linux-mips
[Top] [All Lists]

Re: fadvise on MIPS

To: linux-mips@linux-mips.org
Subject: Re: fadvise on MIPS
From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
Date: Mon, 19 Feb 2007 02:15:15 +0900 (JST)
In-reply-to: <20070217.004812.03978264.anemo@mba.ocn.ne.jp>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <20070217.004329.108739438.anemo@mba.ocn.ne.jp> <20070217.004812.03978264.anemo@mba.ocn.ne.jp>
Sender: linux-mips-bounce@linux-mips.org
On Sat, 17 Feb 2007 00:48:12 +0900 (JST), Atsushi Nemoto <anemo@mba.ocn.ne.jp> 
wrote:
> > 2) On O32, glibc pass a 'long long' argument by hi and lo words, but
> > kernel needs padding word between 'fd' and 'offset' argument.
> > 
> > 3) On N32, glibc pass a 'long long' argument by hi and lo words, but
> > kernel expects a single register value for 'long long' argument.
> 
> And sync_file_range() has some problem too.

Hmm, for other system calls with long long argument, such as
ftruncate64, glibc and uClibc provides mips specific version just for
insert a padding argument.

This is a general rule?  Or for new system call, we can do that
adjustment in kernel to avoid fixing each C libraries?

---
Atsushi Nemoto

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