In some cases we seem to submit two transactions in a row, which
causes us to lose track of the first. If we then cancel the
request, we may still get an interrupt, which traverses a null
ds_run value.
So try to avoid starting a new transaction if the ds_run value
is set.
While this patch avoids the null pointer crash, I've had some
reports of the k3dma driver still getting confused, which
suggests the ds_run/ds_done value handling still isn't quite
right. However, I've not run into an issue recently with it
so I think this patch is worth pushing upstream to avoid the
crash.
Cc: Vinod Koul <vkoul@xxxxxxxxxx>
Cc: Dan Williams <dan.j.williams@xxxxxxxxx>
Cc: ryan@xxxxxxxxx
Cc: aserbinski@xxxxxxxxx
Cc: dmaengine@xxxxxxxxxxxxxxx
Signed-off-by: John Stultz <john.stultz@xxxxxxxxxx>
---
drivers/dma/k3dma.c | 12 +++++++++---
1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/drivers/dma/k3dma.c b/drivers/dma/k3dma.c
index adecea51814f..c5c1aa0dcaed 100644
--- a/drivers/dma/k3dma.c
+++ b/drivers/dma/k3dma.c
@@ -229,9 +229,11 @@ static irqreturn_t k3_dma_int_handler(int irq, void *dev_id)
c = p->vchan;
if (c && (tc1 & BIT(i))) {
spin_lock_irqsave(&c->vc.lock, flags);
- vchan_cookie_complete(&p->ds_run->vd);
- p->ds_done = p->ds_run;
- p->ds_run = NULL;
+ if (p->ds_run != NULL) {
+ vchan_cookie_complete(&p->ds_run->vd);
+ p->ds_done = p->ds_run;
+ p->ds_run = NULL;
+ }
spin_unlock_irqrestore(&c->vc.lock, flags);
}
if (c && (tc2 & BIT(i))) {
@@ -271,6 +273,10 @@ static int k3_dma_start_txd(struct k3_dma_chan *c)
if (BIT(c->phy->idx) & k3_dma_get_chan_stat(d))
return -EAGAIN;
+ /* Avoid losing track of ds_run if a transaction is in flight */
+ if (c->phy->ds_run)
+ return -EAGAIN;
+
if (vd) {
struct k3_dma_desc_sw *ds =
container_of(vd, struct k3_dma_desc_sw, vd);
--
2.17.1
