| Message ID | 2eecd528e85377f03e6fbc5e7d6544b9c9f59cb1.1555330115.git.baolin.wang@linaro.org (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | Fix some bugs and add new feature for Spreadtrum DMA engine | expand |
On 15-04-19, 20:14, Baolin Wang wrote: > We will get a NULL virtual descriptor by vchan_find_desc() when the descriptor > has been submitted, that will crash the kernel when getting the engine status. No that is wrong, status is for descriptor and not engine! > In this case, since the descriptor has been submitted, which means the pointer > 'schan->cur_desc' will point to the current descriptor, then we can use > 'schan->cur_desc' to get the engine status to avoid this issue. Nope, since the descriptor is completed, you return with residue as 0 and DMA_COMPLETE status! > > Signed-off-by: Baolin Wang <baolin.wang@linaro.org> > --- > drivers/dma/sprd-dma.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/dma/sprd-dma.c b/drivers/dma/sprd-dma.c > index 48431e2..e29342a 100644 > --- a/drivers/dma/sprd-dma.c > +++ b/drivers/dma/sprd-dma.c > @@ -625,7 +625,7 @@ static enum dma_status sprd_dma_tx_status(struct dma_chan *chan, > else > pos = 0; > } else if (schan->cur_desc && schan->cur_desc->vd.tx.cookie == cookie) { > - struct sprd_dma_desc *sdesc = to_sprd_dma_desc(vd); > + struct sprd_dma_desc *sdesc = schan->cur_desc; > > if (sdesc->dir == DMA_DEV_TO_MEM) > pos = sprd_dma_get_dst_addr(schan); > -- > 1.7.9.5
Hi Vinod, On Mon, 29 Apr 2019 at 19:36, Vinod Koul <vkoul@kernel.org> wrote: > > On 15-04-19, 20:14, Baolin Wang wrote: > > We will get a NULL virtual descriptor by vchan_find_desc() when the descriptor > > has been submitted, that will crash the kernel when getting the engine status. > > No that is wrong, status is for descriptor and not engine! Sure, will fix the commit message. > > > In this case, since the descriptor has been submitted, which means the pointer > > 'schan->cur_desc' will point to the current descriptor, then we can use > > 'schan->cur_desc' to get the engine status to avoid this issue. > > Nope, since the descriptor is completed, you return with residue as 0 > and DMA_COMPLETE status! No, the descriptor is not completed now. If it is completed, we will return 0 with DMA_COMPLETE status. But now the descriptor is on progress, we should get the descriptor to return current residue. Sorry for confusing description. > > > > > Signed-off-by: Baolin Wang <baolin.wang@linaro.org> > > --- > > drivers/dma/sprd-dma.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/drivers/dma/sprd-dma.c b/drivers/dma/sprd-dma.c > > index 48431e2..e29342a 100644 > > --- a/drivers/dma/sprd-dma.c > > +++ b/drivers/dma/sprd-dma.c > > @@ -625,7 +625,7 @@ static enum dma_status sprd_dma_tx_status(struct dma_chan *chan, > > else > > pos = 0; > > } else if (schan->cur_desc && schan->cur_desc->vd.tx.cookie == cookie) { > > - struct sprd_dma_desc *sdesc = to_sprd_dma_desc(vd); > > + struct sprd_dma_desc *sdesc = schan->cur_desc; > > > > if (sdesc->dir == DMA_DEV_TO_MEM) > > pos = sprd_dma_get_dst_addr(schan); > > -- > > 1.7.9.5 > > -- > ~Vinod
On 29-04-19, 19:49, Baolin Wang wrote: > Hi Vinod, > > On Mon, 29 Apr 2019 at 19:36, Vinod Koul <vkoul@kernel.org> wrote: > > > > On 15-04-19, 20:14, Baolin Wang wrote: > > > We will get a NULL virtual descriptor by vchan_find_desc() when the descriptor > > > has been submitted, that will crash the kernel when getting the engine status. > > > > No that is wrong, status is for descriptor and not engine! > > Sure, will fix the commit message. > > > > > > In this case, since the descriptor has been submitted, which means the pointer > > > 'schan->cur_desc' will point to the current descriptor, then we can use > > > 'schan->cur_desc' to get the engine status to avoid this issue. > > > > Nope, since the descriptor is completed, you return with residue as 0 > > and DMA_COMPLETE status! > > No, the descriptor is not completed now. If it is completed, we will > return 0 with DMA_COMPLETE status. But now the descriptor is on > progress, we should get the descriptor to return current residue. > Sorry for confusing description. OKay will wait for updated description to understand the fix > > > > > > > > > Signed-off-by: Baolin Wang <baolin.wang@linaro.org> > > > --- > > > drivers/dma/sprd-dma.c | 2 +- > > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > > > diff --git a/drivers/dma/sprd-dma.c b/drivers/dma/sprd-dma.c > > > index 48431e2..e29342a 100644 > > > --- a/drivers/dma/sprd-dma.c > > > +++ b/drivers/dma/sprd-dma.c > > > @@ -625,7 +625,7 @@ static enum dma_status sprd_dma_tx_status(struct dma_chan *chan, > > > else > > > pos = 0; > > > } else if (schan->cur_desc && schan->cur_desc->vd.tx.cookie == cookie) { > > > - struct sprd_dma_desc *sdesc = to_sprd_dma_desc(vd); > > > + struct sprd_dma_desc *sdesc = schan->cur_desc; > > > > > > if (sdesc->dir == DMA_DEV_TO_MEM) > > > pos = sprd_dma_get_dst_addr(schan); > > > -- > > > 1.7.9.5 > > > > -- > > ~Vinod > > > > -- > Baolin Wang > Best Regards
diff --git a/drivers/dma/sprd-dma.c b/drivers/dma/sprd-dma.c index 48431e2..e29342a 100644 --- a/drivers/dma/sprd-dma.c +++ b/drivers/dma/sprd-dma.c @@ -625,7 +625,7 @@ static enum dma_status sprd_dma_tx_status(struct dma_chan *chan, else pos = 0; } else if (schan->cur_desc && schan->cur_desc->vd.tx.cookie == cookie) { - struct sprd_dma_desc *sdesc = to_sprd_dma_desc(vd); + struct sprd_dma_desc *sdesc = schan->cur_desc; if (sdesc->dir == DMA_DEV_TO_MEM) pos = sprd_dma_get_dst_addr(schan);
We will get a NULL virtual descriptor by vchan_find_desc() when the descriptor has been submitted, that will crash the kernel when getting the engine status. In this case, since the descriptor has been submitted, which means the pointer 'schan->cur_desc' will point to the current descriptor, then we can use 'schan->cur_desc' to get the engine status to avoid this issue. Signed-off-by: Baolin Wang <baolin.wang@linaro.org> --- drivers/dma/sprd-dma.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)