Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.bugs.dist > #1111100

Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail)

Show all headers | View raw

Hello,

Peter Blackman, le mer. 27 juil. 2022 12:22:17 +0100, a ecrit:
> Back trace on dumatest shows an endless loop. Repeats every ten steps.

Could you manage to get the top of the stack?

> #35 0x080361d6 in DUMA_get_sem () at sem_inc.c:230
> #36 0x080345a1 in _duma_allocate (alignment=1, userSize=100, protectBelow=0,
>     fillByte=255, protectAllocList=1, allocator=EFA_MALLOC,
>     fail=DUMA_FAIL_ENV,
>     filename=0x8037060 <unknown_file> "UNKNOWN (use #include \"duma.h\")",
>     lineno=0) at duma.c:1386
> #37 0x080353b6 in _duma_malloc (size=100,
>     filename=0x8037060 <unknown_file> "UNKNOWN (use #include \"duma.h\")",
>     lineno=0) at duma.c:1970
> #38 0x08035c8b in malloc (size=100) at duma.c:2424
> #39 0x01103ce5 in ?? () from /lib/i386-gnu/libc.so.0.3
> #40 0x010e06a7 in asprintf () from /lib/i386-gnu/libc.so.0.3
> #41 0x010b736a in ?? () from /lib/i386-gnu/libc.so.0.3
> #42 0x010b7469 in __assert_fail () from /lib/i386-gnu/libc.so.0.3
> #43 0x010452cf in pthread_self () from /lib/i386-gnu/libpthread.so.0.3
> #44 0x08036095 in lock () at sem_inc.c:135
> #45 0x080361d6 in DUMA_get_sem () at sem_inc.c:230
> #46 0x080345a1 in _duma_allocate (alignment=1, userSize=100, protectBelow=0,
>     fillByte=255, protectAllocList=1, allocator=EFA_MALLOC,
>     fail=DUMA_FAIL_ENV,
>     filename=0x8037060 <unknown_file> "UNKNOWN (use #include \"duma.h\")",
>     lineno=0) at duma.c:1386
> #47 0x080353b6 in _duma_malloc (size=100,
>     filename=0x8037060 <unknown_file> "UNKNOWN (use #include \"duma.h\")",
>     lineno=0) at duma.c:1970
> #48 0x08035c8b in malloc (size=100) at duma.c:2424

So the loop is because the assertion failure uses printf, which uses an
allocation, which triggers the assertion again etc.

> I'm curious about the step   pthread_self ()  -->  __assert_fail ()
> 
> The man page for pthread_self (Linux)
> https://man7.org/linux/man-pages/man3/pthread_self.3.html
> 
> states that the function always succeeds. Unlikely if it throws an assert.

Yes, that's why it's an assert and not an error: something very wrong is
happening.

Do you have the libc0.3-dbg package installed? That'd help to know
exactly why assertion fails.

> Also raised upstream
> https://github.com/johnsonjh/duma/issues/172
> 
> but the question really is.
> 
> Is this a bug in Hurd, or something that can be fixed in duma?

Most probably what happens (the top of the stack would tell) is that
pthread_self is called very early, before libpthread is initialized, and
it's assert (__pthread_threads); which fails because of that.

Samuel

Back to linux.debian.bugs.dist | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter Blackman <peter@pblackman.plus.com> - 2022-07-27 13:30 +0200
  Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-27 14:00 +0200
    Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-27 19:20 +0200
      Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-27 19:50 +0200
      Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-27 23:40 +0200
        Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 01:10 +0200
          Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-28 17:10 +0200
            Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 17:10 +0200
              Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-28 17:30 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 18:00 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-28 19:10 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-28 19:30 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 19:40 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 20:30 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Peter B <peter@pblackman.plus.com> - 2022-07-28 21:20 +0200
                Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 22:10 +0200
  Bug#1016106: duma: FTBFS on hurd-i386 (builds OK, but most self tests fail) Samuel Thibault <sthibault@debian.org> - 2022-07-28 17:30 +0200

csiph-web