Re: [PATCH v5] arm64: mte: allow async MTE to be upgraded to sync on a per-CPU basis

From: Szabolcs Nagy <szabolcs.nagy@arm.com>
To: Will Deacon <will@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>,
	Peter Collingbourne <pcc@google.com>,
	Vincenzo Frascino <vincenzo.frascino@arm.com>,
	Evgenii Stepanov <eugenis@google.com>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	Tejas Belagod <Tejas.Belagod@arm.com>
Subject: Re: [PATCH v5] arm64: mte: allow async MTE to be upgraded to sync on a per-CPU basis
Date: Mon, 28 Jun 2021 18:21:50 +0100	[thread overview]
Message-ID: <20210628172149.GN13058@arm.com> (raw)
In-Reply-To: <20210628101756.GB5503@willie-the-truck>

The 06/28/2021 11:17, Will Deacon wrote:
> On Fri, Jun 25, 2021 at 03:14:40PM +0100, Szabolcs Nagy wrote:
> > i think a user should be able to ask for sync check
> > mode for a process and get an error if that's not
> > possible.
> 
> Hmm. The question then is what do we do if the sysfs override is changed
> after the application has started running?

if the failure cannot be reported then i think it is not ok
to override.

at least running a process in sync mode sounds useful to me.
with the per cpu override this may need privilege and even
then it may not be possible without significantly affecting
other processes. and sysfs is not a reliable api to figure
out what's going on.

if we need the ability to completely disable sync mode on a
cpu (as opposed to just doing it for performance tuning)
then it can be a boot option, or processes that requested
guaranteed sync mode can be killed.

i'm not against per cpu setting since most code should work
with whatever tag check mode, but if we make that the abi
(that code should work in whatever mode) then i think we
weaken the usability of mte.

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel