Some debuginfod fixlets

Some debuginfod fixlets

[Mark Wielaard]

Hi,

While testing Frank's new spec/rpms for the run-debuginfod-find.sh
testcase I found a couple of issues that I pushed to the debuginfod-
submit branch.

  Add tests/debuginfod-rpms to EXTRA_DISTS.
  Fix two small memory leaks in debuginfod-find and testcase.
  Increase timeout for run-debuginfod-find.sh
  run-debuginfod-find.sh: Use abs_srcdir when copying debuginfod-rpms
  debuginfod: Accept empty comp_dir when cuname is absolute.

With these fixes everything passes make distcheck locally.

The only interesting one is debuginfod: Accept empty comp_dir when
cuname is absolute. This can happen with gcc 4.8.5. In that case we
have a CU DIE with an absolute DW_AT_name, but no DW_AT_comp_dir. And
the files in the debug_line table never reference the comp_dir (dir
entry zero). For example:

 [     b]  compile_unit         abbrev: 1
           producer             (strp) "GNU C 4.8.5 20150623 (Red Hat 4.8.5-39) -mtune=generic -march=x86-64 -g"
           language             (data1) C89 (1)
           name                 (strp) "/home/mark/build/elfutils-obj/prog.c"
           low_pc               (addr) 0x00000000004004ed
           high_pc              (data8) 11 (0x00000000004004f8)
           stmt_list            (sec_offset) 0

Directory table:
 /home/mark/build/elfutils-obj

File name table:
 Entry Dir   Time      Size      Name
 1     1     0         0         prog.c

Hopefully these changes are correct. If not, please let me know, and
apologies for pushing them on your branch.

Cheers,

Mark

Re: Some debuginfod fixlets

[Mark Wielaard]

Hi,

On Sat, 2019-11-16 at 17:42 +0100, Mark Wielaard wrote:
> While testing Frank's new spec/rpms for the run-debuginfod-find.sh
> testcase I found a couple of issues that I pushed to the debuginfod-
> submit branch.
> 
>   Add tests/debuginfod-rpms to EXTRA_DISTS.
>   Fix two small memory leaks in debuginfod-find and testcase.
>   Increase timeout for run-debuginfod-find.sh
>   run-debuginfod-find.sh: Use abs_srcdir when copying debuginfod-rpms
>   debuginfod: Accept empty comp_dir when cuname is absolute.
> 
> With these fixes everything passes make distcheck locally.

There were 3 more patches needed:

 run-debuginfod-find.sh: Explicitly run with /bin/bash
 debuginfod-client: Always initialize suffix.
 run-debuginfod-find.sh: Extend DEBUGINFOD_TIMEOUT when ran under valgrind

All pushed to the debuginfod-submit branch. I did builds on all the
builbot workers (thanks to Dan Horák for preparing the s390x and
ppc64/ppc64le setups). And the new testcase now passes everywhere (some
combinations of debian, fedora and centos on armhf, aarch64, i686,
ppc64, ppc64le, s390x and x86_64).

The last patch extends the DEBUGINFOD_TIMEOUT to 5 minutes when running
under valgrind since valgrind really slows down the client. It seems
the poll/select call turns into a busy loop under valgrind. I am
tempted to disable valgrind when running the run-debuginfod-find.sh
testcase. But it has found a couple of issues, so it does seem useful
for now.

Cheers,

Mark