linux-mips
[Top] [All Lists]

Re: N32 sigset and __COMPAT_ENDIAN_SWAP__

To: Daniel Jacobowitz <dan@debian.org>
Subject: Re: N32 sigset and __COMPAT_ENDIAN_SWAP__
From: Ralf Baechle <ralf@linux-mips.org>
Date: Thu, 8 Jun 2006 18:36:58 +0100
Cc: "Joseph S. Myers" <joseph@codesourcery.com>, linux-mips@linux-mips.org
In-reply-to: <20060608170310.GA23814@nevyn.them.org>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <Pine.LNX.4.64.0606080134480.26638@digraph.polyomino.org.uk> <20060608165136.GA17152@linux-mips.org> <20060608170310.GA23814@nevyn.them.org>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.4.2.1i
On Thu, Jun 08, 2006 at 01:03:10PM -0400, Daniel Jacobowitz wrote:

> Anyway, I was curious if you knew where this code had come from.  I
> didn't see anything to suggest that anyone besides mipsel ever
> used it, but it entered linux-mips.org via a merge from kernel.org,
> just before git history.
> 
> Oh, right, there's a historical import:
> 
> http://www.kernel.org/git/?p=linux/kernel/git/torvalds/old-2.6-bkcvs.git;a=commitdiff;h=32ed691a4efbc1c43584b7b7a6d782528241bb27

> It was copied from sys32_rt_sigtimedwait, which was wrong at least back
> to the initial revision of signal32.c.  I didn't go back any further.

I can further track it into 2.4 or even pre-2.4 where such
__MIPSEB__ / __MIPSEL__ dependencies did exist under arch/mips64/.
I think it was right until a certain point in 2.5 when
get_sigset and put_sigset were implement in a clever way that
automatically takes care of the endianess issue - but the swapping code
outside arch/mips got forgotten.

  Ralf

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