Hi Morimoto-san,
> Current rcar-dmac driver is using spin_lock_irq() / spin_unlock_irq()
> in some functions. But, some other driver might call DMAEngine API
> during interrupt disabled. In such case, rcar-dmac side spin_unlock_irq()
> forcefully allows all interrupts. Therefore, other driver receives
> unexpected interruption, and its exclusive access control will be broken.
> This patch replaces spin_lock_irq() to spin_lock_irqsave(),
> and spin_unlock_irq() to spin_unlock_irqrestore().
>
> Signed-off-by: Jun Watanabe <jun.watanabe.ue@xxxxxxxxxxx>
> Signed-off-by: Hiroyuki Yokoyama <hiroyuki.yokoyama.vx@xxxxxxxxxxxx>
> Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@xxxxxxxxxxx>
< snip >
> @@ -493,15 +494,18 @@ static int rcar_dmac_desc_alloc(struct rcar_dmac_chan *chan, gfp_t gfp)
> static void rcar_dmac_desc_put(struct rcar_dmac_chan *chan,
> struct rcar_dmac_desc *desc)
> {
> - spin_lock_irq(&chan->lock);
> + unsigned long flags;
> +
> + spin_lock_irqsave(&chan->lock, flags);
> list_splice_tail_init(&desc->chunks, &chan->desc.chunks_free);
> list_add_tail(&desc->node, &chan->desc.free);
> - spin_unlock_irq(&chan->lock);
> + spin_unlock_irqrestore(&chan->lock, flags);
> }
This change is already submitted from Laurent-san:
http://thread.gmane.org/gmane.linux.ports.sh.devel/43446/focus=43451
< snip >
> @@ -542,12 +546,13 @@ static void rcar_dmac_desc_recycle_acked(struct rcar_dmac_chan *chan)
> static struct rcar_dmac_desc *rcar_dmac_desc_get(struct rcar_dmac_chan *chan)
> {
> struct rcar_dmac_desc *desc;
> + unsigned long flags;
> int ret;
>
> /* Recycle acked descriptors before attempting allocation. */
> rcar_dmac_desc_recycle_acked(chan);
>
> - spin_lock_irq(&chan->lock);
> + spin_lock_irqsave(&chan->lock, flags);
>
> do {
> if (list_empty(&chan->desc.free)) {
I guess this will conflict the following patch.
But, I'm not sure how to fix conflict with other patch:
http://thread.gmane.org/gmane.linux.ports.sh.devel/43446/focus=43447
< snip >
> @@ -617,9 +623,10 @@ static struct rcar_dmac_xfer_chunk *
> rcar_dmac_xfer_chunk_get(struct rcar_dmac_chan *chan)
> {
> struct rcar_dmac_xfer_chunk *chunk;
> + unsigned long flags;
> int ret;
>
> - spin_lock_irq(&chan->lock);
> + spin_lock_irqsave(&chan->lock, flags);
>
> do {
> if (list_empty(&chan->desc.chunks_free)) {
Same the above.
< snip >
> @@ -1349,8 +1357,9 @@ static irqreturn_t rcar_dmac_isr_channel_thread(int irq, void *dev)
> {
> struct rcar_dmac_chan *chan = dev;
> struct rcar_dmac_desc *desc;
> + unsigned long flags;
>
> - spin_lock_irq(&chan->lock);
> + spin_lock_irqsave(&chan->lock, flags);
Since this function is not dmaengine API, I don't think we need this change.
Best regards,
Yoshihiro Shimoda
--
To unsubscribe from this list: send the line "unsubscribe dmaengine" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
