes.resize(1);
 
-            resolveAsCountIfAggregateFunction(*function_node, function_node_arguments_nodes[0]->getResultType());
+            resolveAggregateFunctionNodeByName(*function_node, "countIf");
 
             if (if_false_condition_value != 1)
             {
@@ -156,15 +158,6 @@ public:
     }
 
 private:
-    static void resolveAsCountIfAggregateFunction(FunctionNode & function_node, const DataTypePtr & argument_type)
-    {
-        AggregateFunctionProperties properties;
-        auto aggregate_function = AggregateFunctionFactory::instance().get(
-            "countIf", NullsAction::EMPTY, {argument_type}, function_node.getAggregateFunction()->getParameters(), properties);
-
-        function_node.resolveAsAggregateFunction(std::move(aggregate_function));
-    }
-
     QueryTreeNodePtr getMultiplyFunction(QueryTreeNodePtr left, QueryTreeNodePtr right)
     {
         auto multiply_function_node = std::make_shared<FunctionNode>("multiply");

src/Analyzer/Passes/UniqInjectiveFunctionsEliminationPass.cpp

@@ -7,6 +7,7 @@
 
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Analyzer/FunctionNode.h>
+#include <Analyzer/Utils.h>
 
 
 namespace DB

src/Analyzer/Passes/UniqToCountPass.cpp

@@ -7,6 +7,7 @@
 #include <Analyzer/FunctionNode.h>
 #include <Analyzer/InDepthQueryTreeVisitor.h>
 #include <Analyzer/QueryNode.h>
+#include <Analyzer/Utils.h>
 
 namespace DB
 {
@@ -184,11 +185,8 @@ public:
         /// Replace uniq of initial query to count
         if (match_subquery_with_distinct() || match_subquery_with_group_by())
         {
-            AggregateFunctionProperties properties;
-            auto aggregate_function = AggregateFunctionFactory::instance().get("count", NullsAction::EMPTY, {}, {}, properties);
-
             function_node->getArguments().getNodes().clear();
-            function_node->resolveAsAggregateFunction(std::move(aggregate_function));
+            resolveAggregateFunctionNodeByName(*function_node, "count");
         }
     }
 };

src/Analyzer/Utils.cpp

@@ -636,16 +636,16 @@ private:
     bool has_function = false;
 };
 
-inline AggregateFunctionPtr resolveAggregateFunction(FunctionNode * function_node)
+inline AggregateFunctionPtr resolveAggregateFunction(FunctionNode & function_node, const String & function_name)
 {
     Array parameters;
-    for (const auto & param : function_node->getParameters())
+    for (const auto & param : function_node.getParameters())
     {
         auto * constant = param->as<ConstantNode>();
         parameters.push_back(constant->getValue());
     }
 
-    const auto & function_node_argument_nodes = function_node->getArguments().getNodes();
+    const auto & function_node_argument_nodes = function_node.getArguments().getNodes();
 
     DataTypes argument_types;
     argument_types.reserve(function_node_argument_nodes.size());
@@ -655,7 +655,7 @@ inline AggregateFunctionPtr resolveAggregateFunction(FunctionNode * function_nod
 
     AggregateFunctionProperties properties;
     auto action = NullsAction::EMPTY;
-    return AggregateFunctionFactory::instance().get(function_node->getFunctionName(), action, argument_types, parameters, properties);
+    return AggregateFunctionFactory::instance().get(function_name, action, argument_types, parameters, properties);
 }
 
 }
@@ -736,11 +736,11 @@ void rerunFunctionResolve(FunctionNode * function_node, ContextPtr context)
     {
         if (name == "nothing" || name == "nothingUInt64" || name == "nothingNull")
             return;
-        function_node->resolveAsAggregateFunction(resolveAggregateFunction(function_node));
+        function_node->resolveAsAggregateFunction(resolveAggregateFunction(*function_node, function_node->getFunctionName()));
     }
     else if (function_node->isWindowFunction())
     {
-        function_node->resolveAsWindowFunction(resolveAggregateFunction(function_node));
+        function_node->resolveAsWindowFunction(resolveAggregateFunction(*function_node, function_node->getFunctionName()));
     }
 }
 
@@ -793,6 +793,18 @@ QueryTreeNodePtr createCastFunction(QueryTreeNodePtr node, DataTypePtr result_ty
     return function_node;
 }
 
+void resolveOrdinaryFunctionNodeByName(FunctionNode & function_node, const String & function_name, const ContextPtr & context)
+{
+    auto function = FunctionFactory::instance().get(function_name, context);
+    function_node.resolveAsFunction(function->build(function_node.getArgumentColumns()));
+}
+
+void resolveAggregateFunctionNodeByName(FunctionNode & function_node, const String & function_name)
+{
+    auto aggregate_function = resolveAggregateFunction(function_node, function_name);
+    function_node.resolveAsAggregateFunction(std::move(aggregate_function));
+}
+
 /** Returns:
   * {_, false} - multiple sources
   * {nullptr, true} - no sources (for constants)

src/Analyzer/Utils.h

@@ -112,6 +112,14 @@ NameSet collectIdentifiersFullNames(const QueryTreeNodePtr & node);
 /// Wrap node into `_CAST` function
 QueryTreeNodePtr createCastFunction(QueryTreeNodePtr node, DataTypePtr result_type, ContextPtr context);
 
+/// Resolves function node as ordinary function with given name.
+/// Arguments and parameters are taken from the node.
+void resolveOrdinaryFunctionNodeByName(FunctionNode & function_node, const String & function_name, const ContextPtr & context);
+
+/// Resolves function node as aggregate function with given name.
+/// Arguments and parameters are taken from the node.
+void resolveAggregateFunctionNodeByName(FunctionNode & function_node, const String & function_name);
+
 /// Checks that node has only one source and returns it
 QueryTreeNodePtr getExpressionSource(const QueryTreeNodePtr & node);
 

src/Backups/BackupIO_AzureBlobStorage.cpp

@@ -29,48 +29,49 @@ namespace ErrorCodes
 }
 
 BackupReaderAzureBlobStorage::BackupReaderAzureBlobStorage(
-    const StorageAzureConfiguration & configuration_,
+    const AzureBlobStorage::ConnectionParams & connection_params_,
+    const String & blob_path_,
     bool allow_azure_native_copy,
     const ReadSettings & read_settings_,
     const WriteSettings & write_settings_,
     const ContextPtr & context_)
     : BackupReaderDefault(read_settings_, write_settings_, getLogger("BackupReaderAzureBlobStorage"))
-    , data_source_description{DataSourceType::ObjectStorage, ObjectStorageType::Azure, MetadataStorageType::None, configuration_.getConnectionURL().toString(), false, false}
+    , data_source_description{DataSourceType::ObjectStorage, ObjectStorageType::Azure, MetadataStorageType::None, connection_params_.getConnectionURL(), false, false}
-    , configuration(configuration_)
+    , connection_params(connection_params_)
+    , blob_path(blob_path_)
 {
-    auto client_ptr = configuration.createClient(/* is_readonly */false, /* attempt_to_create_container */true);
+    auto client_ptr = AzureBlobStorage::getContainerClient(connection_params, /*readonly=*/ false);
-    client_ptr->SetClickhouseOptions(Azure::Storage::Blobs::ClickhouseClientOptions{.IsClientForDisk=true});
+    auto settings_ptr = AzureBlobStorage::getRequestSettingsForBackup(context_->getSettingsRef(), allow_azure_native_copy);
 
-    object_storage = std::make_unique<AzureObjectStorage>("BackupReaderAzureBlobStorage",
+    object_storage = std::make_unique<AzureObjectStorage>(
-                                                          std::move(client_ptr),
+        "BackupReaderAzureBlobStorage",
-                                                          configuration.createSettings(context_),
+        std::move(client_ptr),
-                                                          configuration_.container,
+        std::move(settings_ptr),
-                                                          configuration.getConnectionURL().toString());
+        connection_params.getContainer(),
+        connection_params.getConnectionURL());
 
     client = object_storage->getAzureBlobStorageClient();
-    auto settings_copy = *object_storage->getSettings();
+    settings = object_storage->getSettings();
-    settings_copy.use_native_copy = allow_azure_native_copy;
-    settings = std::make_unique<const AzureObjectStorageSettings>(settings_copy);
 }
 
 BackupReaderAzureBlobStorage::~BackupReaderAzureBlobStorage() = default;
 
 bool BackupReaderAzureBlobStorage::fileExists(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     return object_storage->exists(StoredObject(key));
 }
 
 UInt64 BackupReaderAzureBlobStorage::getFileSize(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     ObjectMetadata object_metadata = object_storage->getObjectMetadata(key);
     return object_metadata.size_bytes;
 }
 
 std::unique_ptr<SeekableReadBuffer> BackupReaderAzureBlobStorage::readFile(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     return std::make_unique<ReadBufferFromAzureBlobStorage>(
         client, key, read_settings, settings->max_single_read_retries,
         settings->max_single_download_retries);
@@ -85,23 +86,23 @@ void BackupReaderAzureBlobStorage::copyFileToDisk(const String & path_in_backup,
         && destination_data_source_description.is_encrypted == encrypted_in_backup)
     {
         LOG_TRACE(log, "Copying {} from AzureBlobStorage to disk {}", path_in_backup, destination_disk->getName());
-        auto write_blob_function = [&](const Strings & blob_path, WriteMode mode, const std::optional<ObjectAttributes> &) -> size_t
+        auto write_blob_function = [&](const Strings & dst_blob_path, WriteMode mode, const std::optional<ObjectAttributes> &) -> size_t
         {
             /// Object storage always uses mode `Rewrite` because it simulates append using metadata and different files.
-            if (blob_path.size() != 2 || mode != WriteMode::Rewrite)
+            if (dst_blob_path.size() != 2 || mode != WriteMode::Rewrite)
                 throw Exception(ErrorCodes::LOGICAL_ERROR,
                                 "Blob writing function called with unexpected blob_path.size={} or mode={}",
-                                blob_path.size(), mode);
+                                dst_blob_path.size(), mode);
 
             copyAzureBlobStorageFile(
                 client,
                 destination_disk->getObjectStorage()->getAzureBlobStorageClient(),
-                configuration.container,
+                connection_params.getContainer(),
-                fs::path(configuration.blob_path) / path_in_backup,
+                fs::path(blob_path) / path_in_backup,
                 0,
                 file_size,
-                /* dest_container */ blob_path[1],
+                /* dest_container */ dst_blob_path[1],
-                /* dest_path */ blob_path[0],
+                /* dest_path */ dst_blob_path[0],
                 settings,
                 read_settings,
                 threadPoolCallbackRunnerUnsafe<void>(getBackupsIOThreadPool().get(), "BackupRDAzure"));
@@ -119,28 +120,33 @@ void BackupReaderAzureBlobStorage::copyFileToDisk(const String & path_in_backup,
 
 
 BackupWriterAzureBlobStorage::BackupWriterAzureBlobStorage(
-    const StorageAzureConfiguration & configuration_,
+    const AzureBlobStorage::ConnectionParams & connection_params_,
+    const String & blob_path_,
     bool allow_azure_native_copy,
     const ReadSettings & read_settings_,
     const WriteSettings & write_settings_,
     const ContextPtr & context_,
     bool attempt_to_create_container)
     : BackupWriterDefault(read_settings_, write_settings_, getLogger("BackupWriterAzureBlobStorage"))
-    , data_source_description{DataSourceType::ObjectStorage, ObjectStorageType::Azure, MetadataStorageType::None, configuration_.getConnectionURL().toString(), false, false}
+    , data_source_description{DataSourceType::ObjectStorage, ObjectStorageType::Azure, MetadataStorageType::None, connection_params_.getConnectionURL(), false, false}
-    , configuration(configuration_)
+    , connection_params(connection_params_)
+    , blob_path(blob_path_)
 {
-    auto client_ptr = configuration.createClient(/* is_readonly */false, attempt_to_create_container);
+    if (!attempt_to_create_container)
-    client_ptr->SetClickhouseOptions(Azure::Storage::Blobs::ClickhouseClientOptions{.IsClientForDisk=true});
+        connection_params.endpoint.container_already_exists = true;
+
+    auto client_ptr = AzureBlobStorage::getContainerClient(connection_params, /*readonly=*/ false);
+    auto settings_ptr = AzureBlobStorage::getRequestSettingsForBackup(context_->getSettingsRef(), allow_azure_native_copy);
+
+    object_storage = std::make_unique<AzureObjectStorage>(
+        "BackupWriterAzureBlobStorage",
+        std::move(client_ptr),
+        std::move(settings_ptr),
+        connection_params.getContainer(),
+        connection_params.getConnectionURL());
 
-    object_storage = std::make_unique<AzureObjectStorage>("BackupWriterAzureBlobStorage",
-                                                          std::move(client_ptr),
-                                                          configuration.createSettings(context_),
-                                                          configuration.container,
-                                                          configuration_.getConnectionURL().toString());
     client = object_storage->getAzureBlobStorageClient();
-    auto settings_copy = *object_storage->getSettings();
+    settings = object_storage->getSettings();
-    settings_copy.use_native_copy = allow_azure_native_copy;
-    settings = std::make_unique<const AzureObjectStorageSettings>(settings_copy);
 }
 
 void BackupWriterAzureBlobStorage::copyFileFromDisk(
@@ -159,18 +165,18 @@ void BackupWriterAzureBlobStorage::copyFileFromDisk(
     {
         /// getBlobPath() can return more than 3 elements if the file is stored as multiple objects in AzureBlobStorage container.
         /// In this case we can't use the native copy.
-        if (auto blob_path = src_disk->getBlobPath(src_path); blob_path.size() == 2)
+        if (auto src_blob_path = src_disk->getBlobPath(src_path); src_blob_path.size() == 2)
         {
             LOG_TRACE(log, "Copying file {} from disk {} to AzureBlobStorag", src_path, src_disk->getName());
             copyAzureBlobStorageFile(
                 src_disk->getObjectStorage()->getAzureBlobStorageClient(),
                 client,
-                /* src_container */ blob_path[1],
+                /* src_container */ src_blob_path[1],
-                /* src_path */ blob_path[0],
+                /* src_path */ src_blob_path[0],
                 start_pos,
                 length,
-                configuration.container,
+                connection_params.getContainer(),
-                fs::path(configuration.blob_path) / path_in_backup,
+                fs::path(blob_path) / path_in_backup,
                 settings,
                 read_settings,
                 threadPoolCallbackRunnerUnsafe<void>(getBackupsIOThreadPool().get(), "BackupWRAzure"));
@@ -188,11 +194,11 @@ void BackupWriterAzureBlobStorage::copyFile(const String & destination, const St
     copyAzureBlobStorageFile(
        client,
        client,
-       configuration.container,
+       connection_params.getContainer(),
-       fs::path(configuration.blob_path)/ source,
+       fs::path(blob_path)/ source,
        0,
        size,
-       /* dest_container */ configuration.container,
+       /* dest_container */ connection_params.getContainer(),
        /* dest_path */ destination,
        settings,
        read_settings,
@@ -206,22 +212,28 @@ void BackupWriterAzureBlobStorage::copyDataToFile(
     UInt64 length)
 {
     copyDataToAzureBlobStorageFile(
-        create_read_buffer, start_pos, length, client, configuration.container,
+        create_read_buffer,
-        fs::path(configuration.blob_path) / path_in_backup, settings,
+        start_pos,
-        threadPoolCallbackRunnerUnsafe<void>(getBackupsIOThreadPool().get(), "BackupWRAzure"));
+        length,
+        client,
+        connection_params.getContainer(),
+        fs::path(blob_path) / path_in_backup,
+        settings,
+        threadPoolCallbackRunnerUnsafe<void>(getBackupsIOThreadPool().get(),
+        "BackupWRAzure"));
 }
 
 BackupWriterAzureBlobStorage::~BackupWriterAzureBlobStorage() = default;
 
 bool BackupWriterAzureBlobStorage::fileExists(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     return object_storage->exists(StoredObject(key));
 }
 
 UInt64 BackupWriterAzureBlobStorage::getFileSize(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     RelativePathsWithMetadata children;
     object_storage->listObjects(key,children,/*max_keys*/0);
     if (children.empty())
@@ -231,7 +243,7 @@ UInt64 BackupWriterAzureBlobStorage::getFileSize(const String & file_name)
 
 std::unique_ptr<ReadBuffer> BackupWriterAzureBlobStorage::readFile(const String & file_name, size_t /*expected_file_size*/)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     return std::make_unique<ReadBufferFromAzureBlobStorage>(
         client, key, read_settings, settings->max_single_read_retries,
         settings->max_single_download_retries);
@@ -239,7 +251,7 @@ std::unique_ptr<ReadBuffer> BackupWriterAzureBlobStorage::readFile(const String
 
 std::unique_ptr<WriteBuffer> BackupWriterAzureBlobStorage::writeFile(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     return std::make_unique<WriteBufferFromAzureBlobStorage>(
         client,
         key,
@@ -251,7 +263,7 @@ std::unique_ptr<WriteBuffer> BackupWriterAzureBlobStorage::writeFile(const Strin
 
 void BackupWriterAzureBlobStorage::removeFile(const String & file_name)
 {
-    String key = fs::path(configuration.blob_path) / file_name;
+    String key = fs::path(blob_path) / file_name;
     StoredObject object(key);
     object_storage->removeObjectIfExists(object);
 }
@@ -260,7 +272,7 @@ void BackupWriterAzureBlobStorage::removeFiles(const Strings & file_names)
 {
     StoredObjects objects;
     for (const auto & file_name : file_names)
-        objects.emplace_back(fs::path(configuration.blob_path) / file_name);
+        objects.emplace_back(fs::path(blob_path) / file_name);
 
     object_storage->removeObjectsIfExist(objects);
 
@@ -270,7 +282,7 @@ void BackupWriterAzureBlobStorage::removeFilesBatch(const Strings & file_names)
 {
     StoredObjects objects;
     for (const auto & file_name : file_names)
-        objects.emplace_back(fs::path(configuration.blob_path) / file_name);
+        objects.emplace_back(fs::path(blob_path) / file_name);
 
     object_storage->removeObjectsIfExist(objects);
 }

src/Backups/BackupIO_AzureBlobStorage.h

@@ -1,12 +1,10 @@
 #pragma once
-
 #include "config.h"
 
 #if USE_AZURE_BLOB_STORAGE
 #include <Backups/BackupIO_Default.h>
 #include <Disks/DiskType.h>
-#include <Interpreters/Context_fwd.h>
+#include <Disks/ObjectStorages/AzureBlobStorage/AzureObjectStorage.h>
-#include <Storages/ObjectStorage/Azure/Configuration.h>
 
 
 namespace DB
@@ -17,7 +15,8 @@ class BackupReaderAzureBlobStorage : public BackupReaderDefault
 {
 public:
     BackupReaderAzureBlobStorage(
-        const StorageAzureConfiguration & configuration_,
+        const AzureBlobStorage::ConnectionParams & connection_params_,
+        const String & blob_path_,
         bool allow_azure_native_copy,
         const ReadSettings & read_settings_,
         const WriteSettings & write_settings_,
@@ -40,16 +39,18 @@ public:
 private:
     const DataSourceDescription data_source_description;
     std::shared_ptr<const Azure::Storage::Blobs::BlobContainerClient> client;
-    StorageAzureConfiguration configuration;
+    AzureBlobStorage::ConnectionParams connection_params;
+    String blob_path;
     std::unique_ptr<AzureObjectStorage> object_storage;
-    std::shared_ptr<const AzureObjectStorageSettings> settings;
+    std::shared_ptr<const AzureBlobStorage::RequestSettings> settings;
 };
 
 class BackupWriterAzureBlobStorage : public BackupWriterDefault
 {
 public:
     BackupWriterAzureBlobStorage(
-        const StorageAzureConfiguration & configuration_,
+        const AzureBlobStorage::ConnectionParams & connection_params_,
+        const String & blob_path_,
         bool allow_azure_native_copy,
         const ReadSettings & read_settings_,
         const WriteSettings & write_settings_,
@@ -87,9 +88,10 @@ private:
 
     const DataSourceDescription data_source_description;
     std::shared_ptr<const Azure::Storage::Blobs::BlobContainerClient> client;
-    StorageAzureConfiguration configuration;
+    AzureBlobStorage::ConnectionParams connection_params;
+    String blob_path;
     std::unique_ptr<AzureObjectStorage> object_storage;
-    std::shared_ptr<const AzureObjectStorageSettings> settings;
+    std::shared_ptr<const AzureBlobStorage::RequestSettings> settings;
 };
 
 }

src/Backups/registerBackupEngineAzureBlobStorage.cpp

@@ -5,6 +5,7 @@
 
 #if USE_AZURE_BLOB_STORAGE
 #include <Backups/BackupIO_AzureBlobStorage.h>
+#include <Disks/ObjectStorages/AzureBlobStorage/AzureBlobStorageCommon.h>
 #include <Backups/BackupImpl.h>
 #include <IO/Archives/hasRegisteredArchiveFileExtension.h>
 #include <Interpreters/Context.h>
@@ -49,7 +50,9 @@ void registerBackupEngineAzureBlobStorage(BackupFactory & factory)
         const String & id_arg = params.backup_info.id_arg;
         const auto & args = params.backup_info.args;
 
-        StorageAzureConfiguration configuration;
+        String blob_path;
+        AzureBlobStorage::ConnectionParams connection_params;
+        auto request_settings = AzureBlobStorage::getRequestSettings(params.context->getSettingsRef());
 
         if (!id_arg.empty())
         {
@@ -59,55 +62,42 @@ void registerBackupEngineAzureBlobStorage(BackupFactory & factory)
             if (!config.has(config_prefix))
                 throw Exception(ErrorCodes::BAD_ARGUMENTS, "There is no collection named `{}` in config", id_arg);
 
-            if (config.has(config_prefix + ".connection_string"))
+            connection_params =
             {
-                configuration.connection_url = config.getString(config_prefix + ".connection_string");
+                .endpoint = AzureBlobStorage::processEndpoint(config, config_prefix),
-                configuration.is_connection_string = true;
+                .auth_method = AzureBlobStorage::getAuthMethod(config, config_prefix),
-                configuration.container = config.getString(config_prefix + ".container");
+                .client_options = AzureBlobStorage::getClientOptions(*request_settings, /*for_disk=*/ true),
-            }
+            };
-            else
-            {
-                configuration.connection_url = config.getString(config_prefix + ".storage_account_url");
-                configuration.is_connection_string = false;
-                configuration.container =  config.getString(config_prefix + ".container");
-                configuration.account_name = config.getString(config_prefix + ".account_name");
-                configuration.account_key =  config.getString(config_prefix + ".account_key");
-
-                if (config.has(config_prefix + ".account_name") && config.has(config_prefix + ".account_key"))
-                {
-                    configuration.account_name = config.getString(config_prefix + ".account_name");
-                    configuration.account_key = config.getString(config_prefix + ".account_key");
-                }
-            }
 
             if (args.size() > 1)
                 throw Exception(ErrorCodes::NUMBER_OF_ARGUMENTS_DOESNT_MATCH,
                                 "Backup AzureBlobStorage requires 1 or 2 arguments: named_collection, [filename]");
 
             if (args.size() == 1)
-                configuration.setPath(args[0].safeGet<String>());
+                blob_path = args[0].safeGet<String>();
-
         }
         else
         {
             if (args.size() == 3)
             {
-                configuration.connection_url = args[0].safeGet<String>();
+                auto connection_url = args[0].safeGet<String>();
-                configuration.is_connection_string = !configuration.connection_url.starts_with("http");
+                auto container_name = args[1].safeGet<String>();
+                blob_path = args[2].safeGet<String>();
 
-                configuration.container =  args[1].safeGet<String>();
+                AzureBlobStorage::processURL(connection_url, container_name, connection_params.endpoint, connection_params.auth_method);
-                configuration.blob_path = args[2].safeGet<String>();
+                connection_params.client_options = AzureBlobStorage::getClientOptions(*request_settings, /*for_disk=*/ true);
             }
             else if (args.size() == 5)
             {
-                configuration.connection_url = args[0].safeGet<String>();
+                connection_params.endpoint.storage_account_url = args[0].safeGet<String>();
-                configuration.is_connection_string = false;
+                connection_params.endpoint.container_name = args[1].safeGet<String>();
+                blob_path = args[2].safeGet<String>();
 
-                configuration.container =  args[1].safeGet<String>();
+                auto account_name = args[3].safeGet<String>();
-                configuration.blob_path = args[2].safeGet<String>();
+                auto account_key = args[4].safeGet<String>();
-                configuration.account_name = args[3].safeGet<String>();
-                configuration.account_key = args[4].safeGet<String>();
 
+                connection_params.auth_method = std::make_shared<Azure::Storage::StorageSharedKeyCredential>(account_name, account_key);
+                connection_params.client_options = AzureBlobStorage::getClientOptions(*request_settings, /*for_disk=*/ true);
             }
             else
             {
@@ -117,16 +107,12 @@ void registerBackupEngineAzureBlobStorage(BackupFactory & factory)
         }
 
         BackupImpl::ArchiveParams archive_params;
-        if (hasRegisteredArchiveFileExtension(configuration.getPath()))
+        if (hasRegisteredArchiveFileExtension(blob_path))
         {
             if (params.is_internal_backup)
                 throw Exception(ErrorCodes::SUPPORT_IS_DISABLED, "Using archives with backups on clusters is disabled");
 
-            auto path = configuration.getPath();
+            archive_params.archive_name = removeFileNameFromURL(blob_path);
-            auto filename = removeFileNameFromURL(path);
-            configuration.setPath(path);
-
-            archive_params.archive_name = filename;
             archive_params.compression_method = params.compression_method;
             archive_params.compression_level = params.compression_level;
             archive_params.password = params.password;
@@ -141,7 +127,8 @@ void registerBackupEngineAzureBlobStorage(BackupFactory & factory)
         if (params.open_mode == IBackup::OpenMode::READ)
         {
             auto reader = std::make_shared<BackupReaderAzureBlobStorage>(
-                configuration,
+                connection_params,
+                blob_path,
                 params.allow_azure_native_copy,
                 params.read_settings,
                 params.write_settings,
@@ -159,7 +146,8 @@ void registerBackupEngineAzureBlobStorage(BackupFactory & factory)
         else
         {
             auto writer = std::make_shared<BackupWriterAzureBlobStorage>(
-                configuration,
+                connection_params,
+                blob_path,
                 params.allow_azure_native_copy,
                 params.read_settings,
                 params.write_settings,

src/CMakeLists.txt

@@ -607,6 +607,10 @@ if (TARGET ch_contrib::usearch)
     dbms_target_link_libraries(PUBLIC ch_contrib::usearch)
 endif()
 
+if (TARGET ch_contrib::prometheus_protobufs)
+    dbms_target_link_libraries (PUBLIC ch_contrib::prometheus_protobufs)
+endif()
+
 if (TARGET ch_rust::skim)
     dbms_target_include_directories(PRIVATE $<TARGET_PROPERTY:ch_rust::skim,INTERFACE_INCLUDE_DIRECTORIES>)
     dbms_target_link_libraries(PUBLIC ch_rust::skim)

src/Client/ClientBase.cpp

@@ -1206,11 +1206,8 @@ void ClientBase::receiveResult(ASTPtr parsed_query, Int32 signals_before_stop, b
     if (local_format_error)
         std::rethrow_exception(local_format_error);
 
-    if (cancelled && is_interactive)
+    if (cancelled && is_interactive && !cancelled_printed.exchange(true))
-    {
         output_stream << "Query was cancelled." << std::endl;
-        cancelled_printed = true;
-    }
 }
 
 
@@ -1326,7 +1323,7 @@ void ClientBase::onEndOfStream()
 
     if (is_interactive)
     {
-        if (cancelled && !cancelled_printed)
+        if (cancelled && !cancelled_printed.exchange(true))
             output_stream << "Query was cancelled." << std::endl;
         else if (!written_first_block)
             output_stream << "Ok." << std::endl;

src/Client/ClientBase.h

@@ -6,13 +6,13 @@
 #include <Common/ProgressIndication.h>
 #include <Common/InterruptListener.h>
 #include <Common/ShellCommand.h>
+#include <Common/QueryFuzzer.h>
 #include <Common/Stopwatch.h>
 #include <Common/DNSResolver.h>
 #include <Core/ExternalTable.h>
 #include <Poco/Util/Application.h>
 #include <Interpreters/Context.h>
 #include <Client/Suggest.h>
-#include <Client/QueryFuzzer.h>
 #include <boost/program_options.hpp>
 #include <Storages/StorageFile.h>
 #include <Storages/SelectQueryInfo.h>
@@ -338,8 +338,8 @@ protected:
     bool allow_repeated_settings = false;
     bool allow_merge_tree_settings = false;
 
-    bool cancelled = false;
+    std::atomic_bool cancelled = false;
-    bool cancelled_printed = false;
+    std::atomic_bool cancelled_printed = false;
 
     /// Unpacked descriptors and streams for the ease of use.
     int in_fd = STDIN_FILENO;

src/Client/HedgedConnections.cpp

@@ -195,6 +195,12 @@ void HedgedConnections::sendQuery(
             modified_settings.parallel_replica_offset = fd_to_replica_location[replica.packet_receiver->getFileDescriptor()].offset;
         }
 
+        /// FIXME: Remove once we will make `allow_experimental_analyzer` obsolete setting.
+        /// Make the analyzer being set, so it will be effectively applied on the remote server.
+        /// In other words, the initiator always controls whether the analyzer enabled or not for
+        /// all servers involved in the distributed query processing.
+        modified_settings.set("allow_experimental_analyzer", static_cast<bool>(modified_settings.allow_experimental_analyzer));
+
         replica.connection->sendQuery(timeouts, query, /* query_parameters */ {}, query_id, stage, &modified_settings, &client_info, with_pending_data, {});
         replica.change_replica_timeout.setRelative(timeouts.receive_data_timeout);
         replica.packet_receiver->setTimeout(hedged_connections_factory.getConnectionTimeouts().receive_timeout);

src/Client/MultiplexedConnections.cpp

@@ -150,6 +150,12 @@ void MultiplexedConnections::sendQuery(
         }
     }
 
+    /// FIXME: Remove once we will make `allow_experimental_analyzer` obsolete setting.
+    /// Make the analyzer being set, so it will be effectively applied on the remote server.
+    /// In other words, the initiator always controls whether the analyzer enabled or not for
+    /// all servers involved in the distributed query processing.
+    modified_settings.set("allow_experimental_analyzer", static_cast<bool>(modified_settings.allow_experimental_analyzer));
+
     const bool enable_sample_offset_parallel_processing = settings.max_parallel_replicas > 1 && settings.allow_experimental_parallel_reading_from_replicas == 0;
 
     size_t num_replicas = replica_states.size();

src/Columns/ColumnObject.cpp

@@ -1093,4 +1093,10 @@ void ColumnObject::finalize()
     checkObjectHasNoAmbiguosPaths(getKeys());
 }
 
+void ColumnObject::updateHashFast(SipHash & hash) const
+{
+    for (const auto & entry : subcolumns)
+        for (auto & part : entry->data.data)
+            part->updateHashFast(hash);
+}
 }

src/Columns/ColumnObject.h

@@ -242,7 +242,7 @@ public:
     const char * skipSerializedInArena(const char *) const override { throwMustBeConcrete(); }
     void updateHashWithValue(size_t, SipHash &) const override { throwMustBeConcrete(); }
     void updateWeakHash32(WeakHash32 &) const override { throwMustBeConcrete(); }
-    void updateHashFast(SipHash &) const override { throwMustBeConcrete(); }
+    void updateHashFast(SipHash & hash) const override;
     void expand(const Filter &, bool) override { throwMustBeConcrete(); }
     bool hasEqualValues() const override { throwMustBeConcrete(); }
     size_t byteSizeAt(size_t) const override { throwMustBeConcrete(); }

src/Common/CgroupsMemoryUsageObserver.cpp

@@ -12,7 +12,9 @@
 #include <base/getMemoryAmount.h>
 #include <base/sleep.h>
 
+#include <cstdint>
 #include <filesystem>
+#include <memory>
 #include <optional>
 
 #include "config.h"
@@ -22,24 +24,169 @@
 #define STRINGIFY(x) STRINGIFY_HELPER(x)
 #endif
 
+using namespace DB;
 
 namespace DB
 {
 
 namespace ErrorCodes
 {
-    extern const int CANNOT_CLOSE_FILE;
+extern const int FILE_DOESNT_EXIST;
-    extern const int CANNOT_OPEN_FILE;
+extern const int INCORRECT_DATA;
-    extern const int FILE_DOESNT_EXIST;
-    extern const int INCORRECT_DATA;
 }
 
-CgroupsMemoryUsageObserver::CgroupsMemoryUsageObserver(std::chrono::seconds wait_time_)
+}
-    : log(getLogger("CgroupsMemoryUsageObserver"))
+
-    , wait_time(wait_time_)
+namespace
-    , memory_usage_file(log)
 {
-    LOG_INFO(log, "Initialized cgroups memory limit observer, wait time is {} sec", wait_time.count());
+
+/// Format is
+///   kernel 5
+///   rss 15
+///   [...]
+uint64_t readMetricFromStatFile(ReadBufferFromFile & buf, const std::string & key)
+{
+    while (!buf.eof())
+    {
+        std::string current_key;
+        readStringUntilWhitespace(current_key, buf);
+        if (current_key != key)
+        {
+            std::string dummy;
+            readStringUntilNewlineInto(dummy, buf);
+            buf.ignore();
+            continue;
+        }
+
+        assertChar(' ', buf);
+        uint64_t value = 0;
+        readIntText(value, buf);
+        return value;
+    }
+
+    throw Exception(ErrorCodes::INCORRECT_DATA, "Cannot find '{}' in '{}'", key, buf.getFileName());
+}
+
+struct CgroupsV1Reader : ICgroupsReader
+{
+    explicit CgroupsV1Reader(const std::filesystem::path & stat_file_dir) : buf(stat_file_dir / "memory.stat") { }
+
+    uint64_t readMemoryUsage() override
+    {
+        std::lock_guard lock(mutex);
+        buf.rewind();
+        return readMetricFromStatFile(buf, "rss");
+    }
+
+private:
+    std::mutex mutex;
+    ReadBufferFromFile buf TSA_GUARDED_BY(mutex);
+};
+
+struct CgroupsV2Reader : ICgroupsReader
+{
+    explicit CgroupsV2Reader(const std::filesystem::path & stat_file_dir)
+        : current_buf(stat_file_dir / "memory.current"), stat_buf(stat_file_dir / "memory.stat")
+    {
+    }
+
+    uint64_t readMemoryUsage() override
+    {
+        std::lock_guard lock(mutex);
+        current_buf.rewind();
+        stat_buf.rewind();
+
+        int64_t mem_usage = 0;
+        /// memory.current contains a single number
+        /// the reason why we subtract it described here: https://github.com/ClickHouse/ClickHouse/issues/64652#issuecomment-2149630667
+        readIntText(mem_usage, current_buf);
+        mem_usage -= readMetricFromStatFile(stat_buf, "inactive_file");
+        chassert(mem_usage >= 0, "Negative memory usage");
+        return mem_usage;
+    }
+
+private:
+    std::mutex mutex;
+    ReadBufferFromFile current_buf TSA_GUARDED_BY(mutex);
+    ReadBufferFromFile stat_buf TSA_GUARDED_BY(mutex);
+};
+
+/// Caveats:
+/// - All of the logic in this file assumes that the current process is the only process in the
+///   containing cgroup (or more precisely: the only process with significant memory consumption).
+///   If this is not the case, then other processe's memory consumption may affect the internal
+///   memory tracker ...
+/// - Cgroups v1 and v2 allow nested cgroup hierarchies. As v1 is deprecated for over half a
+///   decade and will go away at some point, hierarchical detection is only implemented for v2.
+/// - I did not test what happens if a host has v1 and v2 simultaneously enabled. I believe such
+///   systems existed only for a short transition period.
+
+std::optional<std::string> getCgroupsV2Path()
+{
+    if (!cgroupsV2Enabled())
+        return {};
+
+    if (!cgroupsV2MemoryControllerEnabled())
+        return {};
+
+    String cgroup = cgroupV2OfProcess();
+    auto current_cgroup = cgroup.empty() ? default_cgroups_mount : (default_cgroups_mount / cgroup);
+
+    /// Return the bottom-most nested current memory file. If there is no such file at the current
+    /// level, try again at the parent level as memory settings are inherited.
+    while (current_cgroup != default_cgroups_mount.parent_path())
+    {
+        const auto current_path = current_cgroup / "memory.current";
+        const auto stat_path = current_cgroup / "memory.stat";
+        if (std::filesystem::exists(current_path) && std::filesystem::exists(stat_path))
+            return {current_cgroup};
+        current_cgroup = current_cgroup.parent_path();
+    }
+    return {};
+}
+
+std::optional<std::string> getCgroupsV1Path()
+{
+    auto path = default_cgroups_mount / "memory/memory.stat";
+    if (!std::filesystem::exists(path))
+        return {};
+    return {default_cgroups_mount / "memory"};
+}
+
+std::pair<std::string, CgroupsMemoryUsageObserver::CgroupsVersion> getCgroupsPath()
+{
+    auto v2_path = getCgroupsV2Path();
+    if (v2_path.has_value())
+        return {*v2_path, CgroupsMemoryUsageObserver::CgroupsVersion::V2};
+
+    auto v1_path = getCgroupsV1Path();
+    if (v1_path.has_value())
+        return {*v1_path, CgroupsMemoryUsageObserver::CgroupsVersion::V1};
+
+    throw Exception(ErrorCodes::FILE_DOESNT_EXIST, "Cannot find cgroups v1 or v2 current memory file");
+}
+
+}
+
+namespace DB
+{
+
+CgroupsMemoryUsageObserver::CgroupsMemoryUsageObserver(std::chrono::seconds wait_time_)
+    : log(getLogger("CgroupsMemoryUsageObserver")), wait_time(wait_time_)
+{
+    const auto [cgroup_path, version] = getCgroupsPath();
+
+    if (version == CgroupsVersion::V2)
+        cgroup_reader = std::make_unique<CgroupsV2Reader>(cgroup_path);
+    else
+        cgroup_reader = std::make_unique<CgroupsV1Reader>(cgroup_path);
+
+    LOG_INFO(
+        log,
+        "Will read the current memory usage from '{}' (cgroups version: {}), wait time is {} sec",
+        cgroup_path,
+        (version == CgroupsVersion::V1) ? "v1" : "v2",
+        wait_time.count());
 }
 
 CgroupsMemoryUsageObserver::~CgroupsMemoryUsageObserver()
@@ -79,12 +226,13 @@ void CgroupsMemoryUsageObserver::setMemoryUsageLimits(uint64_t hard_limit_, uint
         {
             LOG_WARNING(log, "Exceeded soft memory limit ({})", ReadableSize(soft_limit_));
 
-#if USE_JEMALLOC
+#    if USE_JEMALLOC
             LOG_INFO(log, "Purging jemalloc arenas");
             mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".purge", nullptr, nullptr, nullptr, 0);
-#endif
+#    endif
             /// Reset current usage in memory tracker. Expect zero for free_memory_in_allocator_arenas as we just purged them.
-            uint64_t memory_usage = memory_usage_file.readMemoryUsage();
+            uint64_t memory_usage = cgroup_reader->readMemoryUsage();
+            LOG_TRACE(log, "Read current memory usage {} bytes ({}) from cgroups", memory_usage, ReadableSize(memory_usage));
             MemoryTracker::setRSS(memory_usage, 0);
 
             LOG_INFO(log, "Purged jemalloc arenas. Current memory usage is {}", ReadableSize(memory_usage));
@@ -104,152 +252,6 @@ void CgroupsMemoryUsageObserver::setOnMemoryAmountAvailableChangedFn(OnMemoryAmo
     on_memory_amount_available_changed = on_memory_amount_available_changed_;
 }
 
-namespace
-{
-
-/// Caveats:
-/// - All of the logic in this file assumes that the current process is the only process in the
-///   containing cgroup (or more precisely: the only process with significant memory consumption).
-///   If this is not the case, then other processe's memory consumption may affect the internal
-///   memory tracker ...
-/// - Cgroups v1 and v2 allow nested cgroup hierarchies. As v1 is deprecated for over half a
-///   decade and will go away at some point, hierarchical detection is only implemented for v2.
-/// - I did not test what happens if a host has v1 and v2 simultaneously enabled. I believe such
-///   systems existed only for a short transition period.
-
-std::optional<std::string> getCgroupsV2FileName()
-{
-    if (!cgroupsV2Enabled())
-        return {};
-
-    if (!cgroupsV2MemoryControllerEnabled())
-        return {};
-
-    String cgroup = cgroupV2OfProcess();
-    auto current_cgroup = cgroup.empty() ? default_cgroups_mount : (default_cgroups_mount / cgroup);
-
-    /// Return the bottom-most nested current memory file. If there is no such file at the current
-    /// level, try again at the parent level as memory settings are inherited.
-    while (current_cgroup != default_cgroups_mount.parent_path())
-    {
-        auto path = current_cgroup / "memory.current";
-        if (std::filesystem::exists(path))
-            return {path};
-        current_cgroup = current_cgroup.parent_path();
-    }
-    return {};
-}
-
-std::optional<std::string> getCgroupsV1FileName()
-{
-    auto path = default_cgroups_mount / "memory/memory.stat";
-    if (!std::filesystem::exists(path))
-        return {};
-    return {path};
-}
-
-std::pair<std::string, CgroupsMemoryUsageObserver::CgroupsVersion> getCgroupsFileName()
-{
-    auto v2_file_name = getCgroupsV2FileName();
-    if (v2_file_name.has_value())
-        return {*v2_file_name, CgroupsMemoryUsageObserver::CgroupsVersion::V2};
-
-    auto v1_file_name = getCgroupsV1FileName();
-    if (v1_file_name.has_value())
-        return {*v1_file_name, CgroupsMemoryUsageObserver::CgroupsVersion::V1};
-
-    throw Exception(ErrorCodes::FILE_DOESNT_EXIST, "Cannot find cgroups v1 or v2 current memory file");
-}
-
-}
-
-CgroupsMemoryUsageObserver::MemoryUsageFile::MemoryUsageFile(LoggerPtr log_)
-    : log(log_)
-{
-    std::tie(file_name, version) = getCgroupsFileName();
-
-    LOG_INFO(log, "Will read the current memory usage from '{}' (cgroups version: {})", file_name, (version == CgroupsVersion::V1) ? "v1" : "v2");
-
-    fd = ::open(file_name.data(), O_RDONLY);
-    if (fd == -1)
-        ErrnoException::throwFromPath(
-            (errno == ENOENT) ? ErrorCodes::FILE_DOESNT_EXIST : ErrorCodes::CANNOT_OPEN_FILE,
-            file_name, "Cannot open file '{}'", file_name);
-}
-
-CgroupsMemoryUsageObserver::MemoryUsageFile::~MemoryUsageFile()
-{
-    assert(fd != -1);
-    if (::close(fd) != 0)
-    {
-        try
-        {
-            ErrnoException::throwFromPath(
-                ErrorCodes::CANNOT_CLOSE_FILE,
-                file_name, "Cannot close file '{}'", file_name);
-        }
-        catch (const ErrnoException &)
-        {
-            tryLogCurrentException(log, __PRETTY_FUNCTION__);
-        }
-    }
-}
-
-uint64_t CgroupsMemoryUsageObserver::MemoryUsageFile::readMemoryUsage() const
-{
-    /// File read is probably not read is thread-safe, just to be sure
-    std::lock_guard lock(mutex);
-
-    ReadBufferFromFileDescriptor buf(fd);
-    buf.rewind();
-
-    uint64_t mem_usage = 0;
-
-    switch (version)
-    {
-        case CgroupsVersion::V1:
-        {
-            /// Format is
-            ///   kernel 5
-            ///   rss 15
-            ///   [...]
-            std::string key;
-            bool found_rss = false;
-
-            while (!buf.eof())
-            {
-                readStringUntilWhitespace(key, buf);
-                if (key != "rss")
-                {
-                    std::string dummy;
-                    readStringUntilNewlineInto(dummy, buf);
-                    buf.ignore();
-                    continue;
-                }
-
-                assertChar(' ', buf);
-                readIntText(mem_usage, buf);
-                found_rss = true;
-                break;
-            }
-
-            if (!found_rss)
-                throw Exception(ErrorCodes::INCORRECT_DATA, "Cannot find 'rss' in '{}'", file_name);
-
-            break;
-        }
-        case CgroupsVersion::V2:
-        {
-            readIntText(mem_usage, buf);
-            break;
-        }
-    }
-
-    LOG_TRACE(log, "Read current memory usage {} from cgroups", ReadableSize(mem_usage));
-
-    return mem_usage;
-}
-
 void CgroupsMemoryUsageObserver::startThread()
 {
     if (!thread.joinable())
@@ -301,7 +303,8 @@ void CgroupsMemoryUsageObserver::runThread()
             std::lock_guard<std::mutex> limit_lock(limit_mutex);
             if (soft_limit > 0 && hard_limit > 0)
             {
-                uint64_t memory_usage = memory_usage_file.readMemoryUsage();
+                uint64_t memory_usage = cgroup_reader->readMemoryUsage();
+                LOG_TRACE(log, "Read current memory usage {} bytes ({}) from cgroups", memory_usage, ReadableSize(memory_usage));
                 if (memory_usage > hard_limit)
                 {
                     if (last_memory_usage <= hard_limit)

src/Common/CgroupsMemoryUsageObserver.h

@@ -3,11 +3,19 @@
 #include <Common/ThreadPool.h>
 
 #include <chrono>
+#include <memory>
 #include <mutex>
 
 namespace DB
 {
 
+struct ICgroupsReader
+{
+    virtual ~ICgroupsReader() = default;
+
+    virtual uint64_t readMemoryUsage() = 0;
+};
+
 /// Does two things:
 /// 1. Periodically reads the memory usage of the process from Linux cgroups.
 ///    You can specify soft or hard memory limits:
@@ -61,27 +69,12 @@ private:
     uint64_t last_memory_usage = 0;        /// how much memory does the process use
     uint64_t last_available_memory_amount; /// how much memory can the process use
 
-    /// Represents the cgroup virtual file that shows the memory consumption of the process's cgroup.
-    struct MemoryUsageFile
-    {
-    public:
-        explicit MemoryUsageFile(LoggerPtr log_);
-        ~MemoryUsageFile();
-        uint64_t readMemoryUsage() const;
-    private:
-        LoggerPtr log;
-        mutable std::mutex mutex;
-        int fd TSA_GUARDED_BY(mutex) = -1;
-        CgroupsVersion version;
-        std::string file_name;
-    };
-
-    MemoryUsageFile memory_usage_file;
-
     void stopThread();
 
     void runThread();
 
+    std::unique_ptr<ICgroupsReader> cgroup_reader;
+
     std::mutex thread_mutex;
     std::condition_variable cond;
     ThreadFromGlobalPool thread;

src/Common/CollectionOfDerived.h

@@ -0,0 +1,184 @@
+#pragma once
+
+#include <base/defines.h>
+
+#include <Common/Exception.h>
+
+#include <algorithm>
+#include <memory>
+#include <typeindex>
+#include <vector>
+#include <string>
+
+
+namespace DB
+{
+
+namespace ErrorCodes
+{
+    extern const int LOGICAL_ERROR;
+}
+
+/* This is a collections of objects derived from ItemBase.
+*  Collection contains no more than one instance for each derived type.
+*  The derived type is used to access the instance.
+*/
+
+template<class ItemBase>
+class CollectionOfDerivedItems
+{
+public:
+    using Self = CollectionOfDerivedItems<ItemBase>;
+    using ItemPtr = std::shared_ptr<ItemBase>;
+
+private:
+    struct Rec
+    {
+        std::type_index type_idx;
+        ItemPtr ptr;
+
+        bool operator<(const Rec & other) const
+        {
+            return type_idx < other.type_idx;
+        }
+
+        bool operator<(const std::type_index & value) const
+        {
+            return type_idx < value;
+        }
+
+        bool operator==(const Rec & other) const
+        {
+            return type_idx == other.type_idx;
+        }
+    };
+    using Records = std::vector<Rec>;
+
+public:
+    void swap(Self & other) noexcept
+    {
+        records.swap(other.records);
+    }
+
+    void clear()
+    {
+        records.clear();
+    }
+
+    bool empty() const
+    {
+        return records.empty();
+    }
+
+    size_t size() const
+    {
+        return records.size();
+    }
+
+    Self clone() const
+    {
+        Self result;
+        result.records.reserve(records.size());
+        for (const auto & rec : records)
+            result.records.emplace_back(rec.type_idx, rec.ptr->clone());
+        return result;
+    }
+
+    void append(Self && other)
+    {
+        auto middle_idx = records.size();
+        std::move(other.records.begin(), other.records.end(), std::back_inserter(records));
+        std::inplace_merge(records.begin(), records.begin() + middle_idx, records.end());
+        chassert(isUniqTypes());
+    }
+
+    template <class T>
+    void add(std::shared_ptr<T> info)
+    {
+        static_assert(std::is_base_of_v<ItemBase, T>, "Template parameter must inherit items base class");
+        return addImpl(std::type_index(typeid(T)), std::move(info));
+    }
+
+    template <class T>
+    std::shared_ptr<T> get() const
+    {
+        static_assert(std::is_base_of_v<ItemBase, T>, "Template parameter must inherit items base class");
+        auto it = getImpl(std::type_index(typeid(T)));
+        if (it == records.cend())
+            return nullptr;
+        auto cast = std::dynamic_pointer_cast<T>(it->ptr);
+        chassert(cast);
+        return cast;
+    }
+
+    template <class T>
+    std::shared_ptr<T> extract()
+    {
+        static_assert(std::is_base_of_v<ItemBase, T>, "Template parameter must inherit items base class");
+        auto it = getImpl(std::type_index(typeid(T)));
+        if (it == records.cend())
+            return nullptr;
+        auto cast = std::dynamic_pointer_cast<T>(it->ptr);
+        chassert(cast);
+
+        records.erase(it);
+        return cast;
+    }
+
+    std::string debug() const
+    {
+        std::string result;
+
+        for (auto & rec : records)
+        {
+            result.append(rec.type_idx.name());
+            result.append(" ");
+        }
+
+        return result;
+    }
+
+private:
+    bool isUniqTypes() const
+    {
+        auto uniq_it = std::adjacent_find(records.begin(), records.end());
+
+        return uniq_it == records.end();
+    }
+
+    void addImpl(std::type_index type_idx, ItemPtr item)
+    {
+        auto it = std::lower_bound(records.begin(), records.end(), type_idx);
+
+        if (it == records.end())
+        {
+            records.emplace_back(type_idx, item);
+            return;
+        }
+
+        if (it->type_idx == type_idx)
+            throw Exception(ErrorCodes::LOGICAL_ERROR, "inserted items must be unique by their type, type {} is inserted twice", type_idx.name());
+
+
+        records.emplace(it, type_idx, item);
+
+        chassert(isUniqTypes());
+    }
+
+    Records::const_iterator getImpl(std::type_index type_idx) const
+    {
+        auto it = std::lower_bound(records.cbegin(), records.cend(), type_idx);
+
+        if (it == records.cend())
+            return records.cend();
+
+        if (it->type_idx != type_idx)
+            return records.cend();
+
+        return it;
+    }
+
+    Records records;
+};
+
+}

src/Common/ConcurrentBoundedQueue.h

@@ -1,8 +1,6 @@
 #pragma once
 
 #include <deque>
-#include <type_traits>
-#include <atomic>
 #include <condition_variable>
 #include <mutex>
 #include <optional>
@@ -200,22 +198,18 @@ public:
       */
     bool finish()
     {
-        bool was_finished_before = false;
-
         {
             std::lock_guard lock(queue_mutex);
 
             if (is_finished)
                 return true;
 
-            was_finished_before = is_finished;
             is_finished = true;
         }
 
         pop_condition.notify_all();
         push_condition.notify_all();
-
+        return false;
-        return was_finished_before;
     }
 
     /// Returns if queue is finished

src/Common/Config/ConfigProcessor.cpp

@@ -316,7 +316,6 @@ void ConfigProcessor::mergeRecursive(XMLDocumentPtr config, Node * config_root,
                 }
                 else if (replace)
                 {
-                    with_element.removeAttribute("replace");
                     NodePtr new_node = config->importNode(with_node, true);
                     config_root->replaceChild(new_node, config_node);
                 }

src/Common/Coverage.cpp

@@ -0,0 +1,65 @@
+#include <Common/Coverage.h>
+
+#if defined(SANITIZE_COVERAGE)
+
+#include <fcntl.h>
+#include <unistd.h>
+
+#include <string>
+#include <vector>
+
+#include <Common/IO.h>
+#include <base/coverage.h>
+
+#include <fmt/format.h>
+
+/// Macros to avoid using strlen(), since it may fail if SSE is not supported.
+#define writeError(data) do \
+    { \
+        static_assert(__builtin_constant_p(data)); \
+        if (!writeRetry(STDERR_FILENO, data, sizeof(data) - 1)) \
+            _Exit(1); \
+    } while (false)
+
+__attribute__((no_sanitize("coverage")))
+void dumpCoverage()
+{
+    /// A user can request to dump the coverage information into files at exit.
+    /// This is useful for non-server applications such as clickhouse-format or clickhouse-client,
+    /// that cannot introspect it with SQL functions at runtime.
+
+    /// The CLICKHOUSE_WRITE_COVERAGE environment variable defines a prefix for a filename 'prefix.pid'
+    /// containing the list of addresses of covered .
+
+    /// The format is even simpler than Clang's "sancov": an array of 64-bit addresses, native byte order, no header.
+
+    if (const char * coverage_filename_prefix = getenv("CLICKHOUSE_WRITE_COVERAGE")) // NOLINT(concurrency-mt-unsafe)
+    {
+        auto dump = [](const std::string & name, auto span)
+        {
+            /// Write only non-zeros.
+            std::vector<uintptr_t> data;
+            data.reserve(span.size());
+            for (auto addr : span)
+                if (addr)
+                    data.push_back(addr);
+
+            int fd = ::open(name.c_str(), O_WRONLY | O_CREAT | O_TRUNC | O_CLOEXEC, 0400);
+            if (-1 == fd)
+            {
+                writeError("Cannot open a file to write the coverage data\n");
+            }
+            else
+            {
+                if (!writeRetry(fd, reinterpret_cast<const char *>(data.data()), data.size() * sizeof(data[0])))
+                    writeError("Cannot write the coverage data to a file\n");
+                if (0 != ::close(fd))
+                    writeError("Cannot close the file with coverage data\n");
+            }
+        };
+
+        dump(fmt::format("{}.{}", coverage_filename_prefix, getpid()), getCumulativeCoverage());
+    }
+}
+#endif
+

src/Common/Coverage.h

@@ -0,0 +1,5 @@
+#pragma once
+
+#if defined(SANITIZE_COVERAGE)
+void dumpCoverage();
+#endif

src/Common/EnvironmentChecks.cpp

@@ -0,0 +1,234 @@
+#include <Common/EnvironmentChecks.h>
+#include <Common/IO.h>
+
+#include <fmt/format.h>
+
+#include <csignal>
+#include <csetjmp>
+#include <cstdint>
+
+#include <tuple>
+
+#include <unistd.h>
+
+namespace
+{
+
+enum class InstructionFail : uint8_t
+{
+    NONE = 0,
+    SSE3 = 1,
+    SSSE3 = 2,
+    SSE4_1 = 3,
+    SSE4_2 = 4,
+    POPCNT = 5,
+    AVX = 6,
+    AVX2 = 7,
+    AVX512 = 8
+};
+
+auto instructionFailToString(InstructionFail fail)
+{
+    switch (fail)
+    {
+#define ret(x) return std::make_tuple(STDERR_FILENO, x, sizeof(x) - 1)
+        case InstructionFail::NONE:
+            ret("NONE");
+        case InstructionFail::SSE3:
+            ret("SSE3");
+        case InstructionFail::SSSE3:
+            ret("SSSE3");
+        case InstructionFail::SSE4_1:
+            ret("SSE4.1");
+        case InstructionFail::SSE4_2:
+            ret("SSE4.2");
+        case InstructionFail::POPCNT:
+            ret("POPCNT");
+        case InstructionFail::AVX:
+            ret("AVX");
+        case InstructionFail::AVX2:
+            ret("AVX2");
+        case InstructionFail::AVX512:
+            ret("AVX512");
+#undef ret
+    }
+}
+
+
+sigjmp_buf jmpbuf;
+
+[[noreturn]] void sigIllCheckHandler(int, siginfo_t *, void *)
+{
+    siglongjmp(jmpbuf, 1);
+}
+
+/// Check if necessary SSE extensions are available by trying to execute some sse instructions.
+/// If instruction is unavailable, SIGILL will be sent by kernel.
+void checkRequiredInstructionsImpl(volatile InstructionFail & fail)
+{
+#if defined(__SSE3__)
+    fail = InstructionFail::SSE3;
+    __asm__ volatile ("addsubpd %%xmm0, %%xmm0" : : : "xmm0");
+#endif
+
+#if defined(__SSSE3__)
+    fail = InstructionFail::SSSE3;
+    __asm__ volatile ("pabsw %%xmm0, %%xmm0" : : : "xmm0");
+
+#endif
+
+#if defined(__SSE4_1__)
+    fail = InstructionFail::SSE4_1;
+    __asm__ volatile ("pmaxud %%xmm0, %%xmm0" : : : "xmm0");
+#endif
+
+#if defined(__SSE4_2__)
+    fail = InstructionFail::SSE4_2;
+    __asm__ volatile ("pcmpgtq %%xmm0, %%xmm0" : : : "xmm0");
+#endif
+
+    /// Defined by -msse4.2
+#if defined(__POPCNT__)
+    fail = InstructionFail::POPCNT;
+    {
+        uint64_t a = 0;
+        uint64_t b = 0;
+        __asm__ volatile ("popcnt %1, %0" : "=r"(a) :"r"(b) :);
+    }
+#endif
+
+#if defined(__AVX__)
+    fail = InstructionFail::AVX;
+    __asm__ volatile ("vaddpd %%ymm0, %%ymm0, %%ymm0" : : : "ymm0");
+#endif
+
+#if defined(__AVX2__)
+    fail = InstructionFail::AVX2;
+    __asm__ volatile ("vpabsw %%ymm0, %%ymm0" : : : "ymm0");
+#endif
+
+#if defined(__AVX512__)
+    fail = InstructionFail::AVX512;
+    __asm__ volatile ("vpabsw %%zmm0, %%zmm0" : : : "zmm0");
+#endif
+
+    fail = InstructionFail::NONE;
+}
+
+/// Macros to avoid using strlen(), since it may fail if SSE is not supported.
+#define writeError(data) do \
+    { \
+        static_assert(__builtin_constant_p(data)); \
+        if (!writeRetry(STDERR_FILENO, data, sizeof(data) - 1)) \
+            _Exit(1); \
+    } while (false)
+
+/// Check SSE and others instructions availability. Calls exit on fail.
+/// This function must be called as early as possible, even before main, because static initializers may use unavailable instructions.
+void checkRequiredInstructions()
+{
+    struct sigaction sa{};
+    struct sigaction sa_old{};
+    sa.sa_sigaction = sigIllCheckHandler;
+    sa.sa_flags = SA_SIGINFO;
+    auto signal = SIGILL;
+    if (sigemptyset(&sa.sa_mask) != 0
+        || sigaddset(&sa.sa_mask, signal) != 0
+        || sigaction(signal, &sa, &sa_old) != 0)
+    {
+        /// You may wonder about strlen.
+        /// Typical implementation of strlen is using SSE4.2 or AVX2.
+        /// But this is not the case because it's compiler builtin and is executed at compile time.
+
+        writeError("Can not set signal handler\n");
+        _Exit(1);
+    }
+
+    volatile InstructionFail fail = InstructionFail::NONE;
+
+    if (sigsetjmp(jmpbuf, 1))
+    {
+        writeError("Instruction check fail. The CPU does not support ");
+        if (!std::apply(writeRetry, instructionFailToString(fail)))
+            _Exit(1);
+        writeError(" instruction set.\n");
+        _Exit(1);
+    }
+
+    checkRequiredInstructionsImpl(fail);
+
+    if (sigaction(signal, &sa_old, nullptr))
+    {
+        writeError("Can not set signal handler\n");
+        _Exit(1);
+    }
+}
+
+struct Checker
+{
+    Checker()
+    {
+        checkRequiredInstructions();
+    }
+} checker
+#ifndef OS_DARWIN
+    __attribute__((init_priority(101)))    /// Run before other static initializers.
+#endif
+;
+
+}
+
+
+#if !defined(USE_MUSL)
+/// NOTE: We will migrate to full static linking or our own dynamic loader to make this code obsolete.
+void checkHarmfulEnvironmentVariables(char ** argv)
+{
+    std::initializer_list<const char *> harmful_env_variables = {
+        /// The list is a selection from "man ld-linux".
+        "LD_PRELOAD",
+        "LD_LIBRARY_PATH",
+        "LD_ORIGIN_PATH",
+        "LD_AUDIT",
+        "LD_DYNAMIC_WEAK",
+        /// The list is a selection from "man dyld" (osx).
+        "DYLD_LIBRARY_PATH",
+        "DYLD_FALLBACK_LIBRARY_PATH",
+        "DYLD_VERSIONED_LIBRARY_PATH",
+        "DYLD_INSERT_LIBRARIES",
+    };
+
+    bool require_reexec = false;
+    for (const auto * var : harmful_env_variables)
+    {
+        if (const char * value = getenv(var); value && value[0]) // NOLINT(concurrency-mt-unsafe)
+        {
+            /// NOTE: setenv() is used over unsetenv() since unsetenv() marked as harmful
+            if (setenv(var, "", true)) // NOLINT(concurrency-mt-unsafe) // this is safe if not called concurrently
+            {
+                fmt::print(stderr, "Cannot override {} environment variable", var);
+                _exit(1);
+            }
+            require_reexec = true;
+        }
+    }
+
+    if (require_reexec)
+    {
+        /// Use execvp() over execv() to search in PATH.
+        ///
+        /// This should be safe, since:
+        /// - if argv[0] is relative path - it is OK
+        /// - if argv[0] has only basename, the it will search in PATH, like shell will do.
+        ///
+        /// Also note, that this (search in PATH) because there is no easy and
+        /// portable way to get absolute path of argv[0].
+        /// - on linux there is /proc/self/exec and AT_EXECFN
+        /// - but on other OSes there is no such thing (especially on OSX).
+        ///
+        /// And since static linking will be done someday anyway,
+        /// let's not pollute the code base with special cases.
+        int error = execvp(argv[0], argv);
+        _exit(error);
+    }
+}
+#endif

src/Common/EnvironmentChecks.h

@@ -0,0 +1,5 @@
+#pragma once
+
+#if !defined(USE_MUSL)
+void checkHarmfulEnvironmentVariables(char ** argv);
+#endif