[Bug default/30307] New: arch specific self compare problem with bpftrace

[Bug default/30307] New: arch specific self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

            Bug ID: 30307
           Summary: arch specific self compare problem with bpftrace
           Product: libabigail
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: default
          Assignee: dodji at redhat dot com
          Reporter: woodard at redhat dot com
                CC: libabigail at sourceware dot org
  Target Milestone: ---

This is being split off from 29673 because I do not think that this is the same
bug.

This was with the latest trunk. 
Note that this problem only seems to occur for s390x

fedabipkgdiff --self-compare -a --from fc37 bpftrace
Comparing the ABI of binaries between bpftrace-0.16.0-3.fc37.aarch64.rpm and
bpftrace-0.16.0-3.fc37.aarch64.rpm:

==== SELF CHECK SUCCEEDED for 'bpftrace' ====
==== SELF CHECK SUCCEEDED for 'bpftrace-aotrt' ====

Comparing the ABI of binaries between bpftrace-0.16.0-3.fc37.ppc64le.rpm and
bpftrace-0.16.0-3.fc37.ppc64le.rpm:

==== SELF CHECK SUCCEEDED for 'bpftrace' ====
==== SELF CHECK SUCCEEDED for 'bpftrace-aotrt' ====

Comparing the ABI of binaries between bpftrace-0.16.0-3.fc37.s390x.rpm and
bpftrace-0.16.0-3.fc37.s390x.rpm:

======== comparing'bpftrace' to itself wrongly yielded result: ===========
  Functions changes summary: 0 Removed, 0 Changed (96 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed (1 filtered out), 0 Added
variable

===SELF CHECK FAILED for 'bpftrace'
======== comparing'bpftrace-aotrt' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 0 Changed (96 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed (1 filtered out), 0 Added
variable

===SELF CHECK FAILED for 'bpftrace-aotrt'

Comparing the ABI of binaries between bpftrace-0.16.0-3.fc37.x86_64.rpm and
bpftrace-0.16.0-3.fc37.x86_64.rpm:

==== SELF CHECK SUCCEEDED for 'bpftrace' ====
==== SELF CHECK SUCCEEDED for 'bpftrace-aotrt' ====

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] arch specific self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

--- Comment #1 from Ben Woodard <woodard at redhat dot com> ---
btrfs-progs has a similar arch specific error. However, instead of affecting
s390x it affects ppc64le:

Comparing the ABI of binaries between btrfs-progs-6.2.2-1.fc37.ppc64le.rpm and
btrfs-progs-6.2.2-1.fc37.ppc64le.rpm:

======== comparing'btrfs' to itself wrongly yielded result: ===========
  Functions changes summary: 0 Removed, 0 Changed (62 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfs'
==== SELF CHECK SUCCEEDED for 'btrfs-convert' ====
==== SELF CHECK SUCCEEDED for 'mkfs.btrfs' ====
======== comparing'btrfs-image' to itself wrongly yielded result: ===========
  Functions changes summary: 0 Removed, 0 Changed (61 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfs-image'
======== comparing'btrfstune' to itself wrongly yielded result: ===========
  Functions changes summary: 0 Removed, 0 Changed (61 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfstune'
======== comparing'btrfs-map-logical' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 0 Changed (61 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfs-map-logical'
======== comparing'btrfs-select-super' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 0 Changed (61 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfs-select-super'
======== comparing'btrfs-find-root' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 0 Changed (61 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'btrfs-find-root'

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] arch specific self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

--- Comment #2 from Ben Woodard <woodard at redhat dot com> ---
darktable seems to have a similar problem however in this case it is ppc64le
and x86_64

Comparing the ABI of binaries between darktable-4.2.1-1.fc37.x86_64.rpm and
darktable-4.2.1-1.fc37.x86_64.rpm:

======== comparing'libdarktable.so' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 0 Changed (88 filtered out), 0 Added
functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

===SELF CHECK FAILED for 'libdarktable.so'

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] arch specific self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

--- Comment #3 from Ben Woodard <woodard at redhat dot com> ---
The problem seeems to be some confustion between a pointer with a typedef'd
type vs the underlying structure.

$ abipkgdiff --self-check  --harmless --d1
/home/ben/.cache/fedabipkgdiff/downloads/darktable-tools-noise-debuginfo-4.2.1-1.fc37.x86_64.rpm
 --d1
/home/ben/.cache/fedabipkgdiff/downloads/darktable-debuginfo-4.2.1-1.fc37.x86_64.rpm
/home/ben/.cache/fedabipkgdiff/downloads/darktable-4.2.1-1.fc37.x86_64.rpm 

======== comparing'libdarktable.so' to itself wrongly yielded result:
===========
  Functions changes summary: 0 Removed, 88 Changed, 0 Added functions
  Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

  88 functions with some indirect sub-type change:

    [C] 'function void _dev_add_history_item(dt_develop_t*, dt_iop_module_t*,
gboolean, gboolean)' at develop.c:990:1 has some indirect sub-type changes:
      parameter 1 of type 'dt_develop_t*' has sub-type changes:
        in pointed to type 'typedef dt_develop_t' at develop.h:336:1:
          underlying type 'struct dt_develop_t' at develop.h:149:1 changed:
            type size hasn't changed
            3 data member changes:
              type of 'dt_iop_module_t* gui_module' changed:
                in pointed to type 'struct dt_iop_module_t' at imageop.h:183:1:
                  type size hasn't changed
                  52 data member changes:
                    type of 'void (dt_iop_module_so_t*)* init_presets' changed:
                      in pointed to type 'function type void
(dt_iop_module_so_t*)':
                        parameter 1 of type 'dt_iop_module_so_t*' has sub-type
changes:
                          in pointed to type 'struct dt_iop_module_so_t' at
imageop.h:155:1:
                            type size hasn't changed
                            49 data member changes:
                              type of 'void (dt_iop_module_so_t*)*
init_presets' changed:
                                in pointed to type 'function type void
(dt_iop_module_so_t*)':
                                  parameter 1 of type 'dt_iop_module_so_t*' has
sub-type changes:
                                    pointed to type 'struct dt_iop_module_so_t'
changed, as being reported
<snip>

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

Ben Woodard <woodard at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|arch specific self compare  |self compare problem with
                   |problem with bpftrace       |bpftrace

--- Comment #4 from Ben Woodard <woodard at redhat dot com> ---
Enblightenment seems to have a similar problem however it is not arch specific.
This calls into question whether the problem is more an arch specific problem
or if it is some sort of  race condition in doing type canonicalization. When
doing the underlying abipkgdiff it seems to be the same kind of confusion of
typedef and a struct.

type of 'E_Zone* zone' changed:
in pointed to type 'typedef E_Zone' at e_zone.h:16:1:
underlying type 'struct _E_Zone' changed, as being reported

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] self compare problem with bpftrace

[woodard at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

Ben Woodard <woodard at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Blocks|                            |29413


Referenced Bugs:

https://sourceware.org/bugzilla/show_bug.cgi?id=29413
[Bug 29413] libabigail 2.3 self-compare metabug
-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug default/30307] self compare problem with bpftrace

[dodji at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=30307

dodji at redhat dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #5 from dodji at redhat dot com ---
The bpftrace issue seems to be solved by commit
https://sourceware.org/git/?p=libabigail.git;a=commit;h=4bbf997eb7fcf55a04302b26ac18ae5252336164.

There seem to be a slight btrfs-progs self compare change (all filtered out)
which makes that one different from the this one.

I am still closing this issue as the bpftrace one was quite substantial and
specific.  Maybe the btrfs-progs one should be opened separately.

-- 
You are receiving this mail because:
You are on the CC list for the bug.