[PATCH] binutils: xfail "gnu-debuglink (objdump 1)" on Solaris/amd64 [PR33243,PR33389]

[Jan Beulich]

On 26.01.2026 10:09, Rainer Orth wrote:
> Hi Jan,
> 
>> On 26.01.2026 09:03, Rainer Orth wrote:
>>>> On 25.01.2026 16:04, Rainer Orth wrote:
>>>>> As reported in PR binutils/33243, one test FAILs on Solaris/amd64:
>>>>>
>>>>> FAIL: gnu-debuglink (objdump 1)
>>>>>
>>>>> 30c30,32
>>>>> <   400660:     ff 35 b2 15 00 00 ff 25 b4 15 00 00 0f 1f 40 00     .5.....%......@.
>>>>> --- 
>>>>>>   400660:     ff 35 b2 15 00 00       push   0x15b2(%rip)        # 401c18 <_GLOBAL_OFFSET_TABLE_+0x8>
>>>>>>   400666:     ff 25 b4 15 00 00       jmp    *0x15b4(%rip)        # 401c20 <_GLOBAL_OFFSET_TABLE_+0x10>
>>>>>>   40066c:     0f 1f 40 00             nopl   0x0(%rax)
>>>>>
>>>>> This is another instance of PR binutils/33389, so this patch xfail's the
>>>>> test.
>>>>
>>>> Well, yes, but why is the XFAIL limited to x86-64?
>>>
>>> because the failure is 64-bit only.  On i386-pc-solaris2.11 the test
>>> PASSes just fine.
>>
>> Well, I suspected as much, but the referenced PR 33389 isn't target specific
>> afaict. Should it actually be (much like 33243 is)? Then again it would be
>> somewhat unexpected for such a (mis)behavior to be target specific. H.J.'s
>> original fix attempt also was entirely target independent.
> 
> I don't know: PR 33243 was 64-bit-only, as is the example HJ gives in
> 33389, which was prompted by me mentioning the issue at some point.  I
> know I've seen this several times in the past, but cannot remember if
> that was only on 64-bit.

It all being pretty odd / unclear, the patch here is still okay for the time
being.

Jan