From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alim Akhtar Subject: Re: [PATCH v2 3/4] mmc: dw_mmc: Convert to mmc_send_tuning() Date: Wed, 24 Dec 2014 19:37:15 +0530 Message-ID: References: <1417780783-12939-1-git-send-email-ulf.hansson@linaro.org> <1417780783-12939-4-git-send-email-ulf.hansson@linaro.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Return-path: Received: from mail-yh0-f47.google.com ([209.85.213.47]:45947 "EHLO mail-yh0-f47.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751602AbaLXOH4 (ORCPT ); Wed, 24 Dec 2014 09:07:56 -0500 Received: by mail-yh0-f47.google.com with SMTP id f73so4033822yha.20 for ; Wed, 24 Dec 2014 06:07:56 -0800 (PST) In-Reply-To: Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Ulf Hansson Cc: "linux-mmc@vger.kernel.org" , Chris Ball , Seungwon Jeon , Jaehoon Chung , Shawn Guo , Sascha Hauer , Aisheng Dong , Stephen Boyd , Minda Chen , Barry Song Hi Ulf, On Mon, Dec 22, 2014 at 8:11 PM, Ulf Hansson wrote: > On 20 December 2014 at 14:18, Alim Akhtar wrote: >> Hi Ulf, >> >> On Mon, Dec 8, 2014 at 3:40 PM, Ulf Hansson wrote: >>> On 6 December 2014 at 13:43, Alim Akhtar wrote: >>>> Hi Ulf, >>>> >>>> On Fri, Dec 5, 2014 at 5:29 PM, Ulf Hansson wrote: >>>>> Instead of having a local hack taking care of sending the tuning >>>>> command and as well to verify the response pattern, let's convert to >>>>> the common mmc_send_tuning() API. >>>>> >>>>> This change affects the Exynos variant, since it's the only one which >>>>> support the dw_mmc's ->execute_tuning() callback. >>>>> >>>>> Signed-off-by: Ulf Hansson >>>>> --- >>>>> >>> >>> Alim, thanks for helping out testing! >>> >>>> With this change HS200 mode does not work on exynos5800 peach-pi board. >>>> I got below error while testing this series: >>>> >>>> mmc0: tuning execution failed >>>> mmc0: error -5 whilst initialising MMC card >>>> >>>> Though, your's next branch with commit _a1d06b4_ works fine in HS200 mode. >>> >>> I was looking into the details of what change my patchset introduces >>> for dw_mmc-exynos. Apparently, dw_mmc-exynos was using the >>> MMC_STOP_TRANSMISSION to end the tuning reqeust (CMD21/19). The new >>> mmc_send_tuning() API doesn't, which also conforms to what the eMMC/SD >>> specifications states. >>> >>> Do you have any idea of why dw_mmc-exynos was using >>> MMC_STOP_TRANSMISSION for CMD19/21? >>> >>> To see if my theory is correct, could you try out the following patch >>> on top of $subject patch? >>> BTW, I have queued patch 1 and 2, from this patchset available on my >>> next branch. >>> >>> >>> From e1ac35bb0e90254275ec7088f41e6e2d59e48367 Mon Sep 17 00:00:00 2001 >>> From: Ulf Hansson >>> Date: Mon, 8 Dec 2014 10:58:48 +0100 >>> Subject: [PATCH] mmc: core: End tuning request with stop command >>> >>> Not to be merged! >>> >>> This patch adds the MMC_STOP_TRANSMISSION command to end a tuning >>> request. For some reason dw_mmc seems to need this to complete the >>> tuning request without errors. >>> >>> Signed-off-by: Ulf Hansson >>> --- >>> drivers/mmc/core/mmc_ops.c | 5 +++++ >>> 1 file changed, 5 insertions(+) >>> >>> diff --git a/drivers/mmc/core/mmc_ops.c b/drivers/mmc/core/mmc_ops.c >>> index 3b044c5..aa79e0c 100644 >>> --- a/drivers/mmc/core/mmc_ops.c >>> +++ b/drivers/mmc/core/mmc_ops.c >>> @@ -551,6 +551,7 @@ int mmc_send_tuning(struct mmc_host *host) >>> { >>> struct mmc_request mrq = {NULL}; >>> struct mmc_command cmd = {0}; >>> + struct mmc_command stop = {0}; >>> struct mmc_data data = {0}; >>> struct scatterlist sg; >>> struct mmc_ios *ios = &host->ios; >>> @@ -576,10 +577,14 @@ int mmc_send_tuning(struct mmc_host *host) >>> >>> mrq.cmd = &cmd; >>> mrq.data = &data; >>> + mrq.stop = &stop; >>> >>> cmd.opcode = opcode; >>> cmd.flags = MMC_RSP_R1 | MMC_CMD_ADTC; >>> >>> + stop.opcode = MMC_STOP_TRANSMISSION; >>> + stop.flags = MMC_RSP_R1B | MMC_CMD_AC; >>> + >>> data.blksz = size; >>> data.blocks = 1; >>> data.flags = MMC_DATA_READ; >>> -- >>> 1.9.1 >>> >> Sorry for delay in testing this suggested patch, I would say this >> certainly helps, but still I need to change sample phase to make it >> work with generic tuning function. >> So with your's adding STOP_TRANSMISSION command and below change, >> HS200 works well on exynos5800 >> >> diff --git a/arch/arm/boot/dts/exynos5800-peach-pi.dts >> b/arch/arm/boot/dts/exynos5800-peach-pi.dts >> index e8fdda8..e0f0337 100644 >> --- a/arch/arm/boot/dts/exynos5800-peach-pi.dts >> +++ b/arch/arm/boot/dts/exynos5800-peach-pi.dts >> @@ -555,7 +555,7 @@ >> card-detect-delay = <200>; >> clock-frequency = <400000000>; >> samsung,dw-mshc-ciu-div = <3>; >> - samsung,dw-mshc-sdr-timing = <0 4>; >> + samsung,dw-mshc-sdr-timing = <2 4>; >> >> This basically change the clock-sample phase with which the tuning >> process starts. >> >> I didn't find anything is exynos documentations which suggest >> STOP_TRANSMISSION is needed, may be Seungwon might be know as he wrote >> this. >> >> Are you ok to add STOP_TRANSMISSION in generic tuning code (probably >> with a quirk, so that other hosts are still happy)? > > No, I don't want that. > > If special treatment are needed for dw_mmc to handle a tuning request, > dw_mmc should be able to take care by itself. How about below patch, > do you think that could work? > > From c2e5783beaa322178f2336b4bff4bdf13da76f56 Mon Sep 17 00:00:00 2001 > From: Ulf Hansson > Date: Mon, 22 Dec 2014 15:24:14 +0100 > Subject: [PATCH] mmc: dw_mmc: Generate stop commands for failed tuning > requests > > It's seems like dw_mmc internal logic expects failed data transfers to > be ended using a stop command. Let the tuning requests also fall into > this category, since there are data transfer involved. > > Signed-off-by: Ulf Hansson > --- Yes, this works for me. So, Reviewed-by: Alim Akhtar Tested-by: Alim Akhtar > drivers/mmc/host/dw_mmc.c | 4 +++- > 1 file changed, 3 insertions(+), 1 deletion(-) > > diff --git a/drivers/mmc/host/dw_mmc.c b/drivers/mmc/host/dw_mmc.c > index a1b80e5..976382db 100644 > --- a/drivers/mmc/host/dw_mmc.c > +++ b/drivers/mmc/host/dw_mmc.c > @@ -314,7 +314,9 @@ static u32 dw_mci_prep_stop_abort(struct dw_mci > *host, struct mmc_command *cmd) > if (cmdr == MMC_READ_SINGLE_BLOCK || > cmdr == MMC_READ_MULTIPLE_BLOCK || > cmdr == MMC_WRITE_BLOCK || > - cmdr == MMC_WRITE_MULTIPLE_BLOCK) { > + cmdr == MMC_WRITE_MULTIPLE_BLOCK || > + cmdr == MMC_SEND_TUNING_BLOCK || > + cmdr == MMC_SEND_TUNING_BLOCK_HS200) { > stop->opcode = MMC_STOP_TRANSMISSION; > stop->arg = 0; > stop->flags = MMC_RSP_R1B | MMC_CMD_AC; > -- > 1.9.1 > >> If so, I can send the above DT change as a fix. > > Please do! > > Again, I appreciate all your efforts in helping out! > > Kind regards > Uffe -- Regards, Alim