Skip to content

Commit

Permalink
dmaengine: qcom: bam_dma: Fix resource leak
Browse files Browse the repository at this point in the history
bam_dma_terminate_all() will leak resources if any of the transactions are
committed to the hardware (present in the desc fifo), and not complete.
Since bam_dma_terminate_all() does not cause the hardware to be updated,
the hardware will still operate on any previously committed transactions.
This can cause memory corruption if the memory for the transaction has been
reassigned, and will cause a sync issue between the BAM and its client(s).

Fix this by properly updating the hardware in bam_dma_terminate_all().

Fixes: e7c0fe2 ("dmaengine: add Qualcomm BAM dma driver")
Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@gmail.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20191017152606.34120-1-jeffrey.l.hugo@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
  • Loading branch information
Jeffrey Hugo authored and Vinod Koul committed Oct 18, 2019
1 parent ec1ac30 commit 7667819
Showing 1 changed file with 19 additions and 0 deletions.
19 changes: 19 additions & 0 deletions drivers/dma/qcom/bam_dma.c
Original file line number Diff line number Diff line change
Expand Up @@ -694,6 +694,25 @@ static int bam_dma_terminate_all(struct dma_chan *chan)

/* remove all transactions, including active transaction */
spin_lock_irqsave(&bchan->vc.lock, flag);
/*
* If we have transactions queued, then some might be committed to the
* hardware in the desc fifo. The only way to reset the desc fifo is
* to do a hardware reset (either by pipe or the entire block).
* bam_chan_init_hw() will trigger a pipe reset, and also reinit the
* pipe. If the pipe is left disabled (default state after pipe reset)
* and is accessed by a connected hardware engine, a fatal error in
* the BAM will occur. There is a small window where this could happen
* with bam_chan_init_hw(), but it is assumed that the caller has
* stopped activity on any attached hardware engine. Make sure to do
* this first so that the BAM hardware doesn't cause memory corruption
* by accessing freed resources.
*/
if (!list_empty(&bchan->desc_list)) {
async_desc = list_first_entry(&bchan->desc_list,
struct bam_async_desc, desc_node);
bam_chan_init_hw(bchan, async_desc->dir);
}

list_for_each_entry_safe(async_desc, tmp,
&bchan->desc_list, desc_node) {
list_add(&async_desc->vd.node, &bchan->vc.desc_issued);
Expand Down

0 comments on commit 7667819

Please sign in to comment.