From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeremie Samuel Subject: Re: [PATCH 0/8] sdhci: Move real work out of an atomic context Date: Thu, 11 Jul 2013 10:28:33 +0200 Message-ID: <51DE6CB1.7040609@parrot.com> References: <1373384652-21958-1-git-send-email-jeremie.samuel.ext@parrot.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1"; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from co202.xi-lite.net ([149.6.83.202]:45655 "EHLO co202.xi-lite.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755342Ab3GKI2v (ORCPT ); Thu, 11 Jul 2013 04:28:51 -0400 In-Reply-To: Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Philip Rakity Cc: Chris Ball , "linux-mmc@vger.kernel.org" , Matthieu Castet , =?ISO-8859-1?Q?Gr=E9gor_?= =?ISO-8859-1?Q?BOIRIE?= On 09/07/2013 17:52, Philip Rakity wrote: > On Jul 9, 2013, at 4:44 PM, Jeremie Samuel wrote: > >> Hi all, >> >> Currently the sdhci driver does everything in the atomic context. >> And what is worse, PIO transfers are made from the IRQ handler. >> >> Some patches were already submitted to solve this issue. But there were >> rejected because they involved new issues. >> >> This set of patches is an evolution of an old patch from Anton Vorontsov. >> I tried to fix all the problems involved by the patches. I tested it for >> several time now with SD cards and SDIO. >> >> So, this patch set reworks sdhci code to avoid atomic context, >> almost completely. >> >> Thanks, >> > Running DDR50, SDR104 or HS200 what is the performance impact ? I've realised performance tests with iozone benchmark and the performance impact was approximatively 0.5%. > > >> Jeremie Samuel >> >> Jeremie Samuel (8): >> sdhci: Turn timeout timer into delayed work >> sdhci: Turn tuning timeout timer into delayed work >> sdhci: Use work structs instead of tasklets >> sdhci: Use threaded IRQ handler >> sdhci: Delay led blinking >> sdhci: Turn host->lock into a mutex >> sdhci: Get rid of mdelay()s where it is safe and makes sense >> sdhci: Use jiffies instead of a timeout counter >> >> drivers/mmc/host/sdhci.c | 327 ++++++++++++++++++++++----------------------- >> include/linux/mmc/sdhci.h | 13 +- >> 2 files changed, 168 insertions(+), 172 deletions(-) >> >> -- >> 1.7.10.4 >> >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-mmc" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html