[Bug 2101149] Re: [SRU] Set default file_bigendian in struct tep_handle

[Pragyansh Chaturvedi]

** Changed in: libtraceevent (Ubuntu)
       Status: New => In Progress

** Changed in: libtraceevent (Ubuntu)
     Assignee: (unassigned) => Pragyansh Chaturvedi (r41k0u)

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to libtraceevent in Ubuntu.
https://bugs.launchpad.net/bugs/2101149

Title:
  [SRU] Set default file_bigendian in struct tep_handle

Status in libtraceevent package in Ubuntu:
  In Progress

Bug description:
  [ Impact ]

   * This relates to https://bugs.launchpad.net/ubuntu/+source/libtracefs/+bug/2062118.
     The failures on s390x are primarily because tep_alloc sets file
     endianness to LITTLE_ENDIAN by default (with no checks)

   * The changes are already accepted into plucky and upstream:
     https://code.launchpad.net/~r41k0u/ubuntu/+source/libtraceevent/+git/libtraceevent/+merge/477042

   * This will fix autopkgtest segfaults of libtracefs on oracular and
  noble for s390x.

  [ Test Plan ]

   * Clone libtracefs (oracular-devel or noble-devel), compile it and the tests (make && make test),
     then run the tests as root (utest/trace-utest), on an s390x machine (running oracular or noble)

   * You can test the fix by installing the patched version of libtraceevent for the release you are
     testing and trying the libtracefs tests again.

  [ Where problems could occur ]

   * This is not likely to break anything, as we can make a safe assumption
     to set file_bigendian to the same value as host_bigendian by default.
     If these are different, the user can set it using event-parse-api

   * In any case, this change seems logical. This does not affect any other
     little-endian architectures, and I think this won't be breaking any workarounds
     people did around this problem.

   * But in case we have some regressions as a result of this change, then there surely would
     be some workaround to this in a patch, which can then be reverted.

  [ Other Info ]

   * This does not fix anything in libtraceevent but does fix the tests for libtracefs on s390x.
     But I have opened this bug against libtraceevent.

   * I see a workaround around this problem in the last patch in d/p/series, but this change does not affect
     it

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/libtraceevent/+bug/2101149/+subscriptions