Return-path: Received: from smtp.codeaurora.org ([198.145.29.96]:37883 "EHLO smtp.codeaurora.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754120AbcJLIBk (ORCPT ); Wed, 12 Oct 2016 04:01:40 -0400 From: Kalle Valo To: Larry Finger Cc: devel@driverdev.osuosl.org, linux-wireless@vger.kernel.org, "Stable \[ 4 . 8+ \]" , Julia Lawall , Thorsten Leemhuis Subject: Re: [PATCH] rtlwifi: Fix regression caused by commit d86e64768859 References: <20161010233211.13519-1-Larry.Finger@lwfinger.net> Date: Wed, 12 Oct 2016 10:53:01 +0300 In-Reply-To: <20161010233211.13519-1-Larry.Finger@lwfinger.net> (Larry Finger's message of "Mon, 10 Oct 2016 18:32:11 -0500") Message-ID: <87r37mf1gi.fsf@kamboji.qca.qualcomm.com> (sfid-20161012_100142_392375_28D88F6C) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-wireless-owner@vger.kernel.org List-ID: Larry Finger writes: > In commit d86e64768859 ("rtlwifi: rtl818x: constify local structures"), > the configuration struct for most of the drivers was changed to be > constant. The problem is that five of the modified drivers need to be > able to update the firmware name based on the exact model of the card. > As the file names were stored in one of the members of that struct, > these drivers would fail with a kernel BUG splat when they tried to > update the firmware name. > > Rather than reverting the previous commit, I used a suggestion by > Johannes Berg and made the firmware file name pointers be local to > the routines that update the software variables. > > The configuration struct of rtl8192cu, which was not touched in the > previous patch, is now constantfied. > > Fixes: d86e64768859 ("rtlwifi: rtl818x: constify local structures") > Suggested-by: Johannes Berg > Signed-off-by: Larry Finger > Cc: Stable [4.8+] This should be: Cc: Stable # 4.8 And the plus sign isn't needed as I'll push this to 4.9. > Cc: Julia Lawall > --- > Kalle, > > My apologies for letting these bugs to get by my review and testing. No worries, this is business as usual :) Excellent that you noticed this so early. And this again shows that simple cleanup patches are not without cost, there is a clear risk. Has anyone done any statistics how much cleanup patches cause regressions? My guess is that in a 5% ballpark of the cleanup patches have some sort of issues. Of course in this case the cleanup patch revealed a design flaw in the driver, so in this case it was very useful. > As they affect kernel 4.8, please push this patch as soon as possible. > To reiterate, this patch replaces the one entitled 'Revert "rtlwifi: > rtl818x: constify local structures"' Ok, understood. Many thanks for making this super clear for me, makes my life a lot easier. I'm hoping that this will make it to 4.9-rc2. CCing Thorsten to get this regression to his radar. -- Kalle Valo