Annoy indexes fell out of favor in the community, at least when it comes
to vector databases. Such indexes work okay-ish low dimensions but they
suffers badly from a curse of dimensionality which makes them inapt for
a high number of dimensions.
Now that Annoy is gone, issue (*) also disappears and we can drop
'no-ubsan', 'no-cpu-aarch64', and 'no-asan' from tests.
(*) spotify/annoy#456
Registers usearch and annoy properly via configure_config.cmake and
config.h.in like all other 3rd party libs, instead of (mis)using
target_compile_definitions.
No directory 'SimSIMD-map' exists, the build only worked because SimSIMD
support in usearch was (accidentally?) disabled. This commit corrects
the build description. SimSIMD support in usearch will be enabled by a
later commit.
The problem here is that ignorelist did not work by some reason, if I
will look at the ignored functions it should not contain any TSan
interseption code, while it does:
$ lldb-13 clickhouse
(lldb) target create "clickhouse"
disas -n rd_avg_rollover
Current executable set to '/home/azat/ch/tmp/tsan-test/clickhouse' (x86_64).
(lldb) disas -n rd_avg_rollover
clickhouse`rd_kafka_stats_emit_avg:
clickhouse[0x1cbf84a7] <+39>: leaq 0x30(%r15), %r12
clickhouse[0x1cbf84ab] <+43>: movq %r12, %rdi
clickhouse[0x1cbf84ae] <+46>: callq 0x1ccdad40 ; rdk_thread_mutex_lock at tinycthread.c:111
clickhouse[0x1cbf84b3] <+51>: leaq 0x58(%r15), %rdi
clickhouse[0x1cbf84b7] <+55>: callq 0x71b5390 ; __tsan_read4
clickhouse[0x1cbf84bc] <+60>: cmpl $0x0, 0x58(%r15)
clickhouse[0x1cbf84c1] <+65>: je 0x1cbf8595 ; <+277> [inlined] rd_avg_rollover + 238 at rdavg.h
clickhouse[0x1cbf84c7] <+71>: leaq -0xc8(%rbp), %rdi
clickhouse[0x1cbf84ce] <+78>: xorl %esi, %esi
clickhouse[0x1cbf84d0] <+80>: callq 0x1ccdac80 ; rdk_thread_mutex_init at tinycthread.c:62
clickhouse[0x1cbf84d5] <+85>: leaq 0x5c(%r15), %rdi
clickhouse[0x1cbf84d9] <+89>: callq 0x71b5390 ; __tsan_read4
(lldb) disas -n rd_avg_calc
clickhouse`rd_kafka_broker_ops_io_serve:
clickhouse[0x1cbdf086] <+1990>: leaq 0x5a4(%rbx), %rdi
clickhouse[0x1cbdf08d] <+1997>: callq 0x71b5390 ; __tsan_read4
clickhouse[0x1cbdf092] <+2002>: cmpl $0x0, 0x5a4(%rbx)
clickhouse[0x1cbdf099] <+2009>: je 0x1cbdf12b ; <+2155> [inlined] rd_kafka_broker_timeout_scan + 719 at rdkafka_broker.c
I guess the reason is that they had been inlined
So now rd_avg_calc() guarded with a mutex.
Refs: https://github.com/ClickHouse/librdkafka/pull/11
Fixes: https://github.com/ClickHouse/ClickHouse/issues/60939
Signed-off-by: Azat Khuzhin <a.khuzhin@semrush.com>
