From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Message-ID: <18610.20951.764057.574721@cargo.ozlabs.ibm.com> Date: Mon, 25 Aug 2008 16:31:51 +1000 From: Paul Mackerras To: Arnd Bergmann Subject: Re: [Cbe-oss-dev] powerpc/cell/oprofile: fix mutex locking for spu-oprofile In-Reply-To: <200808211014.42683.arnd@arndb.de> References: <1217620879.15667.145.camel@carll-linux-desktop> <200808201519.19453.arnd@arndb.de> <18604.60738.523963.886786@drongo.ozlabs.ibm.com> <200808211014.42683.arnd@arndb.de> Cc: Robert Richter , linux-kernel , linuxppc-dev@ozlabs.org, oprofile-list@lists.sourceforge.net, cel , cbe-oss-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Arnd Bergmann writes: > The patch does not fix a regression, the spu-oprofile code basically never > worked. With the current code in Linux, samples in the profile buffer > can get corrupted because reader and writer to that buffer use different > locks for accessing it. It took us several iterations to come up with > a solution that does not introduce other problems and I didn't want to > push an earlier version that would need more fixups. > > Since rc4 is out now, I understand if you feel more comfortable with > putting the patch into -next instead of -merge. Linus has been getting stricter about only putting in fixes for regressions and serious bugs (see his recent email to Dave Airlie on LKML for instance). I assume that the corruption is just in the data that is supplied to userspace and doesn't extend to any kernel data structures. If it does then we have a much stronger argument for pushing this stuff for 2.6.27. > Note that the second patch is trivial and fixes an oopsable condition > of the kernel, so at least that should still go into 2.6.27. OK, I'll cherry-pick that one for my next batch for Linus. Paul.