From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?ISO-8859-1?Q?Per_F=F6rlin?= Subject: Re: [PATCH] mmc: mmc_test: align max_seg_size Date: Fri, 18 Nov 2011 10:25:20 +0100 Message-ID: <4EC62480.2000308@stericsson.com> References: <1321268664-22006-1-git-send-email-per.forlin@stericsson.com> <4EC485E8.4040706@codeaurora.org> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Return-path: Received: from eu1sys200aog113.obsmtp.com ([207.126.144.135]:48757 "EHLO eu1sys200aog113.obsmtp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754292Ab1KRJ0U (ORCPT ); Fri, 18 Nov 2011 04:26:20 -0500 In-Reply-To: <4EC485E8.4040706@codeaurora.org> Sender: linux-mmc-owner@vger.kernel.org List-Id: linux-mmc@vger.kernel.org To: Sujit Reddy Thumma Cc: "linux-mmc@vger.kernel.org" , Ulf HANSSON , Linus Walleij , Chris Ball On 11/17/2011 04:56 AM, Sujit Reddy Thumma wrote: > On 11/14/2011 4:34 PM, Per Forlin wrote: >> If max_seg_size is unaligned, mmc_test_map_sg() may create sg element >> sizes that are not aligned with 512 byte. Fix, align max_seg_size at >> mmc_test_area_init(). >> >> Signed-off-by: Per Forlin >> --- >> drivers/mmc/card/mmc_test.c | 1 + >> 1 files changed, 1 insertions(+), 0 deletions(-) >> >> diff --git a/drivers/mmc/card/mmc_test.c b/drivers/mmc/card/mmc_test.c >> index b038c4a..5848998 100644 >> --- a/drivers/mmc/card/mmc_test.c >> +++ b/drivers/mmc/card/mmc_test.c >> @@ -1581,6 +1581,7 @@ static int mmc_test_area_init(struct mmc_test_card *test, int erase, int fill) >> >> t->max_segs = test->card->host->max_segs; >> t->max_seg_sz = test->card->host->max_seg_size; >> + t->max_seg_sz -= t->max_seg_sz % 512; > > Shouldn't we align this to host->max_blk_size supported by controller? It's up to each test case to set the test transfer size. It's not necessary to align max_size but it wouldn't make any harm either. All tests use a multiple of 512 blocks, but this may be changed of course in the future to support large blocks. The max_seg_sz must be aligned otherwise the host driver may get sg-elements that are hard to handle, aligned to 1 byte in the worst case. The test case don't control how the transfer is sg-mapped. /Per