debcrafters-packages team mailing list archive

[Simon Chopin <2125424@xxxxxxxxxxxxxxxxxx>]

Alright, I've spent an inordinate amount of time trying to figure this
out.

First thing first, it's reproducible on a questing arm64 porterbox
running a 6.17 kernel, even though the failing build is on a system
running a Noble kernel.

Second, the reproduction is... unstable. Sometimes it's one test that
fails, sometimes it's 3. And even within one test (trace_clock.gen.test
is my goto), it's never the same binary that calls clock_gettime64,
which means that when you then go and execute the binary by hand, it
likely will *not* call call_gettime64.

Which brings me to my third point: I managed to successfully build strace on armhf against a freshly-compiled glibc that had the recent changes reverted, as supplied by slyon in another investigation: https://launchpad.net/~slyon/+archive/ubuntu/lp-2123828-qemu-riscv64-on-s390x for the glibc package, https://launchpad.net/~schopin/+archive/ubuntu/strace-glibc3/+build/31310077 for the successful build.
HOWEVER due to the point 2, it *could* be a fluke, so I have another build running right now.

Now, for the fun technical details:
clock_gettime64 on ARM64 is in the vDSO. What that normally means is that we should *not* be seeing calls to it on strace. I'm guessing there's an initial call when setting up the vDSO at process start, but only under some circumstances. And that's where I'm a bit puzzled: recent glibc changes don't seem to have anything to do with time management, or the vDSO.

To make matter worse, porterboxes are annoyingly isolated from the
internet at large. Tomorrow I'll attempt a bisection on glibc, and while
that runs I'll try to figure out what could be calling this syscall
intermittently...

-- 
You received this bug notification because you are a member of
Debcrafters packages, which is subscribed to glibc in Ubuntu.
https://bugs.launchpad.net/bugs/2125424

Title:
  strace FTBFS against glibc 2.42-0ubuntu3 on armhf

Status in glibc package in Ubuntu:
  New
Status in strace package in Ubuntu:
  New

Bug description:
  strace started to FTBFS on armhf against that version, but was
  building fine against 2.42-0ubuntu1.

  Excerpt from the failing builds (first failing test):

  --- /dev/null	2025-08-02 18:23:40.000000000 +0000
  +++ log	2025-09-22 15:40:20.824144876 +0000
  @@ -0,0 +1 @@
  +clock_gettime64(CLOCK_MONOTONIC, {tv_sec=398, tv_nsec=625531646}) = 0
  + fail_ ../../src/strace -qq -esignal=none -e/clock ../getgid output mismatch
  + warn_ clock.gen.test: failed test: ../../src/strace -qq -esignal=none -e/clock ../getgid output mismatch
  + printf %s\n clock.gen.test: failed test: ../../src/strace -qq -esignal=none -e/clock ../getgid output mismatch
  clock.gen.test: failed test: ../../src/strace -qq -esignal=none -e/clock ../getgid output mismatch
  + exit 1
  FAIL clock.gen.test (exit status: 1)

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/2125424/+subscriptions