Bug#1143816: faiss: autopkgtest needs update for new version of gcc-defaults: test SEGFAULTs

Aurelien Jarno <[email protected]> Fri, 7 Aug 2026 22:22:30 +0200
Newsgroups	gmane.linux.debian.devel.bugs.rc
Message-ID	<[email protected]>
Hi,

On 2026-08-06 17:28, Paul Gevers wrote:
> Source: faiss
> Version: 1.13.2-1
> Severity: serious
> X-Debbugs-CC: [email protected] [email protected]
> Tags: sid forky
> User: [email protected]
> Usertags: needs-update
> Control: affects -1 src:gcc-defaults
> User: [email protected]
> Usertags: riscv64
> 
> Dear maintainer(s),
> 
> With a recent upload of gcc-defaults the autopkgtest of faiss fails in
> testing when that autopkgtest is run with the binary packages of
> gcc-defaults from unstable. It passes when run with only packages from
> testing. In tabular form:
> 
>                        pass            fail
> gcc-defaults           from testing    1.233
> faiss                  from testing    1.13.2-1
> all others             from testing    from testing
> 
> I copied some of the output at the bottom of this report.
> 
> Currently this regression is blocking the migration of gcc-defaults to
> testing [1]. Of course, gcc-defaults shouldn't just break your autopkgtest
> (or even worse, your package), but it seems to me that the change in
> gcc-defaults was intended and your package needs to update to the new
> situation.
> 
> If this is a real problem in your package (and not only in your
> autopkgtest), the right binary package(s) from gcc-defaults should really
> add a versioned Breaks on the unfixed version of (one of your) package(s).
> Note: the Breaks is nice even if the issue is only in the autopkgtest as it
> helps the migration software to figure out the right versions to combine in
> the tests.
> 
> More information about this bug and the reason for filing it can be found on
> https://wiki.debian.org/ContinuousIntegration/RegressionEmailInformation
> 
> Paul
> 
> [1] https://qa.debian.org/excuses.php?package=gcc-defaults
> 
> https://ci.debian.net/data/autopkgtest/testing/riscv64/f/faiss/74133117/log.gz
> 
> 7572s test 119
> 7572s         Start 119: HNSWTest.TEST_search_from_candidate_unbounded
> 7572s 7572s 119: Test command:
> /tmp/autopkgtest.Eu4j7u/autopkgtest_tmp/tests/build/faiss_test
> "--gtest_filter=HNSWTest.TEST_search_from_candidate_unbounded"
> "--gtest_also_run_disabled_tests"
> 7572s 119: Working Directory:
> /tmp/autopkgtest.Eu4j7u/autopkgtest_tmp/tests/build
> 7572s 119: Test timeout computed to be: 10000000
> 7572s 119: WARNING clustering 1000 points to 40 centroids: please provide at
> least 1560 training points
> 7572s 119: Running main() from ./googletest/src/gtest_main.cc
> 7572s 119: Note: Google Test filter =
> HNSWTest.TEST_search_from_candidate_unbounded
> 7572s 119: [==========] Running 1 test from 1 test suite.
> 7572s 119: [----------] Global test environment set-up.
> 7572s 119: [----------] 1 test from HNSWTest
> 7572s 119: [ RUN      ] HNSWTest.TEST_search_from_candidate_unbounded
> 7572s 119/133 Test #119: HNSWTest.TEST_search_from_candidate_unbounded
> .....................***Exception: SegFault  0.14 sec
> 
> 
> 7577s 7577s 99% tests passed, 1 tests failed out of 133
> 7577s 7577s Total Test time (real) = 444.00 sec
> 7577s 7577s The following tests FAILED:
> 7577s 	119 - HNSWTest.TEST_search_from_candidate_unbounded (SEGFAULT)
> 7577s Errors while running CTest

While the test seemed to fail reliably on the debci host, I wasn't able 
to reproduce the problem. After giving it back, it succeeded [1] with 
gcc-16, so it's not a regression, but rather an unreliable test.

I also noticed that the autopkgtest time varies quite a lot (up to a 
factor 3 for a given debci runner), so I guess the high level of 
concurrent autopkgtests job following the gcc-default upload explains 
that.

I have therefore tried to reproduce the issue under high load, and I 
have been successful. Here is the corresponding backtrace:

| GNU gdb (Debian 17.2-1+b1) 17.2
| Copyright (C) 2025 Free Software Foundation, Inc.
| License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
| This is free software: you are free to change and redistribute it.
| There is NO WARRANTY, to the extent permitted by law.
| Type "show copying" and "show warranty" for details.
| This GDB was configured as "riscv64-linux-gnu".
| Type "show configuration" for configuration details.
| For bug reporting instructions, please see:
| <https://www.gnu.org/software/gdb/bugs/>.
| Find the GDB manual and other documentation resources online at:
|     <http://www.gnu.org/software/gdb/documentation/>.
| 
| For help, type "help".
| Type "apropos word" to search for commands related to "word"...
| Reading symbols from /home/aurel32/faiss/autopkgtest_tmp/tests/build/faiss_test...
| (No debugging symbols found in /home/aurel32/faiss/autopkgtest_tmp/tests/build/faiss_test)
| [New LWP 16849]
| [New LWP 16848]
| [New LWP 16850]
| [New LWP 16851]
| [Thread debugging using libthread_db enabled]
| Using host libthread_db library "/usr/lib/riscv64-linux-gnu/libthread_db.so.1".
| Core was generated by `/home/aurel32/faiss/autopkgtest_tmp/tests/build/faiss_test --gtest_filter=HNSWT'.
| Program terminated with signal SIGSEGV, Segmentation fault.
| #0  0x0000002ad7aeb6c2 in faiss::fvec_L2sqr_batch_4(float const*, float const*, float const*, float const*, float const*, unsigned long, float&, float&, float&, float&) ()
| [Current thread is 1 (Thread 0x3faf4420e0 (LWP 16849))]
| (gdb) bt
| #0  0x0000002ad7aeb6c2 in faiss::fvec_L2sqr_batch_4(float const*, float const*, float const*, float const*, float const*, unsigned long, float&, float&, float&, float&) ()
| #1  0x0000002ad7823f0c in faiss::(anonymous namespace)::FlatL2Dis::distances_batch_4(long, long, long, long, float&, float&, float&, float&) ()
| #2  0x0000002ad79014f6 in faiss::search_neighbors_to_add(faiss::HNSW&, faiss::DistanceComputer&, std::priority_queue<faiss::HNSW::NodeDistCloser, std::vector<faiss::HNSW::NodeDistCloser, std::allocator<faiss::HNSW::NodeDistCloser> >, std::
| less<faiss::HNSW::NodeDistCloser> >&, int, float, int, faiss::VisitedTable&, bool) ()
| #3  0x0000002ad7902f94 in faiss::HNSW::add_links_starting_from(faiss::DistanceComputer&, int, int, float, int, omp_lock_t*, faiss::VisitedTable&, bool) ()
| #4  0x0000002ad79033d2 in faiss::HNSW::add_with_locks(faiss::DistanceComputer&, int, int, std::vector<omp_lock_t, std::allocator<omp_lock_t> >&, faiss::VisitedTable&, bool) ()
| #5  0x0000002ad78498b0 in faiss::(anonymous namespace)::hnsw_add_vertices(faiss::IndexHNSW&, unsigned long, unsigned long, float const*, bool, bool) [clone ._omp_fn.0] ()
| #6  0x0000003fafeeff3e in ?? () from /usr/lib/riscv64-linux-gnu/libgomp.so.1
| #7  0x0000003faf64e330 in ?? () from /usr/lib/riscv64-linux-gnu/libc.so.6
| #8  0x0000003faf69b670 in ?? () from /usr/lib/riscv64-linux-gnu/libc.so.6
| (gdb) disassemble
| Dump of assembler code for function _ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_:
|    0x0000002ad7aeb59e <+0>:     ld      t2,0(sp)
|    0x0000002ad7aeb5a0 <+2>:     beqz    a5,0x2ad7aeb7e0 <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+578>
|    0x0000002ad7aeb5a4 <+6>:     slli    a5,a5,0x2
|    0x0000002ad7aeb5a6 <+8>:     addi    t3,a5,-4
|    0x0000002ad7aeb5aa <+12>:    fmv.w.x fa5,zero
|    0x0000002ad7aeb5ae <+16>:    srli    t0,t3,0x2
|    0x0000002ad7aeb5b2 <+20>:    addi    t1,t0,1
|    0x0000002ad7aeb5b6 <+24>:    andi    t5,t1,3
|    0x0000002ad7aeb5ba <+28>:    fmv.s   ft5,fa5
|    0x0000002ad7aeb5be <+32>:    fmv.s   ft3,fa5
|    0x0000002ad7aeb5c2 <+36>:    fmv.s   fa4,fa5
|    0x0000002ad7aeb5c6 <+40>:    li      t1,0
|    0x0000002ad7aeb5c8 <+42>:    beqz    t5,0x2ad7aeb6a6 <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+264>
|    0x0000002ad7aeb5cc <+46>:    li      t4,1
|    0x0000002ad7aeb5ce <+48>:    beq     t5,t4,0x2ad7aeb658 <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+186>
|    0x0000002ad7aeb5d2 <+52>:    li      t6,2
|    0x0000002ad7aeb5d4 <+54>:    beq     t5,t6,0x2ad7aeb60e <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+112>
|    0x0000002ad7aeb5d8 <+58>:    flw     fa3,0(a0)
|    0x0000002ad7aeb5dc <+62>:    flw     fa4,0(a1)
|    0x0000002ad7aeb5e0 <+66>:    flw     ft1,0(a2)
|    0x0000002ad7aeb5e4 <+70>:    flw     ft0,0(a3)
|    0x0000002ad7aeb5e8 <+74>:    flw     fa2,0(a4)
|    0x0000002ad7aeb5ec <+78>:    fsub.s  ft2,fa3,fa4
|    0x0000002ad7aeb5f0 <+82>:    fsub.s  ft3,fa3,ft1
|    0x0000002ad7aeb5f4 <+86>:    fsub.s  ft4,fa3,ft0
|    0x0000002ad7aeb5f8 <+90>:    fsub.s  ft6,fa3,fa2
|    0x0000002ad7aeb5fc <+94>:    li      t1,4
|    0x0000002ad7aeb5fe <+96>:    fmadd.s fa4,ft2,ft2,fa5
|    0x0000002ad7aeb602 <+100>:   fmadd.s ft3,ft3,ft3,fa5
|    0x0000002ad7aeb606 <+104>:   fmadd.s ft5,ft4,ft4,fa5
|    0x0000002ad7aeb60a <+108>:   fmadd.s fa5,ft6,ft6,fa5
|    0x0000002ad7aeb60e <+112>:   add     t0,a0,t1
|    0x0000002ad7aeb612 <+116>:   add     t3,a1,t1
|    0x0000002ad7aeb616 <+120>:   add     t5,a2,t1
|    0x0000002ad7aeb61a <+124>:   add     t4,a3,t1
|    0x0000002ad7aeb61e <+128>:   add     t6,a4,t1
|    0x0000002ad7aeb622 <+132>:   flw     ft7,0(t0)
|    0x0000002ad7aeb626 <+136>:   flw     fa0,0(t3)
|    0x0000002ad7aeb62a <+140>:   flw     fa1,0(t5)
|    0x0000002ad7aeb62e <+144>:   flw     fa6,0(t4)
|    0x0000002ad7aeb632 <+148>:   flw     fa7,0(t6)
|    0x0000002ad7aeb636 <+152>:   fsub.s  ft8,ft7,fa0
|    0x0000002ad7aeb63a <+156>:   fsub.s  ft9,ft7,fa1
|    0x0000002ad7aeb63e <+160>:   fsub.s  ft10,ft7,fa6
|    0x0000002ad7aeb642 <+164>:   fsub.s  ft11,ft7,fa7
|    0x0000002ad7aeb646 <+168>:   addi    t1,t1,4
|    0x0000002ad7aeb648 <+170>:   fmadd.s fa4,ft8,ft8,fa4
|    0x0000002ad7aeb64c <+174>:   fmadd.s ft3,ft9,ft9,ft3
|    0x0000002ad7aeb650 <+178>:   fmadd.s ft5,ft10,ft10,ft5
|    0x0000002ad7aeb654 <+182>:   fmadd.s fa5,ft11,ft11,fa5
|    0x0000002ad7aeb658 <+186>:   add     t0,a0,t1
|    0x0000002ad7aeb65c <+190>:   add     t3,a1,t1
|    0x0000002ad7aeb660 <+194>:   add     t5,a2,t1
|    0x0000002ad7aeb664 <+198>:   add     t4,a3,t1
|    0x0000002ad7aeb668 <+202>:   add     t6,a4,t1
|    0x0000002ad7aeb66c <+206>:   flw     fa3,0(t0)
|    0x0000002ad7aeb670 <+210>:   flw     ft1,0(t3)
|    0x0000002ad7aeb674 <+214>:   flw     ft2,0(t5)
|    0x0000002ad7aeb678 <+218>:   flw     fa2,0(t4)
|    0x0000002ad7aeb67c <+222>:   flw     ft0,0(t6)
|    0x0000002ad7aeb680 <+226>:   fsub.s  ft4,fa3,ft1
|    0x0000002ad7aeb684 <+230>:   fsub.s  ft6,fa3,ft2
|    0x0000002ad7aeb688 <+234>:   fsub.s  ft7,fa3,fa2
|    0x0000002ad7aeb68c <+238>:   fsub.s  fa0,fa3,ft0
|    0x0000002ad7aeb690 <+242>:   addi    t1,t1,4
|    0x0000002ad7aeb692 <+244>:   fmadd.s fa4,ft4,ft4,fa4
|    0x0000002ad7aeb696 <+248>:   fmadd.s ft3,ft6,ft6,ft3
|    0x0000002ad7aeb69a <+252>:   fmadd.s ft5,ft7,ft7,ft5
|    0x0000002ad7aeb69e <+256>:   fmadd.s fa5,fa0,fa0,fa5
|    0x0000002ad7aeb6a2 <+260>:   beq     a5,t1,0x2ad7aeb7cc <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+558>
|    0x0000002ad7aeb6a6 <+264>:   add     t0,a0,t1
|    0x0000002ad7aeb6aa <+268>:   add     t3,a1,t1
|    0x0000002ad7aeb6ae <+272>:   flw     fa6,0(t0)
|    0x0000002ad7aeb6b2 <+276>:   flw     fa7,0(t3)
|    0x0000002ad7aeb6b6 <+280>:   add     t5,a2,t1
|    0x0000002ad7aeb6ba <+284>:   add     t6,a4,t1
|    0x0000002ad7aeb6be <+288>:   flw     ft8,0(t5)
| => 0x0000002ad7aeb6c2 <+292>:   flw     ft9,0(t6)
|    0x0000002ad7aeb6c6 <+296>:   add     t4,a3,t1
|    0x0000002ad7aeb6ca <+300>:   fsub.s  ft10,fa6,fa7
|    0x0000002ad7aeb6ce <+304>:   flw     fa1,0(t4)
|    0x0000002ad7aeb6d2 <+308>:   add     t3,a0,t1
|    0x0000002ad7aeb6d6 <+312>:   add     t5,a1,t1
|    0x0000002ad7aeb6da <+316>:   flw     fa0,4(t3)
|    0x0000002ad7aeb6de <+320>:   fsub.s  ft11,fa6,ft8
|    0x0000002ad7aeb6e2 <+324>:   fsub.s  fa3,fa6,ft9
|    0x0000002ad7aeb6e6 <+328>:   flw     fa2,4(t5)
|    0x0000002ad7aeb6ea <+332>:   fsub.s  ft7,fa6,fa1
|    0x0000002ad7aeb6ee <+336>:   add     t6,a2,t1
|    0x0000002ad7aeb6f2 <+340>:   add     t4,a3,t1
|    0x0000002ad7aeb6f6 <+344>:   add     t0,a4,t1
|    0x0000002ad7aeb6fa <+348>:   fmadd.s ft2,ft10,ft10,fa4
|    0x0000002ad7aeb6fe <+352>:   flw     ft1,4(t6)
|    0x0000002ad7aeb702 <+356>:   fsub.s  fa4,fa0,fa2
|    0x0000002ad7aeb706 <+360>:   flw     ft0,4(t4)
|    0x0000002ad7aeb70a <+364>:   flw     ft6,4(t0)
|    0x0000002ad7aeb70e <+368>:   add     t5,a0,t1
|    0x0000002ad7aeb712 <+372>:   add     t6,a1,t1
|    0x0000002ad7aeb716 <+376>:   fmadd.s ft4,ft11,ft11,ft3
|    0x0000002ad7aeb71a <+380>:   fmadd.s ft10,fa3,fa3,fa5
|    0x0000002ad7aeb71e <+384>:   flw     ft11,8(t6)
|    0x0000002ad7aeb722 <+388>:   flw     fa5,8(t5)
|    0x0000002ad7aeb726 <+392>:   fmadd.s ft3,ft7,ft7,ft5
|    0x0000002ad7aeb72a <+396>:   fsub.s  ft9,fa0,ft6
|    0x0000002ad7aeb72e <+400>:   fsub.s  fa6,fa0,ft1
|    0x0000002ad7aeb732 <+404>:   fsub.s  fa7,fa0,ft0
|    0x0000002ad7aeb736 <+408>:   add     t0,a2,t1
|    0x0000002ad7aeb73a <+412>:   add     t4,a3,t1
|    0x0000002ad7aeb73e <+416>:   add     t3,a4,t1
|    0x0000002ad7aeb742 <+420>:   fmadd.s ft8,fa4,fa4,ft2
|    0x0000002ad7aeb746 <+424>:   fsub.s  fa3,fa5,ft11
|    0x0000002ad7aeb74a <+428>:   flw     fa1,8(t0)
|    0x0000002ad7aeb74e <+432>:   flw     ft7,8(t4)
|    0x0000002ad7aeb752 <+436>:   flw     ft5,8(t3)
|    0x0000002ad7aeb756 <+440>:   fmadd.s fa4,ft9,ft9,ft10
|    0x0000002ad7aeb75a <+444>:   fmadd.s ft1,fa6,fa6,ft4
|    0x0000002ad7aeb75e <+448>:   fsub.s  ft6,fa5,fa1
|    0x0000002ad7aeb762 <+452>:   fmadd.s ft0,fa7,fa7,ft3
|    0x0000002ad7aeb766 <+456>:   fsub.s  fa2,fa5,ft7
|    0x0000002ad7aeb76a <+460>:   fsub.s  fa6,fa5,ft5
|    0x0000002ad7aeb76e <+464>:   add     t0,a0,t1
|    0x0000002ad7aeb772 <+468>:   add     t6,a1,t1
|    0x0000002ad7aeb776 <+472>:   add     t5,a2,t1
|    0x0000002ad7aeb77a <+476>:   add     t4,a3,t1
|    0x0000002ad7aeb77e <+480>:   add     t3,a4,t1
|    0x0000002ad7aeb782 <+484>:   flw     fa7,12(t0)
|    0x0000002ad7aeb786 <+488>:   fmadd.s fa0,fa3,fa3,ft8
|    0x0000002ad7aeb78a <+492>:   flw     ft3,12(t3)
|    0x0000002ad7aeb78e <+496>:   flw     ft4,12(t6)
|    0x0000002ad7aeb792 <+500>:   flw     ft8,12(t5)
|    0x0000002ad7aeb796 <+504>:   flw     ft2,12(t4)
|    0x0000002ad7aeb79a <+508>:   fmadd.s fa5,fa6,fa6,fa4
|    0x0000002ad7aeb79e <+512>:   fsub.s  fa3,fa7,ft3
|    0x0000002ad7aeb7a2 <+516>:   fsub.s  ft9,fa7,ft4
|    0x0000002ad7aeb7a6 <+520>:   fmadd.s ft10,ft6,ft6,ft1
|    0x0000002ad7aeb7aa <+524>:   fsub.s  ft11,fa7,ft8
|    0x0000002ad7aeb7ae <+528>:   fmadd.s fa1,fa2,fa2,ft0
|    0x0000002ad7aeb7b2 <+532>:   fsub.s  ft7,fa7,ft2
|    0x0000002ad7aeb7b6 <+536>:   addi    t1,t1,16
|    0x0000002ad7aeb7b8 <+538>:   fmadd.s fa4,ft9,ft9,fa0
|    0x0000002ad7aeb7bc <+542>:   fmadd.s ft3,ft11,ft11,ft10
|    0x0000002ad7aeb7c0 <+546>:   fmadd.s fa5,fa3,fa3,fa5
|    0x0000002ad7aeb7c4 <+550>:   fmadd.s ft5,ft7,ft7,fa1
|    0x0000002ad7aeb7c8 <+554>:   bne     a5,t1,0x2ad7aeb6a6 <_ZN5faiss18fvec_L2sqr_batch_4EPKfS1_S1_S1_S1_mRfS2_S2_S2_+264>
|    0x0000002ad7aeb7cc <+558>:   fsw     fa4,0(a6)
|    0x0000002ad7aeb7d0 <+562>:   ld      a0,8(sp)
|    0x0000002ad7aeb7d2 <+564>:   fsw     ft3,0(a7)
|    0x0000002ad7aeb7d6 <+568>:   fsw     ft5,0(t2)
|    0x0000002ad7aeb7da <+572>:   fsw     fa5,0(a0)
|    0x0000002ad7aeb7de <+576>:   ret
|    0x0000002ad7aeb7e0 <+578>:   fmv.w.x fa5,zero
|    0x0000002ad7aeb7e4 <+582>:   ld      a0,8(sp)
|    0x0000002ad7aeb7e6 <+584>:   fmv.s   fa4,fa5
|    0x0000002ad7aeb7ea <+588>:   fmv.s   ft3,fa5
|    0x0000002ad7aeb7ee <+592>:   fmv.s   ft5,fa5
|    0x0000002ad7aeb7f2 <+596>:   fsw     fa4,0(a6)
|    0x0000002ad7aeb7f6 <+600>:   fsw     ft3,0(a7)
|    0x0000002ad7aeb7fa <+604>:   fsw     ft5,0(t2)
|    0x0000002ad7aeb7fe <+608>:   fsw     fa5,0(a0)
|    0x0000002ad7aeb802 <+612>:   ret

At this stage, I haven't debugged it further. I'll try to see if I can 
reproduce the issue on another architecture. Given it's not a 
regression, I guess the severity can be decreased.

Regards
Aurelien

[1] https://ci.debian.net/packages/f/faiss/testing/riscv64/74151428/

-- 
Aurelien Jarno                          GPG: 4096R/1DDD8C9B
[email protected]                     http://aurel32.net
signature.asc (application/pgp-signature, 833 B)