Re: [PATCH v3 04/10] x86/fpu/xstate: Correct guest fpstate size calculation

[Chang S. Bae]

On 3/7/2025 6:49 PM, Chao Gao wrote:
> On Fri, Mar 07, 2025 at 01:37:15PM -0800, Chang S. Bae wrote:
> > On 3/7/2025 8:41 AM, Chao Gao wrote:
> > > diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c
> > > index 6166a928d3f5..adc34914634e 100644
> > > --- a/arch/x86/kernel/fpu/core.c
> > > +++ b/arch/x86/kernel/fpu/core.c
> > > @@ -218,7 +218,7 @@ bool fpu_alloc_guest_fpstate(struct fpu_guest *gfpu)
> > >    	struct fpstate *fpstate;
> > >    	unsigned int size;
> > > -	size = fpu_user_cfg.default_size + ALIGN(offsetof(struct fpstate, regs), 64);
> > > +	size = fpu_kernel_cfg.default_size + ALIGN(offsetof(struct fpstate, regs), 64);
> > >    	fpstate = vzalloc(size);
> > >    	if (!fpstate)
> > >    		return false;
> >
> > BTW, did you ever base this series on the tip/master branch? The fix has
> > already been merged there:
> >
> >   1937e18cc3cf ("x86/fpu: Fix guest FPU state buffer allocation size")
>
> Thanks for the information. I will remove this patch.

But, I think there is a fallout that someone should follow up:

The merged patch ensures size consistency between 
fpu_alloc_guest_fpstate() and fpstate_realloc(), maintaining a 
consistent reference to the kernel buffer size. However, within 
fpu_alloc_guest_fpstate(), fpu_guest->xfeatures should also be adjusted 
accordingly for consistency. Instead of referencing fpu_user_cfg, it 
should reference fpu_kernel_cfg.

Thanks,
CHang