Re: [PATCH v2 2/5] gdbserver/aarch64: Enable FPMR for AArch64 in gdbserver on Linux

[Luis <luis.machado.foss@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1746 bytes --]

On Mon, Oct 13, 2025, 15:53 Ezra Sitorus <ezra.sitorus@arm.com> wrote:

> On Mon, Oct 13, 2025 at 02:40:07PM +0100, Richard Earnshaw wrote:
> > On 11/10/2025 12:50, Luis wrote:
> > > On 07/10/2025 13:31, Ezra.Sitorus@arm.com wrote:
> > > > From: Ezra Sitorus <ezra.sitorus@arm.com>
> > > >
> > > > Support FPMR in gdbserver.
> > >
> > > Nit. Make it "Add support for FPMR register set in gdbserver."
> > >
> > > It is a bit more descriptive.
> >
> > But grammatically wrong, since FPMR is short for Floating Point Mode
> > Register.  So then you'd have:
> >
> > Add support for the floating point mode register register set in
> > gdbserver!!!!
> >
> > R.
> >
> > >
>
> I agree with Richard that it should just be FPMR but I think FPMR register
> set makes sense too. Linux kernel has REGSET_FPMR, as well as REGSET_GPR
> and REGSET_FPR. Maybe you're meant to think something along the lines of a
> 'register set called FPMR, which only has the FPMR'?
>

I meant FPMR the register set exposed by the Linux Kernel via ptrace.
That's what these changes are dealing with.

But either way is fine by me. You could use:

"Add support for FPMR in gdbserver."


> Which means something like this:
>
> > > > +/* Fill BUF with the FPMR register from the regcache.� */
> > > > +
> > > > +static void
> > > > +aarch64_fill_fpmr_regset (struct regcache *regcache, void *buf)
>
> should just be:
>
> Fill BUF with the FPMR from the regcache.
>
> but this works too:
>
> Fill BUF with the FPMR register set from the regcache.
>

Technically it is the register set we're dealing with, but it has a single
register. Again, as long as it is clear I'm fine either way. It is a minor
detail.


> Ezra
>

[-- Attachment #2: Type: text/html, Size: 3065 bytes --]