[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1517669207.31953.120.camel@infradead.org>
Date: Sat, 03 Feb 2018 14:46:47 +0000
From: David Woodhouse <dwmw2@...radead.org>
To: Alan Cox <gnomes@...rguk.ukuu.org.uk>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>,
Thomas Gleixner <tglx@...utronix.de>,
KarimAllah Ahmed <karahmed@...zon.de>, sironi@...zon.de,
the arch/x86 maintainers <x86@...nel.org>,
KVM list <kvm@...r.kernel.org>,
Paolo Bonzini <pbonzini@...hat.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Borislav Petkov <bp@...en8.de>,
Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH] KVM: x86: Reduce retpoline performance impact in
slot_handle_level_range()
On Fri, 2018-02-02 at 21:23 +0000, Alan Cox wrote:
> In addition the problem with switch() is that gcc might decide in some
> cases that the best way to implement your switch is an indirect call
> from a lookup table.
That's also true of the
if (ptr == usualfunction)
usualfunction();
else
*ptr();
construct. Especially if GCC doesn't take into account the increased
cost of indirect branches with retpoline.
> For the simple case how about wrapping the if into
>
> call_likely(foo->bar, usualfunction, args)
>
> as a companion to
>
> foo->bar(args)
>
> that can resolve to nothing special on architectures that don't need it,
> an if/else case on platforms with spectre, and potentially clever
> stuff on any platform where you can beat the compiler by knowing
> probabilities it can't infer ?
Yeah. I'm keen on being able to use something like alternatives to
*change* 'usualfunction' at runtime though. I suspect it'll be a win
for stuff like dma_ops.
But I'm also keen to actually base such things on real data, not just
go randomly "optimising" stuff just because we can. Let's try to make
sure we fix up the real bottlenecks, and not just go crazy.
Download attachment "smime.p7s" of type "application/x-pkcs7-signature" (5213 bytes)
Powered by blists - more mailing lists