[PATCH 0/5] SFrame: add support for .cfi_b_key_frame

[Indu Bhagat]

On 12/19/22 7:36 AM, Nick Clifton wrote:
> Hi Indu,
> 
>> - Regression tested cross build of several targets on an x86_64 host 
>> and an
>> aarch64 host using a regression script that checks for failures in 
>> gas, ld,
>> binutils, libctf and libsframe.
> 
> With this patch series applied on top of today's (19 dec) sources I see one
> new failure in the gas testsuite for a toolchain configured as 
> aarch64-linux-gnu:
> 
>    regexp_diff match failure
>    regexp "^    Num FREs: 4$"
>    line   "    Num FREs: 6"
>    regexp_diff match failure
>    regexp "^    0+0008 +sp\+16 +c-16 +c-8\[s\] +$"
>    line   "    0000000000000004  sp+0      u         u[s]         "
>    regexp_diff match failure
>    regexp "^    func idx \[1\]: pc = 0x0, size = 20 bytes, pauth = B key$"
>    line   "    0000000000000008  sp+16     c-16      c-8[s]       "
>    regexp_diff match failure
>    regexp "^    STARTPC + CFA + FP +  RA +$"
>    line   "    func idx [1]: pc = 0x0, size = 20 bytes, pauth = B key"
>    regexp_diff match failure
>    regexp "^    0+0000 +sp\+0 +u +u +$"
>    line   "    STARTPC         CFA       FP        RA           "
>    regexp_diff match failure
>    regexp "^    0+0008 +sp\+16 +c-16 +c-8\[s\] +$"
>    line   "    0000000000000000  sp+0      u         u            "
>    FAIL: SFrame cfi_b_key_frame and cfi_negate_ra_state composite test
> 
> Possibly this is because another patch needs to be applied first ?  
> Please could
> you check and let me know.
> 

This series now needs to be rebased with some changes to the testsuite 
necessary.  This is because the V2 version of the series which added 
support for .cfi_negate_ra_state was committed with "[s]" being emitted 
when RA is in register or stack.

Will rebase, post V2 after retesting.

Thanks
Indu