[From nobody Sun Jul 5 16:40:50 2009 Return-Path: <torvalds@osdl.org> X-Original-To: andrea@wotan.suse.de Received: from hermes.suse.de (hermes.suse.de [149.44.160.1]) by wotan.suse.de (Postfix) with ESMTP id 8347D224B33 for <andrea@wotan.suse.de>; Mon, 28 Feb 2005 17:24:01 +0100 (CET) Received: by hermes.suse.de (Postfix) id 7BAE0151676; Mon, 28 Feb 2005 17:24:01 +0100 (CET) Received: from scanhost.suse.de (scanhost.suse.de [149.44.160.36]) by hermes.suse.de (Postfix) with ESMTP id 752791519E3 for <andrea@suse.de>; Mon, 28 Feb 2005 17:24:01 +0100 (CET) Received: from hermes.suse.de ([149.44.160.1]) by scanhost.suse.de (scanhost [149.44.160.36]) (amavisd-new, port 10025) with ESMTP id 06256-11 for <andrea@suse.de>; Mon, 28 Feb 2005 17:23:59 +0100 (CET) Received: from Cantor.suse.de (mail-ex.suse.de [195.135.220.2]) (using TLSv1 with cipher EDH-RSA-DES-CBC3-SHA (168/168 bits)) (No client certificate requested) by hermes.suse.de (Postfix) with ESMTP id EF1F2151676 for <andrea@suse.de>; Mon, 28 Feb 2005 17:23:58 +0100 (CET) Received: from smtp.osdl.org (fire.osdl.org [65.172.181.4]) (using TLSv1 with cipher EDH-RSA-DES-CBC3-SHA (168/168 bits)) (No client certificate requested) by Cantor.suse.de (Postfix) with ESMTP id 6B463152BF80 for <andrea@suse.de>; Mon, 28 Feb 2005 17:23:58 +0100 (CET) Received: from shell0.pdx.osdl.net (fw.osdl.org [65.172.181.6]) by smtp.osdl.org (8.12.8/8.12.8) with ESMTP id j1SGNsqi025667 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Mon, 28 Feb 2005 08:23:56 -0800 Received: from localhost (shell0.pdx.osdl.net [10.9.0.31]) by shell0.pdx.osdl.net (8.13.1/8.11.6) with ESMTP id j1SGNreE025663; Mon, 28 Feb 2005 08:23:54 -0800 Date: Mon, 28 Feb 2005 08:25:07 -0800 (PST) From: Linus Torvalds <torvalds@osdl.org> To: Andrea Arcangeli <andrea@suse.de> Cc: Andrew Morton <akpm@osdl.org>, linux-kernel@vger.kernel.org Subject: Re: two pipe bugfixes In-Reply-To: <20050228042544.GA8742@opteron.random> Message-ID: <Pine.LNX.4.58.0502272143500.25732@ppc970.osdl.org> References: <20050228042544.GA8742@opteron.random> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Received-SPF: pass (domain of torvalds@osdl.org designates 65.172.181.6 as permitted sender) X-MIMEDefang-Filter: osdl$Revision: 1.103 $ X-Virus-Scanned: by amavisd-new at scanhost.suse.de X-Spam-Checker-Version: SpamAssassin 3.0.0 (2004-09-13) on opteron.random X-Spam-Level: X-Spam-Status: No, score=-2.8 required=5.0 tests=AWL,BAYES_00 autolearn=ham version=3.0.0 On Mon, 28 Feb 2005, Andrea Arcangeli wrote: > > and it started to return this since 2.6.11-rc: > > pipe([3, 4]) = 0 > write(4, "qqqqq", 5) = 5 > select(5, [4], [4], [4], {0, 0}) = 2 (in [4], out [4], left {0, > 0}) > close(3) = 0 > write(4, "qqqqq", 5) = 5 Ahh, yes. The write merging code doesn't check for no readers. > IMHO the really wrong thing is that we always set POLLIN (even for > output filedescriptors that will never allow any data to be read). However, that has always been true. Look at the old code: it would set POLLIN for a non-empty pipe for both readers and writers (and do POLLOUT for empty pipes both for readers and writers). In fact, your very own original strace shows that - it shows "in [4]" even though fd 4 is a write-only fd. The new code does nothing really different. POLLIN is still there for a non-empty pipe, just like it was before. It's just that when you have multiple buffers, POLLOUT can _also_ be true, since even if you have _some_ data in the pipe, you can still do a write of a full PIPE_BUF. So the difference is not at all the one you're talking about, and the "bug" you claim to fix was there before too. The fact is that if this broke python-twisted, then it just happened to work before by mistake. And python-twisted is just plain bogus. That said, I agree with the fact that it's probably not the right thing to do, and never was. And if fixing it makes a difference to python-twisted, then hey, that's a benefit, but not a reason for the patch. I don't agree with your patch, though - I don't like your lack of parenthesis ;) Linus ]