Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1483154 > unrolled thread

[PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data

Started byPeter Ujfalusi <peter.ujfalusi@ti.com>
First post2016-09-14 12:20 +0200
Last post2016-09-14 13:00 +0200
Articles 4 — 3 participants

Back to article view | Back to linux.kernel

Contents

  [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data Peter Ujfalusi <peter.ujfalusi@ti.com> - 2016-09-14 12:20 +0200
    Re: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data Arnd Bergmann <arnd@arndb.de> - 2016-09-14 12:30 +0200
      Re: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid  random data Peter Ujfalusi <peter.ujfalusi@ti.com> - 2016-09-14 13:00 +0200
      Re: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid  random data Ulf Hansson <ulf.hansson@linaro.org> - 2016-09-14 13:00 +0200

#1483154 — [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data

FromPeter Ujfalusi <peter.ujfalusi@ti.com>
Date2016-09-14 12:20 +0200
Subject[PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data
Message-ID<shfsm-830-17@gated-at.bofh.it>
It is wrong to not initialize the dma_slave_config struct as the DMAengine
driver might look at non initialized (random data) fields and tries to
interpret it.

Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
---
Hi,

would it be possible to send this patch for 4.8? I have omap-dma DMAengine
driver changes pending, but they break the MMC/SD because of the uninitialized
fields in dma_slave_config.

Thanks,
Peter

 drivers/mmc/host/omap_hsmmc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c
index 24ebc9a8de89..0a5640156159 100644
--- a/drivers/mmc/host/omap_hsmmc.c
+++ b/drivers/mmc/host/omap_hsmmc.c
@@ -1409,7 +1409,7 @@ static int omap_hsmmc_pre_dma_transfer(struct omap_hsmmc_host *host,
 static int omap_hsmmc_setup_dma_transfer(struct omap_hsmmc_host *host,
 					struct mmc_request *req)
 {
-	struct dma_slave_config cfg;
+	struct dma_slave_config cfg = {};
 	struct dma_async_tx_descriptor *tx;
 	int ret = 0, i;
 	struct mmc_data *data = req->data;
--
2.10.0

[toc] | [next] | [standalone]

#1483157

FromArnd Bergmann <arnd@arndb.de>
Date2016-09-14 12:30 +0200
Message-ID<shfC1-86s-27@gated-at.bofh.it>
In reply to#1483154
On Wednesday, September 14, 2016 1:10:48 PM CEST Peter Ujfalusi wrote:
> diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c
> index 24ebc9a8de89..0a5640156159 100644
> --- a/drivers/mmc/host/omap_hsmmc.c
> +++ b/drivers/mmc/host/omap_hsmmc.c
> @@ -1409,7 +1409,7 @@ static int omap_hsmmc_pre_dma_transfer(struct omap_hsmmc_host *host,
>  static int omap_hsmmc_setup_dma_transfer(struct omap_hsmmc_host *host,
>                                         struct mmc_request *req)
>  {
> -       struct dma_slave_config cfg;
> +       struct dma_slave_config cfg = {};
>         struct dma_async_tx_descriptor *tx;
>         int ret = 0, i;
>         struct mmc_data *data = req->data;
> 

This change looks correct and is certainly the simplest solution if you want to
get it into v4.8, but generally speaking, I think it would be nicer to combine
it with the initialization of the fields:

	struct dma_slave_config cfg = {
		  .src_addr = host->mapbase + OMAP_HSMMC_DATA;
        .dst_addr = host->mapbase + OMAP_HSMMC_DATA;
        .src_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
        .dst_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
        .src_maxburst = data->blksz / 4;
        .dst_maxburst = data->blksz / 4;
	};

	Arnd

[toc] | [prev] | [next] | [standalone]

#1483175 — Re: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data

FromPeter Ujfalusi <peter.ujfalusi@ti.com>
Date2016-09-14 13:00 +0200
SubjectRe: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data
Message-ID<shg54-8gk-13@gated-at.bofh.it>
In reply to#1483157
On 09/14/16 13:23, Arnd Bergmann wrote:
> On Wednesday, September 14, 2016 1:10:48 PM CEST Peter Ujfalusi wrote:
>> diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c
>> index 24ebc9a8de89..0a5640156159 100644
>> --- a/drivers/mmc/host/omap_hsmmc.c
>> +++ b/drivers/mmc/host/omap_hsmmc.c
>> @@ -1409,7 +1409,7 @@ static int omap_hsmmc_pre_dma_transfer(struct omap_hsmmc_host *host,
>>  static int omap_hsmmc_setup_dma_transfer(struct omap_hsmmc_host *host,
>>                                         struct mmc_request *req)
>>  {
>> -       struct dma_slave_config cfg;
>> +       struct dma_slave_config cfg = {};
>>         struct dma_async_tx_descriptor *tx;
>>         int ret = 0, i;
>>         struct mmc_data *data = req->data;
>>
> 
> This change looks correct and is certainly the simplest solution if you want to
> get it into v4.8, but generally speaking, I think it would be nicer to combine
> it with the initialization of the fields:
> 
> 	struct dma_slave_config cfg = {
> 		  .src_addr = host->mapbase + OMAP_HSMMC_DATA;
>         .dst_addr = host->mapbase + OMAP_HSMMC_DATA;
>         .src_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
>         .dst_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
>         .src_maxburst = data->blksz / 4;
>         .dst_maxburst = data->blksz / 4;

I opted to not do this because the data->blksz is validated later in the
function (twice).

It is not a big issue, but looks a bit odd.

-- 
Péter

[toc] | [prev] | [next] | [standalone]

#1483179 — Re: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data

FromUlf Hansson <ulf.hansson@linaro.org>
Date2016-09-14 13:00 +0200
SubjectRe: [PATCH] mmc: omap_hsmmc: Initialize dma_slave_config to avoid random data
Message-ID<shg54-8gk-11@gated-at.bofh.it>
In reply to#1483157
On 14 September 2016 at 12:23, Arnd Bergmann <arnd@arndb.de> wrote:
> On Wednesday, September 14, 2016 1:10:48 PM CEST Peter Ujfalusi wrote:
>> diff --git a/drivers/mmc/host/omap_hsmmc.c b/drivers/mmc/host/omap_hsmmc.c
>> index 24ebc9a8de89..0a5640156159 100644
>> --- a/drivers/mmc/host/omap_hsmmc.c
>> +++ b/drivers/mmc/host/omap_hsmmc.c
>> @@ -1409,7 +1409,7 @@ static int omap_hsmmc_pre_dma_transfer(struct omap_hsmmc_host *host,
>>  static int omap_hsmmc_setup_dma_transfer(struct omap_hsmmc_host *host,
>>                                         struct mmc_request *req)
>>  {
>> -       struct dma_slave_config cfg;
>> +       struct dma_slave_config cfg = {};
>>         struct dma_async_tx_descriptor *tx;
>>         int ret = 0, i;
>>         struct mmc_data *data = req->data;
>>
>
> This change looks correct and is certainly the simplest solution if you want to
> get it into v4.8, but generally speaking, I think it would be nicer to combine
> it with the initialization of the fields:
>
>         struct dma_slave_config cfg = {
>                   .src_addr = host->mapbase + OMAP_HSMMC_DATA;
>         .dst_addr = host->mapbase + OMAP_HSMMC_DATA;
>         .src_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
>         .dst_addr_width = DMA_SLAVE_BUSWIDTH_4_BYTES;
>         .src_maxburst = data->blksz / 4;
>         .dst_maxburst = data->blksz / 4;
>         };

This seems like a small an easy change as well, so I wouldn't mind
picking up a v2 adopting this pattern instead.

Kind regards
Uffe

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web