[Bug 1967518] [NEW] ltrace - build on ppc64el fails

[Olivier Gayot]

Public bug reported:

The build on ppc64el fails with the following error:

plt.c: In function ‘arch_elf_add_plt_entry’:
plt.c:781:17: error: ‘libsym’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
  781 |                 free(libsym);
      |                 ^~~~~~~~~~~~
cc1: all warnings being treated as errors


This seems to be happening because our back-port of this upstream patch is wrong:

commit 7f20cb21a0d021ab3b12d8a5187beb728f339b0f
Author:     Thierry Fauck <thierry at linux.vnet.ibm.com>
AuthorDate: Tue May 13 07:48:24 2014 -0400
Commit:     Petr Machata <pmachata at gmail.com>
CommitDate: Wed Nov 11 19:08:43 2015 +0100

    Support for powerpc64 arch ppc64el

    Signed-off-by: Thierry Fauck <thierry at linux.vnet.ibm.com>

            Add support for ppc64le proc and ELF ABIv2.
            Provides support for irelative and wchar


The original patch leans on a jump to a "fail" label (that is declared near the beginning of the function and does the following):

  free(name);
  return PLT_FAIL;

In the 0.7.3 code, however, a different definition of the "fail" label
exists. It is declared later in the function and relies on the libsym
variable being initialized:

  free(name);
  free(libsym);
  return plt_fail;

** Affects: ltrace (Ubuntu)
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to ltrace in Ubuntu.
https://bugs.launchpad.net/bugs/1967518

Title:
  ltrace - build on ppc64el fails

Status in ltrace package in Ubuntu:
  New

Bug description:
  The build on ppc64el fails with the following error:

  plt.c: In function ‘arch_elf_add_plt_entry’:
  plt.c:781:17: error: ‘libsym’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
    781 |                 free(libsym);
        |                 ^~~~~~~~~~~~
  cc1: all warnings being treated as errors

  
  This seems to be happening because our back-port of this upstream patch is wrong:

  commit 7f20cb21a0d021ab3b12d8a5187beb728f339b0f
  Author:     Thierry Fauck <thierry at linux.vnet.ibm.com>
  AuthorDate: Tue May 13 07:48:24 2014 -0400
  Commit:     Petr Machata <pmachata at gmail.com>
  CommitDate: Wed Nov 11 19:08:43 2015 +0100

      Support for powerpc64 arch ppc64el

      Signed-off-by: Thierry Fauck <thierry at linux.vnet.ibm.com>

              Add support for ppc64le proc and ELF ABIv2.
              Provides support for irelative and wchar

  
  The original patch leans on a jump to a "fail" label (that is declared near the beginning of the function and does the following):

    free(name);
    return PLT_FAIL;

  In the 0.7.3 code, however, a different definition of the "fail" label
  exists. It is declared later in the function and relies on the libsym
  variable being initialized:

    free(name);
    free(libsym);
    return plt_fail;

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/ltrace/+bug/1967518/+subscriptions