LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Juergen Borleis <jbe@pengutronix.de>
To: linux-mmc@vger.kernel.org
Cc: linux-kernel@vger.kernel.org, festevam@gmail.com,
	wsa+renesas@sang-engineering.com, dianders@chromium.org,
	ulf.hansson@linaro.org, kernel@pengutronix.de
Subject: [PATCH 2/2] mmc: mxcmmc: don't expect a DMA operation if DMA seems present
Date: Thu, 29 Jul 2021 09:29:29 +0200	[thread overview]
Message-ID: <20210729072929.10267-3-jbe@pengutronix.de> (raw)
In-Reply-To: <20210729072929.10267-1-jbe@pengutronix.de>

Change the driver's default behaviour from DMA to PIO for each request.
Preparing DMA can fail or be prevented by other causes. Switch to a DMA
operation only, if it is really possible.

This avoids a NULL pointer dereference on shutdown in mxcmci_start_cmd()
where it tries to store a DMA callback configuration into mxcmci_host::desc
which isn't setup a this point of time.

Signed-off-by: Juergen Borleis <jbe@pengutronix.de>
---
 drivers/mmc/host/mxcmmc.c | 24 ++++++++++++++----------
 1 file changed, 14 insertions(+), 10 deletions(-)

diff --git a/drivers/mmc/host/mxcmmc.c b/drivers/mmc/host/mxcmmc.c
index 611f827..41feea9 100644
--- a/drivers/mmc/host/mxcmmc.c
+++ b/drivers/mmc/host/mxcmmc.c
@@ -293,14 +293,13 @@ static int mxcmci_setup_data(struct mxcmci_host *host, struct mmc_data *data)
 	mxcmci_writew(host, blksz, MMC_REG_BLK_LEN);
 	host->datasize = datasize;
 
-	if (!mxcmci_use_dma(host))
-		return 0;
+	if (host->dma == NULL)
+		return 0; /* Keep PIO */
 
+	/* Avoid the use of DMA on short transfers, e.g. non-sectors for example */
 	for_each_sg(data->sg, sg, data->sg_len, i) {
-		if (sg->offset & 3 || sg->length & 3 || sg->length < 512) {
-			host->do_dma = 0;
-			return 0;
-		}
+		if (sg->offset & 3 || sg->length & 3 || sg->length < 512)
+			return 0; /* Keep PIO */
 	}
 
 	if (data->flags & MMC_DATA_READ) {
@@ -325,9 +324,11 @@ static int mxcmci_setup_data(struct mxcmci_host *host, struct mmc_data *data)
 	if (!host->desc) {
 		dma_unmap_sg(host->dma->device->dev, data->sg, data->sg_len,
 				host->dma_dir);
-		host->do_dma = 0;
-		return 0; /* Fall back to PIO */
+		return 0; /* Keep PIO */
 	}
+
+	/* DMA is possible */
+	host->do_dma = 1;
 	wmb();
 
 	dmaengine_submit(host->desc);
@@ -747,8 +748,11 @@ static void mxcmci_request(struct mmc_host *mmc, struct mmc_request *req)
 	host->req = req;
 	host->cmdat &= ~CMD_DAT_CONT_INIT;
 
-	if (host->dma)
-		host->do_dma = 1;
+	/*
+	 * Default always to PIO. DMA will be enabled in
+	 * mxcmci_setup_data() if possible
+	 */
+	host->do_dma = 0;
 
 	if (req->data) {
 		error = mxcmci_setup_data(host, req->data);
-- 
2.20.1


      parent reply	other threads:[~2021-07-29  7:29 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-07-29  7:29 Make the i.MX MMC device using DMA again Juergen Borleis
2021-07-29  7:29 ` [PATCH 1/2] mmc: mxcmmc: really configure the DMA on demand Juergen Borleis
2021-08-11 11:36   ` Ulf Hansson
2021-07-29  7:29 ` Juergen Borleis [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210729072929.10267-3-jbe@pengutronix.de \
    --to=jbe@pengutronix.de \
    --cc=dianders@chromium.org \
    --cc=festevam@gmail.com \
    --cc=kernel@pengutronix.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mmc@vger.kernel.org \
    --cc=ulf.hansson@linaro.org \
    --cc=wsa+renesas@sang-engineering.com \
    --subject='Re: [PATCH 2/2] mmc: mxcmmc: don'\''t expect a DMA operation if DMA seems present' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).