[PATCH 0/4] Fix relocatable links of SFrame sections

[Indu Bhagat]

On 4/1/25 1:25 AM, Jens Remus wrote:
> On 31.03.2025 20:52, Indu Bhagat wrote:
> 
>> This patch series resolves two PRs related to relocatable links
>> involving SFrame sections:
>>
>> PR libsframe/32589 - function start address is zero in SFrame section 
>> dump
>> PR ld/32666 - Incorrect .rela.sframe when using ld -r
>>
>> The first patch in the series addresses a thinko in GAS/ld.bfd related
>> to SFrame FDE func_start_addr.  Subsequent patches are the bugfixes.
>> Patch 3 and 4 will be combined at commit time; the fix for PR ld/32666
>> currently remains split in patch 3 and 4 for ease of review.
>>
>> This series was previously posted as RFC earlier:
>> https://sourceware.org/pipermail/binutils/2025-March/139853.html
> 
>> Indu Bhagat (4):
>>    gas: ld: sframe: FDE func start address is from start of section
>>    objdump, readelf: sframe: apply relocations before textual dump
>>    ld: bfd: sframe: fix incorrect r_offset in RELA entries
>>    ld: bfd: sframe: fix incorrect r_addend in RELA entries
> 
> Did you consider to include or squash my related patch "[RFC PATCH] ld:
> bfd: sframe: Update section size also for relocatable links" into your
> series?  Are there any issues with it?
> 
> https://inbox.sourceware.org/binutils/20250312163640.218942-1- 
> jremus@linux.ibm.com/
> 

Given that without your patch, relocatable links will have issues in 
cross builds, I should include it in this series.

Thanks and apologies for the delay in getting back on your patch,
Indu

> N.B. I will have to touch flip_sframe() to ignore trailing padding bytes
> of zero with my upcoming s390x SFrame support, which would then hide the
> issue of excessive zero padding in merged .sframe in relocatable links.
> 
> Thanks and regards,
> Jens