Re: [PATCH v3 0/5] arm: Add support for multiple stacks on Cortex-M

[Christophe Lyon <christophe.lyon@arm.com>]

On 3/11/22 10:40, Yvan Roux wrote:
> Hi,
> 
> I'll let Christophe address the comments and update the patchset, but I was
> able to test it on an arm linux plateform I don't see any regression in the
> testsuite.

Thanks, good news!

> 
> Sorry I wasn't on the list when the patches were submitted, but I think there
> is a small issue this one:
> 
> [PATCH v3 1/5] gdb/arm: Fix prologue analysis to support vpush
> 
> +	      /* Calculate offsets of saved registers.  */
> +	      for (; number > 0; number--)
> +		{
> +		  addr = pv_add_constant (addr, -8);
> +		  stack.store (addr, 8, pv_register (ARM_D0_REGNUM + number, 0));
> +		}
> 
> Here the D registers to store are computed by adding number to D0 but the first
> register should be the one encoded by D:Vd value which are bits 6 of insn and
> 15..12 of inst2
> 

Indeed, thanks for noticing.

I have a WIP patchset we plan to submit shortly.

Thanks,

Christophe

> Cheers,
> Yvan
> 
> On Sun, Mar 06, 2022 at 10:19:52PM +0100, Christophe Lyon via Gdb-patches wrote:
>>
>>
>> On 3/6/22 11:48, Joel Brobecker wrote:
>>>>>> Will you be able to review this patch series from Christophe?
>>>>> I see that the v3 series was posted on Feb 4th, while on Feb 6th
>>>>> I reviewed v2 :-(. This is really unfortunate that we don't have
>>>>> a better system for tracking whether a version of a series is
>>>>> trully the latest version.
>>>>
>>>> No problem, there's not much difference between v2 and v3,
>>>> so your comments still apply. I should have answered earlier.
>>>>
>>>>>
>>>>> Can you take a look at the comments and question I asked in v2,
>>>>> and publish a v4 which takes those into account? Please Cc: me
>>>>> on the submission of the patches, this will help me remember
>>>>> to prioritize those patches.
>>>>
>>>>
>>>> Sure!
>>>>
>>>> Luis and I changed jobs at the same time, so there's a bit
>>>> of lag :-)
>>>
>>> No problem!
>>>
>>> Just for the avoidance of doubt and you don't end up waiting on me,
>>> my understanding is that the next step is for you to produce a v4
>>> with some of my comments addressed, and to post answers/questions
>>> for my comments. Is that right?
>>>
>>
>> Yes, that's what I meant :-)
>>
>> Thanks,
>>
>> Christophe
>