From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bartlomiej Zolnierkiewicz Subject: Re: [PATCH 0/3] SATA: Fine-tuning for two function implementations Date: Thu, 11 May 2017 17:01:34 +0200 Message-ID: <3428511.iH6kg31xz3@amdc3058> References: <20170428215334.GH22354@htj.duckdns.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7Bit Return-path: In-reply-to: <20170428215334.GH22354@htj.duckdns.org> Sender: linux-kernel-owner@vger.kernel.org To: Tejun Heo Cc: SF Markus Elfring , linux-ide@vger.kernel.org, Hans de Goede , LKML , kernel-janitors@vger.kernel.org List-Id: linux-ide@vger.kernel.org Hi, On Friday, April 28, 2017 05:53:34 PM Tejun Heo wrote: > Hello, > > On Tue, Apr 18, 2017 at 10:00:37PM +0200, SF Markus Elfring wrote: > > From: Markus Elfring > > Date: Tue, 18 Apr 2017 21:54:32 +0200 > > > > A few update suggestions were taken into account > > from static source code analysis. > > Hmmm, allocs -> callocs. Are these actually beneficial? If so, why? > Because one multiplication is rolled into the call? Each conversion (i.e. I tried the one from patch #1) seems to add an extra 24 bytes to the resulting code size (using gcc 4.8.4 for ARM32 cross-compilation) so I don't see much point in the automatic conversions. Only instances containing size calculations with real possibility for integer overflows should be converted and the patchset under discussion contains no such instances. Best regards, -- Bartlomiej Zolnierkiewicz Samsung R&D Institute Poland Samsung Electronics