Re: [ PATCH 0/3] Fix GDB reverse execution behavior

[Guinevere Larsen <blarsen@redhat.com>]

On 23/11/2023 00:33, Carl Love wrote:
> GDB developers:
>
> The patch "[PATCH 2/2 ver 8] Fix reverse stepping multiple contiguous
> PC ranges over the line table",
>
> https://sourceware.org/pipermail/gdb-patches/2023-August/201889.html
>
> was posted and pinged several times without any response from the
> developers.  It was reviewed by Guinevere Larsen. The original patch 2
> was intended to fix failures on PowerPC and AArch64.  An additional
> architecture independent change and test was added per feedback from
> the community.
>
> Based on some private discussions on how best to get the patch
> approved, it was suggested that I should break the patch in two to make
> it easier to review.  Specifically, separate out the PowerPC and
> AArch64 fixes from the generic architecture independent changes.  The
> original patch 2 has been split up as suggested.  The hope is this will
> make it easier to review and get approval.
>
> The two new patches are functionally identical to the previously posted
> patch 2.
>
> The patch series now consists of three patches.
>
> Patch 1, adds a new option to gdb_compile to either generate or not
> generate the line table information.  No change from the prior version.
>
> Patch 2, fix the specific GDB issues with reverse stepping over a line
> with multiple statements in the same line for PowerPC and AArch64.
> Behavior on X86-64 did not change.
>
> Patch 3, fix the behavior of GDB for all architectures when executing
> the next command in reverse on a line that contains multiple function
> calls.
>
> The series of three patches has been tested on PowerPC, X86 and AArch64
> with no regression errors.  Please let me know if these patches are
> acceptable for mainline.  Note, Luis Machado did the AArch64 testing.
> Thanks.
>
>                              Carl
>               
>
I've taken a look at this and also see no regressions. I seem to recall 
them fixing a clang regression at some point, which they don't anymore, 
but these have been in the works for so long I think its better to take 
them as is and check the clang regression later.

Reviewed-By: Guinevere Larsen <blarsen@redhat.com>

On a side note, b4 didn't really like this series and I'm not sure why. 
My guess is that all 3 patches have the same email subject, but it may 
also be that I needed --3way when using git am. Either way, just 
mentioning to gather knowledge on what b4 likes or doesn't like.

-- 
Cheers,
Guinevere Larsen
She/Her/Hers