From mboxrd@z Thu Jan 1 00:00:00 1970 From: Kuninori Morimoto Subject: Re: [PATCH 4/4] mmc: SDHI: update sh_mobile_sdhi_of_data for r8a7790 Date: Thu, 03 Apr 2014 17:38:07 -0700 (PDT) Message-ID: <877g769bfa.wl%kuninori.morimoto.gx@gmail.com> References: <533D77C3.7000606@codethink.co.uk> Mime-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka") Content-Type: text/plain; charset=US-ASCII Return-path: Received: from mail-pb0-f43.google.com ([209.85.160.43]:59194 "EHLO mail-pb0-f43.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753960AbaDDAiJ (ORCPT ); Thu, 3 Apr 2014 20:38:09 -0400 Received: by mail-pb0-f43.google.com with SMTP id um1so2628966pbc.16 for ; Thu, 03 Apr 2014 17:38:08 -0700 (PDT) In-Reply-To: <533D77C3.7000606@codethink.co.uk> Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Ben Dooks Cc: Linux MMC List , Kuninori Morimoto , linux-kernel , Rob Taylor , William Towle Hi Ben > We tested this patch with our current kernel, and have a question > about the addition of the of_rcar_gen2_compatible area. > > +static const struct sh_mobile_sdhi_of_data of_rcar_gen2_compatible = { > + .tmio_flags = TMIO_MMC_HAS_IDLE_WAIT | > TMIO_MMC_WRPROTECT_DISABLE, > + .capabilities = MMC_CAP_SD_HIGHSPEED | MMC_CAP_SDIO_IRQ, > + .capabilities2 = MMC_CAP2_NO_MULTI_READ, > +}; > > We have been doing some speed tests and this is causing an order > of magnitude speed loss with a number of cards. We're seeing > 1.2-1.6MB/sec instead of 10+MB/sec we would expect from a Class 10 > or UHS-1 card. > > Is there any reason to stop using the MULTI_READ as the system seems > to work without MMC_CAP2_NO_MULTI_READ. I'm not sure detail, but, Renesas Chip has MULTI_READ bug which happens under minor condition. This driver is sharing code together with tmio driver, but this bug is Renesas Chip only. OTOH, original (?) code seems that it has similar issue (= MMC_CAP2_NO_MULTI_READ). Maybe, this flag is over-kill for Renesas Chip, but, I don't want to add Renesas minor flag to driver. I'm not sure this is good or not. > Also, this series does not seem to be in any of the git trees I > would expect. Does it need re-sending? Which git tree do you mean ?? I guess it will be merged to Linus tree when -rc1 was released