[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4981FA0B.5010101@zytor.com>
Date: Thu, 29 Jan 2009 10:48:43 -0800
From: "H. Peter Anvin" <hpa@...or.com>
To: Christoph Lameter <cl@...ux-foundation.org>
CC: Mathieu Desnoyers <mathieu.desnoyers@...ymtl.ca>,
Rusty Russell <rusty@...tcorp.com.au>,
Tejun Heo <tj@...nel.org>, Ingo Molnar <mingo@...e.hu>,
Herbert Xu <herbert@...dor.apana.org.au>,
akpm@...ux-foundation.org, brgerst@...il.com,
ebiederm@...ssion.com, travis@....com,
linux-kernel@...r.kernel.org, steiner@....com, hugh@...itas.com,
"David S. Miller" <davem@...emloft.net>, netdev@...r.kernel.org
Subject: Re: [PATCH] percpu: add optimized generic percpu accessors
Christoph Lameter wrote:
>
> gcc/glibc support a __thread attribute to variables. As far as I can tell
> this automatically makes gcc perform the relocation to the current
> context using a segment register.
>
> But its a weird ABI http://people.redhat.com/drepper/tls.pdf. After
> reading that I agree that we should stay with the cpu ops and forget about
> the local and thread stuff in gcc/glibc.
>
We have discussed this a number of times. There are several issues with
it; one being that per-cpu != per-thread (we can switch CPU whereever
we're preempted), and another that many versions of gcc uses hardcoded
registers, in particular %fs on 64 bits, but the kernel *has* to use %gs
since there is no swapfs instruction.
-hpa
--
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel. I don't speak on their behalf.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists