tarantool / tarantool/tarantool

SIGSEGV in memtx.perftest

Open
#10,281 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

bug
Dominant language
Lua
Stars
3.7k
Forks
419
Avg merge
1d 23h
Merged PRs (30d)
88

Description

Bug description

  • OS: Linux
  • OS Version: Ubuntu 22.04
  • Architecture: amd64

Tarantool 3.2.0-entrypoint-161-g030e64803e
Target: Linux-x86_64-RelWithDebInfo
Build options: cmake . -DCMAKE_INSTALL_PREFIX=/usr/local -DENABLE_BACKTRACE=TRUE
Compiler: GNU-11.4.0
C_FLAGS: -fexceptions -funwind-tables -fasynchronous-unwind-tables -fno-common -msse2 -Wformat -Wformat-security -Werror=format-security -fstack-protector-strong -fPIC -fmacro-prefix-map=/home/sergeyb/sources/MRG/tarantool=. -std=c11 -Wall -Wextra -Wno-gnu-alignof-expression -fno-gnu89-inline -Wno-cast-function-type -O2 -g -DNDEBUG -ggdb -O2
CXX_FLAGS: -fexceptions -funwind-tables -fasynchronous-unwind-tables -fno-common -msse2 -Wformat -Wformat-security -Werror=format-security -fstack-protector-strong -fPIC -fmacro-prefix-map=/home/sergeyb/sources/MRG/tarantool=. -std=c++11 -Wall -Wextra -Wno-invalid-offsetof -Wno-gnu-alignof-expression -Wno-cast-function-type -O2 -g -DNDEBUG -ggdb -O2

Steps to reproduce

$ ./build/perf/memtx.perftest 
2024-07-20T20:36:52+03:00
Running ./build/perf/memtx.perftest
Run on (8 X 1185.99 MHz CPU s)
CPU Caches:
  L1 Data 48 KiB (x4)
  L1 Instruction 32 KiB (x4)
  L2 Unified 1280 KiB (x4)
  L3 Unified 12288 KiB (x1)
Load Average: 0.88, 0.68, 0.45
mapping 1073741824 bytes for memtx tuple arena...

<snipped>
$ cd perf/lua/
$ ../../build/perf/memtx.perftest 
2024-07-20T20:37:49+03:00
Running ../../build/perf/memtx.perftest
Run on (8 X 854.655 MHz CPU s)
CPU Caches:
  L1 Data 48 KiB (x4)
  L1 Instruction 32 KiB (x4)
  L2 Unified 1280 KiB (x4)
  L3 Unified 12288 KiB (x1)
Load Average: 1.03, 0.77, 0.50
Segmentation fault (core dumped)
$ rm 00000000000000001003.snap 
$ ../../build/perf/memtx.perftest 
2024-07-20T20:38:50+03:00
Running ../../build/perf/memtx.perftest
Run on (8 X 1644.97 MHz CPU s)
CPU Caches:
  L1 Data 48 KiB (x4)
  L1 Instruction 32 KiB (x4)
  L2 Unified 1280 KiB (x4)
  L3 Unified 12288 KiB (x1)
Load Average: 1.09, 0.84, 0.54
mapping 1073741824 bytes for memtx tuple arena...
Actual slab_alloc_factor calculated on the basis of desired slab_alloc_factor = 1.090508
fiber has not yielded for more than 0.500 seconds
-----------------------------------------------------------------------------------------------------
Benchmark                                           Time             CPU   Iterations UserCounters...
-----------------------------------------------------------------------------------------------------
MemtxFixture/TreeGetRandomExistingKeys            802 ns          802 ns       909464 items_per_second=1.24701M/s
MemtxFixture/TreeGet1RandomExistingKey            404 ns          404 ns      1641841 items_per_second=2.4759M/s
MemtxFixture/TreeSelectAll                    6439206 ns      6436792 ns          107 items_per_second=20.3629M/s
MemtxFixture/TreeReplaceRandomExistingKeys       1260 ns         1257 ns       576237 items_per_second=795.323k/s

Actual behavior

I see a segmentation fault in memtx.perftest when it is executed in a directory with existed tarantool files.

(gdb) run
Starting program: /home/sergeyb/sources/MRG/tarantool/build/perf/memtx.perftest 
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
2024-07-20T17:43:35+03:00
Running /home/sergeyb/sources/MRG/tarantool/build/perf/memtx.perftest
Run on (8 X 721.996 MHz CPU s)
CPU Caches:
  L1 Data 48 KiB (x4)
  L1 Instruction 32 KiB (x4)
  L2 Unified 1280 KiB (x4)
  L3 Unified 12288 KiB (x1)
Load Average: 0.39, 0.86, 0.97

Program received signal SIGSEGV, Segmentation fault.
rlist_add_tail (item=0x555555c56760, head=0x555555ba2450 <gc+304>) at /home/sergeyb/sources/MRG/tarantool/src/lib/small/include/small/rlist.h:87
87              item->prev->next = item;
(gdb) 
(gdb) bt
#0  rlist_add_tail (item=0x555555c56760, head=0x555555ba2450 <gc+304>)
    at /home/sergeyb/sources/MRG/tarantool/src/lib/small/include/small/rlist.h:87
#1  gc_add_checkpoint (vclock=0x555555c75e90) at /home/sergeyb/sources/MRG/tarantool/src/box/gc.c:485
#2  0x000055555560eeb8 in memtx_engine_new (snap_dirname=<optimized out>, force_recovery=<optimized out>, tuple_arena_max_size=1073741824, 
    objsize_min=16, dontdump=<optimized out>, granularity=8, allocator=0x555555ad6328 "small", alloc_factor=1.10000002, sort_threads=0, 
    on_indexes_built=0x5555555e4b10 <Memtx::memtx_on_indexes_built_mock_cb()>) at /home/sergeyb/sources/MRG/tarantool/src/box/memtx_engine.cc:1694
#3  0x00005555555e51b4 in Memtx::Memtx (this=0x555555b99c20 <Memtx::instance()::singleton>)
    at /home/sergeyb/sources/MRG/tarantool/perf/memtx.cc:124
#4  0x00005555555e60df in Memtx::instance () at /home/sergeyb/sources/MRG/tarantool/perf/memtx.cc:85
#5  MemtxFixture::SetUp (this=0x555555c3d550) at /home/sergeyb/sources/MRG/tarantool/perf/memtx.cc:221
#6  0x00005555555e4c54 in benchmark::Fixture::SetUp (st=..., this=0x555555c3d550) at /usr/include/benchmark/benchmark.h:1225
#7  benchmark::Fixture::Run (this=0x555555c3d550, st=...) at /usr/include/benchmark/benchmark.h:1216
#8  0x00007ffff7f60a28 in benchmark::internal::BenchmarkInstance::Run(unsigned long, int, benchmark::internal::ThreadTimer*, benchmark::internal::ThreadManager*, benchmark::internal::PerfCountersMeasurement*) const () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#9  0x00007ffff7f7a8e1 in ?? () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#10 0x00007ffff7f7b8a5 in benchmark::internal::BenchmarkRunner::DoNIterations() () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#11 0x00007ffff7f7fa6c in benchmark::internal::BenchmarkRunner::DoOneRepetition() () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#12 0x00007ffff7f6949a in ?? () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#13 0x00007ffff7f6acee in benchmark::RunSpecifiedBenchmarks(benchmark::BenchmarkReporter*, benchmark::BenchmarkReporter*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >) () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#14 0x00007ffff7f6aecc in benchmark::RunSpecifiedBenchmarks() () from /lib/x86_64-linux-gnu/libbenchmark.so.1
#15 0x00005555555e3408 in main (argc=<optimized out>, argv=0x7fffffffdd28) at /home/sergeyb/sources/MRG/tarantool/perf/memtx.cc:390
(gdb) 

Expected behavior

no sigsegv

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Reproduce the crash by running build/perf/memtx.perftest from a directory containing existing Tarantool files, then inspect perf/memtx.cc and the stack through src/box/memtx_engine.cc, src/box/gc.c, and src/lib/small/include/small/rlist.h. Done means the benchmark no longer raises SIGSEGV in that situation and still runs when the directory is clean.

Written by the indexing model from the issue text.

Assessment

Tech stack
c, cpp
Domain
databases, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.