mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH v2] x86/fpu: Use vmemdup_user() in xstateregs_set()
@ 2026-08-05 14:03 Thorsten Blum
  2026-08-31  9:57 ` Thorsten Blum
  0 siblings, 1 reply; 2+ messages in thread
From: Thorsten Blum @ 2026-08-05 14:03 UTC (permalink / raw)
  To: Thomas Gleixner, Ingo Molnar, Borislav Petkov, Dave Hansen, x86,
	H. Peter Anvin
  Cc: Thorsten Blum, linux-kernel

Replace the open-coded vmalloc() and copy_from_user() with
vmemdup_user() to simplify xstateregs_set().

vmemdup_user() returns an ERR_PTR() on failure, preserving the existing
-ENOMEM and -EFAULT error codes. Since vmemdup_user() is backed by
kvmalloc(), use kvfree() to free the buffer instead.

Return early on error and drop the obsolete out label.

Performance should be similar, and xstateregs_set() is not a hot path.
Page alignment should not matter because the buffer is only used as a
memcpy() source in copy_uabi_from_kernel_to_xstate().

The allocation behavior differs because vmemdup_user() uses GFP_USER,
which adds __GFP_HARDWALL and may affect cpuset allocations.

tools/testing/selftests/x86/avx_64 passed on an Alder Lake CPU.

Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
---
Changes in v2:
- Add performance analysis and selftest result to the changelog (Ingo)
- v1: https://lore.kernel.org/r/20260720195534.70111-3-thorsten.blum@linux.dev/
---
 arch/x86/kernel/fpu/regset.c | 17 ++++++-----------
 1 file changed, 6 insertions(+), 11 deletions(-)

diff --git a/arch/x86/kernel/fpu/regset.c b/arch/x86/kernel/fpu/regset.c
index 0986c2200adc..8dc9de732c78 100644
--- a/arch/x86/kernel/fpu/regset.c
+++ b/arch/x86/kernel/fpu/regset.c
@@ -3,7 +3,8 @@
  * FPU register's regset abstraction, for ptrace, core dumps, etc.
  */
 #include <linux/sched/task_stack.h>
-#include <linux/vmalloc.h>
+#include <linux/slab.h>
+#include <linux/string.h>
 
 #include <asm/fpu/api.h>
 #include <asm/fpu/signal.h>
@@ -157,21 +158,15 @@ int xstateregs_set(struct task_struct *target, const struct user_regset *regset,
 		return -EFAULT;
 
 	if (!kbuf) {
-		tmpbuf = vmalloc(count);
-		if (!tmpbuf)
-			return -ENOMEM;
-
-		if (copy_from_user(tmpbuf, ubuf, count)) {
-			ret = -EFAULT;
-			goto out;
-		}
+		tmpbuf = vmemdup_user(ubuf, count);
+		if (IS_ERR(tmpbuf))
+			return PTR_ERR(tmpbuf);
 	}
 
 	fpu_force_restore(fpu);
 	ret = copy_uabi_from_kernel_to_xstate(fpu->fpstate, kbuf ?: tmpbuf, &target->thread.pkru);
 
-out:
-	vfree(tmpbuf);
+	kvfree(tmpbuf);
 	return ret;
 }
 

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [PATCH v2] x86/fpu: Use vmemdup_user() in xstateregs_set()
  2026-08-05 14:03 [PATCH v2] x86/fpu: Use vmemdup_user() in xstateregs_set() Thorsten Blum
@ 2026-08-31  9:57 ` Thorsten Blum
  0 siblings, 0 replies; 2+ messages in thread
From: Thorsten Blum @ 2026-08-31  9:57 UTC (permalink / raw)
  To: Thomas Gleixner, Ingo Molnar, Borislav Petkov, Dave Hansen, x86,
	H. Peter Anvin
  Cc: linux-kernel

On Wed, Aug 05, 2026 at 04:03:45PM +0200, Thorsten Blum wrote:
> Replace the open-coded vmalloc() and copy_from_user() with
> vmemdup_user() to simplify xstateregs_set().
> 
> vmemdup_user() returns an ERR_PTR() on failure, preserving the existing
> -ENOMEM and -EFAULT error codes. Since vmemdup_user() is backed by
> kvmalloc(), use kvfree() to free the buffer instead.
> 
> Return early on error and drop the obsolete out label.
> 
> Performance should be similar, and xstateregs_set() is not a hot path.
> Page alignment should not matter because the buffer is only used as a
> memcpy() source in copy_uabi_from_kernel_to_xstate().
> 
> The allocation behavior differs because vmemdup_user() uses GFP_USER,
> which adds __GFP_HARDWALL and may affect cpuset allocations.
> 
> tools/testing/selftests/x86/avx_64 passed on an Alder Lake CPU.
> 
> Signed-off-by: Thorsten Blum <thorsten.blum@linux.dev>
> ---
> Changes in v2:
> - Add performance analysis and selftest result to the changelog (Ingo)
> - v1: https://lore.kernel.org/r/20260720195534.70111-3-thorsten.blum@linux.dev/
> ---
>  arch/x86/kernel/fpu/regset.c | 17 ++++++-----------
>  1 file changed, 6 insertions(+), 11 deletions(-)

Hi Ingo,

Could you please give this another look now that I've added the
performance analysis to the changelog? Is there anything else missing?

Thanks,
Thorsten

> diff --git a/arch/x86/kernel/fpu/regset.c b/arch/x86/kernel/fpu/regset.c
> index 0986c2200adc..8dc9de732c78 100644
> --- a/arch/x86/kernel/fpu/regset.c
> +++ b/arch/x86/kernel/fpu/regset.c
> @@ -3,7 +3,8 @@
>   * FPU register's regset abstraction, for ptrace, core dumps, etc.
>   */
>  #include <linux/sched/task_stack.h>
> -#include <linux/vmalloc.h>
> +#include <linux/slab.h>
> +#include <linux/string.h>
>  
>  #include <asm/fpu/api.h>
>  #include <asm/fpu/signal.h>
> @@ -157,21 +158,15 @@ int xstateregs_set(struct task_struct *target, const struct user_regset *regset,
>  		return -EFAULT;
>  
>  	if (!kbuf) {
> -		tmpbuf = vmalloc(count);
> -		if (!tmpbuf)
> -			return -ENOMEM;
> -
> -		if (copy_from_user(tmpbuf, ubuf, count)) {
> -			ret = -EFAULT;
> -			goto out;
> -		}
> +		tmpbuf = vmemdup_user(ubuf, count);
> +		if (IS_ERR(tmpbuf))
> +			return PTR_ERR(tmpbuf);
>  	}
>  
>  	fpu_force_restore(fpu);
>  	ret = copy_uabi_from_kernel_to_xstate(fpu->fpstate, kbuf ?: tmpbuf, &target->thread.pkru);
>  
> -out:
> -	vfree(tmpbuf);
> +	kvfree(tmpbuf);
>  	return ret;
>  }
>  

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-08-31  9:57 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-08-05 14:03 [PATCH v2] x86/fpu: Use vmemdup_user() in xstateregs_set() Thorsten Blum
2026-08-31  9:57 ` Thorsten Blum

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®