[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87369gl392.fsf@nanos.tec.linutronix.de>
Date: Mon, 06 Apr 2020 23:21:13 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Benjamin Lamowski <benjamin.lamowski@...nkonzept.com>,
xiaoyao.li@...el.com
Cc: philipp.eppelt@...nkonzept.com, bp@...en8.de, fenghua.yu@...el.com,
hpa@...or.com, linux-kernel@...r.kernel.org, luto@...nel.org,
mingo@...hat.com, nivedita@...m.mit.edu, pbonzini@...hat.com,
peterz@...radead.org, sean.j.christopherson@...el.com,
tony.luck@...el.com, x86@...nel.org
Subject: Re: [PATCH 0/1] x86/split_lock: check split lock feature on initialization
Benjamin Lamowski <benjamin.lamowski@...nkonzept.com> writes:
> During regression testing of our hypervisor[1] with the current git tip,
> we got writes to the TEST_CTRL MSR on hardware that does not support
> split lock detection. While the original split_lock implementation does
> not exhibit this behavior, the reworked initialization from
> dbaba47085b0c unconditionally calls split_lock_verify_msr() from
> split_lock_init().
>
> After the elaborate checks in cpu_set_core_cap_bits() this seems like an
> oversight. The following simple patch fixes our regression by checking
> for X86_FEATURE_SPLIT_LOCK_DETECT before accessing the TEST_CTRL MSR.
No. It's not an oversight, it's a simplification and it's perfectly
legit. rdsmrl_safe() on a unimplemented MSR results in a #GP which is
caught and fixed up. Nothing to see here.
Thanks,
tglx
Powered by blists - more mailing lists