Re: [RFC PATCH v2] x86/fpu: make kernel-mode FPU reliably usable in softirqs

From: Ingo Molnar
Date: Thu Mar 06 2025 - 06:46:25 EST

Next message: Masahiro Yamada: "Re: [PATCH v2] kbuild: install-extmod-build: Fix build when specifying KBUILD_OUTPUT"
Previous message: K Prateek Nayak: "[RFC PATCH 3/3] treewide: pipe: Convert all references to pipe->{head,tail,max_usage,ring_size} to unsigned short"
In reply to: Eric Biggers: "Re: [RFC PATCH v2] x86/fpu: make kernel-mode FPU reliably usable in softirqs"
Next in thread: Peter Zijlstra: "Re: [RFC PATCH v2] x86/fpu: make kernel-mode FPU reliably usable in softirqs"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

* Eric Biggers <ebiggers@xxxxxxxxxx> wrote:

> Deleting the ldmxcsr(MXCSR_DEFAULT) adds about 14 MB/s.

BTW., that's still a +3% performance improvement, so it is worth
considering in addition to the current patch IMO.

Thanks,

Ingo

Next message: Masahiro Yamada: "Re: [PATCH v2] kbuild: install-extmod-build: Fix build when specifying KBUILD_OUTPUT"
Previous message: K Prateek Nayak: "[RFC PATCH 3/3] treewide: pipe: Convert all references to pipe->{head,tail,max_usage,ring_size} to unsigned short"
In reply to: Eric Biggers: "Re: [RFC PATCH v2] x86/fpu: make kernel-mode FPU reliably usable in softirqs"
Next in thread: Peter Zijlstra: "Re: [RFC PATCH v2] x86/fpu: make kernel-mode FPU reliably usable in softirqs"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]