From: Dave Martin Subject: Re: [RFC PATCH] arm64/fpsimd: Don't disable softirq when touching FPSIMD/SVE state Date: Thu, 14 Feb 2019 10:34:52 +0000 Message-ID: <20190214103449.GN3567@e103592.cambridge.arm.com> References: <20190208165513.8435-1-julien.grall@arm.com> <20190213143029.ad2kzg7vtuo3zpjk@linutronix.de> <20190213153630.GK3567@e103592.cambridge.arm.com> <20190213165227.7ekekkxazhbaqxoe@linutronix.de> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Ard Biesheuvel , Julien Grall , linux-arm-kernel , linux-rt-users@vger.kernel.org, Catalin Marinas , Will Deacon , Linux Kernel Mailing List To: Sebastian Andrzej Siewior Return-path: Content-Disposition: inline In-Reply-To: <20190213165227.7ekekkxazhbaqxoe@linutronix.de> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-rt-users.vger.kernel.org On Wed, Feb 13, 2019 at 05:52:27PM +0100, Sebastian Andrzej Siewior wrote: > On 2019-02-13 16:40:00 [+0100], Ard Biesheuvel wrote: > > > > This is equal what x86 is currently doing. The naming is slightly > > > > different, there is irq_fpu_usable(). > > > > > > Yes, I think it's basically the same idea. > > > > > > It's been evolving a bit on both sides, but is quite similar now. > > > > > > > may_use_simd() only exists because we have a generic crypto SIMD > > helper, and so we needed something arch agnostic to wrap around > > irq_fpu_usable() > > My question was more if this is helpful and we want to keep or if > it would be better to remove it and always disable BH as part of SIMD > operations. Wouldn't this arbitrarily increase softirq latency? Unconditionally forbidding SIMD in softirq might make more sense. It depends on how important the use cases are... Cheers ---Dave