Re: [PATCH v13 02/14] unwind_user: Add frame pointer support

[Jens Remus <jremus@linux.ibm.com>]

On 10.07.2025 19:08, Steven Rostedt wrote:
> On Thu, 10 Jul 2025 17:41:36 +0200
> Jens Remus <jremus@linux.ibm.com> wrote:
> 
>> cfa + frame->ra_off could be aligned by chance.  So could
>> cfa + frame->fp_off be as well of course.
>>
>> On s390 the CFA must be aligned (as the SP must be aligned) and the
>> FP and RA offsets from CFA must be aligned, as pointer / 64-bit integers
>> (such as 64-bit register values) must be aligned as well.
>>
>> So the CFA (and/or offset), FP offset, and RA offset could be validated
>> individually.  Not sure if that would be over engineering though.
> 
> I wonder if we should just validate that cfa is aligned? Would that work?
> 
> I would think that ra_off and fp_off should be aligned as well and if
> cfa is aligned then it would still be aligned when adding those offsets.

Makes sense, if the base assumption is that the SFrame information is
valid and the primary intend is to check that the used CFA base register
(i.e. SP or FP) was aligned.

Regards,
Jens
-- 
Jens Remus
Linux on Z Development (D3303)
+49-7031-16-1128 Office
jremus@de.ibm.com

IBM

IBM Deutschland Research & Development GmbH; Vorsitzender des Aufsichtsrats: Wolfgang Wendt; Geschäftsführung: David Faller; Sitz der Gesellschaft: Böblingen; Registergericht: Amtsgericht Stuttgart, HRB 243294
IBM Data Privacy Statement: https://www.ibm.com/privacy/