[Top] [All Lists]

Re: [PATCH V2 1/2] asm-offsets.c: adding #define to break circular depen

To: Ralf Baechle <>
Subject: Re: [PATCH V2 1/2] asm-offsets.c: adding #define to break circular dependency
From: Jim Quinlan <>
Date: Thu, 30 Aug 2012 16:35:35 -0400
Cc: David Daney <>,
Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed;; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=CPbZ80Q2xWfpbZUnGDkytDBwt7c3iBvfah4VBYkcvqY=; b=weHB9kIUNbX1lUe8Z54+TzI2L6oDj3neJuNiF6iE75H/3jcdm5FfPsS2dabEUpoWgI z0p6lmwYh4592oVNGhiIQCZvsxex+lFWPlxt7i/wKiav/pNiGHVIrzFH4wIs00pTbVm6 h+457057KDGiNHjtTq5eNg/5ag6S4YmIiG7aSs5ss8D5M7mGZKGhI73/FHP2hkpsqThF z5e7seWks/MLE3ttBbM0MW4im8rEhLfINvF7ZD5V0ka0o2ubL0e3oKCIHHx1xUHcfKZG W3CzIv50btOL+4G5auFUpyRDp5S78DT7zdFqGhRIirYhib0Mfsw4hyQOzVO4MuMZZUhJ joGA==
In-reply-to: <>
List-archive: <>
List-help: <>
List-id: linux-mips <>
List-owner: <>
List-post: <>
List-software: Ecartis version 1.0.0
List-subscribe: <>
List-unsubscribe: <>
References: <> <> <> <>
Are you saying you want me to outline most of the functions in
asm/bitops.h , or do you want me somehow outline just the "else"
clause that invokes raw_local_irq_{save_restore}?


On Thu, Aug 30, 2012 at 10:15 AM, Ralf Baechle <> wrote:
> On Thu, Aug 30, 2012 at 10:06:30AM -0400, Jim Quinlan wrote:
>> I'm not sure the tangle is so easily undone.  The first dependency I see is
>> asm-offsets.c
>> asm/processors.h
>> linux/cpumask.h
>> linux/kernel.h
>> linux/bitops.h
>> asm/bitops.h
>> linux/irqflags.h
>> asm/irqflags.h
>> When compared to other architectures, the MIPS asm/bitops.h seems to
>> include more files at the top, including linux/irqflags.h.
>> Any suggestions?
> This is because MIPS bitops for some ancient processors which don't have
> atomic operations and Cavium cnMIPS cores where disabling interrupts is
> faster than the atomic operations are implemented by disabling interrupts.
> This makes these atomic operations relativly bloated in terms of code size
> generated and may it'd be a good idea to outline the bits.  With a bit
> of luck we even get better cache locality - and fewer header file
> inclusions.
>   Ralf

<Prev in Thread] Current Thread [Next in Thread>