From: Tejun Heo <tj@kernel.org>
To: Christoph Lameter <cl@linux.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>,
akpm@linux-foundation.org, Yinghai Lu <yinghai@kernel.org>,
Ingo Molnar <mingo@elte.hu>,
Pekka Enberg <penberg@cs.helsinki.fi>,
linux-kernel@vger.kernel.org,
Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Subject: Re: [PATCH] x86: Replace uses of current_cpu_data with this_cpu ops
Date: Wed, 08 Dec 2010 18:57:06 +0100 [thread overview]
Message-ID: <4CFFC6F2.5040304@kernel.org> (raw)
In-Reply-To: <alpine.DEB.2.00.1012081132280.23407@router.home>
Hello, Christoph.
On 12/08/2010 06:33 PM, Christoph Lameter wrote:
> Beginnings of a patch to avoid bitops. If you think this is the right way
> then I will complete it for all uses of cpu_has.
>
> ---
> arch/x86/include/asm/cpufeature.h | 15 +++++++++++++++
> arch/x86/kernel/apic/apic.c | 2 +-
> 2 files changed, 16 insertions(+), 1 deletion(-)
>
> Index: linux-2.6/arch/x86/include/asm/cpufeature.h
> ===================================================================
> --- linux-2.6.orig/arch/x86/include/asm/cpufeature.h 2010-12-08 09:39:31.000000000 -0600
> +++ linux-2.6/arch/x86/include/asm/cpufeature.h 2010-12-08 09:40:02.000000000 -0600
> @@ -221,6 +221,21 @@ extern const char * const x86_power_flag
> ? 1 : \
> test_cpu_cap(c, bit))
>
> +#define this_cpu_has(bit) \
> + (__builtin_constant_p(bit) && \
> + ( (((bit)>>5)==0 && (1UL<<((bit)&31) & REQUIRED_MASK0)) || \
> + (((bit)>>5)==1 && (1UL<<((bit)&31) & REQUIRED_MASK1)) || \
> + (((bit)>>5)==2 && (1UL<<((bit)&31) & REQUIRED_MASK2)) || \
> + (((bit)>>5)==3 && (1UL<<((bit)&31) & REQUIRED_MASK3)) || \
> + (((bit)>>5)==4 && (1UL<<((bit)&31) & REQUIRED_MASK4)) || \
> + (((bit)>>5)==5 && (1UL<<((bit)&31) & REQUIRED_MASK5)) || \
> + (((bit)>>5)==6 && (1UL<<((bit)&31) & REQUIRED_MASK6)) || \
> + (((bit)>>5)==7 && (1UL<<((bit)&31) & REQUIRED_MASK7)) || \
> + (((bit)>>5)==8 && (1UL<<((bit)&31) & REQUIRED_MASK8)) || \
> + (((bit)>>5)==9 && (1UL<<((bit)&31) & REQUIRED_MASK9)) ) \
> + ? 1 : \
It would be nice to factor out the above so that it's not repated for
cpu_has() and this_cpu_has().
> + (this_cpu_read(cpu_info.x86_capabilty) & (1 << bit)))
bit can go beyond unsigned, so 1 << bit doesn't work. Why not just
start with a simple wrapper? ie. sth like
#define this_cpu_has(bit) cpu_has(__this_cpu_ptr(&cpu_info), bit)
We can later optimize in a separate patch along the line of...
#define this_cpu_has(bit) ({ \
__builtin_constant_p(bit) ? \
(cpu_required_bit_test(bit) || \
this_cpu_read(cpu_info.x86_capability + ((bit) >> 5)) & \
(1 << ((bit) & 31))) : \
cpu_has(__this_cpu_ptr(&cpu_info), bit); \
})
Thanks.
--
tejun
next prev parent reply other threads:[~2010-12-08 17:58 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-12-06 17:16 [Use cpuops V1 00/11] Use this_cpu_ops Christoph Lameter
2010-12-06 17:16 ` [Use cpuops V1 01/11] percpucounter: Optimize __percpu_counter_add a bit through the use of this_cpu() options Christoph Lameter
2010-12-07 13:59 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 02/11] vmstat: Optimize zone counter modifications through the use of this cpu operations Christoph Lameter
2010-12-07 14:00 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 03/11] x86: Use this_cpu_ops to optimize code Christoph Lameter
2010-12-06 17:16 ` [Use cpuops V1 04/11] x86: Use this_cpu_ops for current_cpu_data accesses Christoph Lameter
2010-12-07 14:17 ` Tejun Heo
2010-12-07 14:20 ` Tejun Heo
2010-12-07 14:46 ` Christoph Lameter
2010-12-07 14:59 ` Tejun Heo
2010-12-07 15:07 ` Christoph Lameter
2010-12-07 15:08 ` Tejun Heo
2010-12-07 16:21 ` Christoph Lameter
2010-12-08 13:16 ` [PATCH] x86: Replace uses of current_cpu_data with this_cpu ops Tejun Heo
2010-12-08 13:38 ` Eric Dumazet
2010-12-08 15:33 ` Christoph Lameter
2010-12-08 17:17 ` Eric Dumazet
2010-12-08 17:20 ` Tejun Heo
2010-12-08 17:33 ` Christoph Lameter
2010-12-08 17:57 ` Tejun Heo [this message]
2010-12-06 17:16 ` [Use cpuops V1 05/11] core: Replace __get_cpu_var with __this_cpu_read if not used for an address Christoph Lameter
2010-12-07 14:29 ` Tejun Heo
2010-12-07 14:51 ` Christoph Lameter
2010-12-07 15:21 ` Tejun Heo
2010-12-07 15:38 ` Christoph Lameter
2010-12-08 15:24 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 06/11] drivers: " Christoph Lameter
2010-12-07 14:31 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 07/11] kprobes: Use this_cpu_ops Christoph Lameter
2010-12-07 15:03 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 08/11] Fakekey: Simplify speakup_fake_key_pressed through this_cpu_ops Christoph Lameter
2010-12-07 15:07 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 09/11] Connector: Use this_cpu operations Christoph Lameter
2010-12-07 15:08 ` Tejun Heo
2010-12-07 15:36 ` Christoph Lameter
2010-12-06 17:16 ` [Use cpuops V1 10/11] fs: Use this_cpu_xx operations in buffer.c Christoph Lameter
2010-12-07 15:16 ` Tejun Heo
2010-12-06 17:16 ` [Use cpuops V1 11/11] Xen: Use this_cpu_ops Christoph Lameter
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4CFFC6F2.5040304@kernel.org \
--to=tj@kernel.org \
--cc=akpm@linux-foundation.org \
--cc=cl@linux.com \
--cc=eric.dumazet@gmail.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mathieu.desnoyers@efficios.com \
--cc=mingo@elte.hu \
--cc=penberg@cs.helsinki.fi \
--cc=yinghai@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).