linux-kernel - Re: [PATCH 4/4] dma: mxs-dma: Don't use CLKGATE bits in CTRL0 to disable DMA channels

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4EE096D5.4030005@freescale.com>
Date:	Thu, 8 Dec 2011 18:52:05 +0800
From:	Huang Shijie <b32955@...escale.com>
To:	Lothar Waßmann <LW@...O-electronics.de>
CC:	Vinod Koul <vinod.koul@...el.com>,
	Wolfram Sang <w.sang@...gutronix.de>,
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 4/4] dma: mxs-dma: Don't use CLKGATE bits in CTRL0 to
 disable DMA channels

Hi,

[1] i tested the cmd:
  #dd if=/dev/mtd0 > /dev/null & dd if=/dev/mmcblk0 > /dev/null

The `dd if=/dev/mtd0 > /dev/null`  finished successfully;
but `dd if=/dev/mmcblk0 > /dev/null ` will failed.

[2] So i only tested the SD card with:
   #dd if=/dev/mmcblk0 > /dev/null

   it failed too.

[3] I think you have fixed the bug.

Huang Shijie








> This is how the original Freescale code (unintentionally) worked,
> because the code path which would have asserted the CLKGATE bit was
> never actually reached in their code.
> This fixes the nefarious "DMA timout" bug when multiple DMA channels
> (e.g. GPMI NAND and MMC) are used at the same time.
> If a better fix for this problem should be found, the clkgate handling
> could be reinstated.
> See http://lists.infradead.org/pipermail/linux-arm-kernel/2011-September/065228.html
>
> Also reverse the order of mxs_dma_disable_chan() and
> mxs_dma_reset_chan() in mxs_dma_control() because mxs_dma_reset_chan()
> can only work when the DMA channel is enabled.
>
> Signed-off-by: Lothar Waßmann<LW@...O-electronics.de>
> ---
>   drivers/dma/mxs-dma.c |   29 +----------------------------
>   1 files changed, 1 insertions(+), 28 deletions(-)
>
> diff --git a/drivers/dma/mxs-dma.c b/drivers/dma/mxs-dma.c
> index 78336b0..71ba93a 100644
> --- a/drivers/dma/mxs-dma.c
> +++ b/drivers/dma/mxs-dma.c
> @@ -44,7 +44,6 @@
>   #define HW_APBHX_CTRL0				0x000
>   #define BM_APBH_CTRL0_APB_BURST8_EN		(1<<  29)
>   #define BM_APBH_CTRL0_APB_BURST_EN		(1<<  28)
> -#define BP_APBH_CTRL0_CLKGATE_CHANNEL		8
>   #define BP_APBH_CTRL0_RESET_CHANNEL		16
>   #define HW_APBHX_CTRL1				0x010
>   #define HW_APBHX_CTRL2				0x020
> @@ -131,23 +130,6 @@ struct mxs_dma_engine {
>   	struct mxs_dma_chan		mxs_chans[MXS_DMA_CHANNELS];
>   };
>
> -static inline void mxs_dma_clkgate(struct mxs_dma_chan *mxs_chan, int enable)
> -{
> -	struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;
> -	int chan_id = mxs_chan->chan.chan_id;
> -	int set_clr = enable ? MXS_CLR_ADDR : MXS_SET_ADDR;
> -
> -	/* enable apbh channel clock */
> -	if (dma_is_apbh()) {
> -		if (apbh_is_old())
> -			writel(1<<  (chan_id + BP_APBH_CTRL0_CLKGATE_CHANNEL),
> -				mxs_dma->base + HW_APBHX_CTRL0 + set_clr);
> -		else
> -			writel(1<<  chan_id,
> -				mxs_dma->base + HW_APBHX_CTRL0 + set_clr);
> -	}
> -}
> -
>   static void mxs_dma_reset_chan(struct mxs_dma_chan *mxs_chan)
>   {
>   	struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;
> @@ -166,9 +148,6 @@ static void mxs_dma_enable_chan(struct mxs_dma_chan *mxs_chan)
>   	struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;
>   	int chan_id = mxs_chan->chan.chan_id;
>
> -	/* clkgate needs to be enabled before writing other registers */
> -	mxs_dma_clkgate(mxs_chan, 1);
> -
>   	/* set cmd_addr up */
>   	writel(mxs_chan->ccw_phys,
>   		mxs_dma->base + HW_APBHX_CHn_NXTCMDAR(chan_id));
> @@ -179,9 +158,6 @@ static void mxs_dma_enable_chan(struct mxs_dma_chan *mxs_chan)
>
>   static void mxs_dma_disable_chan(struct mxs_dma_chan *mxs_chan)
>   {
> -	/* disable apbh channel clock */
> -	mxs_dma_clkgate(mxs_chan, 0);
> -
>   	mxs_chan->status = DMA_SUCCESS;
>   }
>
> @@ -339,10 +315,7 @@ static int mxs_dma_alloc_chan_resources(struct dma_chan *chan)
>   	if (ret)
>   		goto err_clk;
>
> -	/* clkgate needs to be enabled for reset to finish */
> -	mxs_dma_clkgate(mxs_chan, 1);
>   	mxs_dma_reset_chan(mxs_chan);
> -	mxs_dma_clkgate(mxs_chan, 0);
>
>   	dma_async_tx_descriptor_init(&mxs_chan->desc, chan);
>   	mxs_chan->desc.tx_submit = mxs_dma_tx_submit;
> @@ -542,8 +515,8 @@ static int mxs_dma_control(struct dma_chan *chan, enum dma_ctrl_cmd cmd,
>
>   	switch (cmd) {
>   	case DMA_TERMINATE_ALL:
> -		mxs_dma_disable_chan(mxs_chan);
>   		mxs_dma_reset_chan(mxs_chan);
> +		mxs_dma_disable_chan(mxs_chan);
>   		break;
>   	case DMA_PAUSE:
>   		mxs_dma_pause_chan(mxs_chan);


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/