Re: [PATCH] libgcc: Fix uninitialized RA signing on AArch64 [PR107678]

[Richard Sandiford <richard.sandiford@arm.com>]

Richard Sandiford via Gcc-patches <gcc-patches@gcc.gnu.org> writes:
> Jakub Jelinek <jakub@redhat.com> writes:
>> On Wed, Jan 11, 2023 at 11:59:27AM +0000, Wilco Dijkstra wrote:
>>> Hi,
>>> 
>>> > On 1/10/23 19:12, Jakub Jelinek via Gcc-patches wrote:
>>> >> Anyway, the sooner this makes it into gcc trunk, the better, it breaks quite
>>> >> a lot of stuff.
>>> >
>>> > Yep, please, we're also waiting for this patch for pushing to our gcc13 package.
>>> 
>>> Well I'm waiting for an OK from a maintainer... I believe Jakub can approve it as well.
>>
>> Yes, I can, but am not sure it is appropriate.  While I'm familiar with the
>> unwinder, I'm not familiar with the pointer signing, and AArch64 has active
>> maintainers, so I'd prefer to defer the review to them.
>
> I think my main question is: how clean vs hacky is it to use
> REG_UNDEFINED as effectively a toggle bit, rather than using
> REG_UNDEFINED for its intended purpose?
>
> In the review for earlier (May) patch, I'd asked whether it would
> make sense to add a new enum.  Would that be OK from a target-independent
> point of view?  E.g. maybe REG_TOGGLE_ON.
>
> Although we don't AFAIK support using DW_CFA_undefined with RA signing,
> the failure mode would be non-obvious: it would effectively toggle the
> bit on.
>
> It would be good to remove the definition of RA_SIGNED_BIT as well,
> so that people don't accidentally use it in future.

Sorry, just realised that the patch does do that.  But please remove
the comment too!

Richard