From: Jie Hai <haijie1@huawei.com>
To: <vkoul@kernel.org>
Cc: <dmaengine@vger.kernel.org>, <linux-kernel@vger.kernel.org>
Subject: [PATCH] dmaengine: virt-dma : fix vchan error on multi-thread
Date: Thu, 20 Jul 2023 19:42:12 +0800 [thread overview]
Message-ID: <20230720114212.51224-1-haijie1@huawei.com> (raw)
List desc_allocated was introduced for the case of a transfer
submitted multiple times. But elegating descriptors on the list
causes other problems.
For example, in the multi-thread scenario, which tasks are
continuously created and submitted by each thread. If one of
the threads calls dmaengine_terminate_all, for dirvers using
vchan_get_all_descriptors, all descriptors will be freed. If
there's another thread submitting a transfer A by
vchan_tx_submit, the following results may be generated:
1. desc A is freeing -> visit wrong address of node prep/next.
2. desc A is freed -> visit invalid address of A.
In the above case, calltrace is generated and the system is
suspended. This can be tested by dmatest.
This patch removes desc_allocated from vchan_get_all_descriptors,
and add new function 'vchan_get_all_allocated_descs' to get all
descriptors ever allocated.
And apply vchan_get_all_allocated_descs to free chan resource and
vchan_get_all_descriptors to terminate all transfers, respectively.
This avoids freeing up descriptors in use by other threads.
Signed-off-by: Jie Hai <haijie1@huawei.com>
---
drivers/dma/fsl-edma-common.c | 2 +-
drivers/dma/fsl-qdma.c | 2 +-
drivers/dma/sf-pdma/sf-pdma.c | 2 +-
drivers/dma/virt-dma.h | 20 ++++++++++++++++++--
4 files changed, 21 insertions(+), 5 deletions(-)
diff --git a/drivers/dma/fsl-edma-common.c b/drivers/dma/fsl-edma-common.c
index a06a1575a2a5..c6d2e54ab85d 100644
--- a/drivers/dma/fsl-edma-common.c
+++ b/drivers/dma/fsl-edma-common.c
@@ -674,7 +674,7 @@ void fsl_edma_free_chan_resources(struct dma_chan *chan)
if (edma->drvdata->dmamuxs)
fsl_edma_chan_mux(fsl_chan, 0, false);
fsl_chan->edesc = NULL;
- vchan_get_all_descriptors(&fsl_chan->vchan, &head);
+ vchan_get_all_allocated_descs(&fsl_chan->vchan, &head);
fsl_edma_unprep_slave_dma(fsl_chan);
spin_unlock_irqrestore(&fsl_chan->vchan.lock, flags);
diff --git a/drivers/dma/fsl-qdma.c b/drivers/dma/fsl-qdma.c
index eddb2688f234..5ffd7ba92058 100644
--- a/drivers/dma/fsl-qdma.c
+++ b/drivers/dma/fsl-qdma.c
@@ -311,7 +311,7 @@ static void fsl_qdma_free_chan_resources(struct dma_chan *chan)
LIST_HEAD(head);
spin_lock_irqsave(&fsl_chan->vchan.lock, flags);
- vchan_get_all_descriptors(&fsl_chan->vchan, &head);
+ vchan_get_all_allocated_descs(&fsl_chan->vchan, &head);
spin_unlock_irqrestore(&fsl_chan->vchan.lock, flags);
vchan_dma_desc_free_list(&fsl_chan->vchan, &head);
diff --git a/drivers/dma/sf-pdma/sf-pdma.c b/drivers/dma/sf-pdma/sf-pdma.c
index d1c6956af452..f35dc68e1a7c 100644
--- a/drivers/dma/sf-pdma/sf-pdma.c
+++ b/drivers/dma/sf-pdma/sf-pdma.c
@@ -144,7 +144,7 @@ static void sf_pdma_free_chan_resources(struct dma_chan *dchan)
sf_pdma_disable_request(chan);
kfree(chan->desc);
chan->desc = NULL;
- vchan_get_all_descriptors(&chan->vchan, &head);
+ vchan_get_all_allocated_descs(&chan->vchan, &head);
sf_pdma_disclaim_chan(chan);
spin_unlock_irqrestore(&chan->vchan.lock, flags);
vchan_dma_desc_free_list(&chan->vchan, &head);
diff --git a/drivers/dma/virt-dma.h b/drivers/dma/virt-dma.h
index e9f5250fbe4d..65b4f3bdecf7 100644
--- a/drivers/dma/virt-dma.h
+++ b/drivers/dma/virt-dma.h
@@ -177,13 +177,29 @@ static inline struct virt_dma_desc *vchan_next_desc(struct virt_dma_chan *vc)
static inline void vchan_get_all_descriptors(struct virt_dma_chan *vc,
struct list_head *head)
{
- list_splice_tail_init(&vc->desc_allocated, head);
list_splice_tail_init(&vc->desc_submitted, head);
list_splice_tail_init(&vc->desc_issued, head);
list_splice_tail_init(&vc->desc_completed, head);
list_splice_tail_init(&vc->desc_terminated, head);
}
+/**
+ * vchan_get_all_allocated_descs - obtain all descriptors
+ * @vc: virtual channel to get descriptors from
+ * @head: list of descriptors found
+ *
+ * vc.lock must be held by caller
+ *
+ * Removes all descriptors from internal lists, and provides a list of all
+ * descriptors found
+ */
+static inline void vchan_get_all_allocated_descs(struct virt_dma_chan *vc,
+ struct list_head *head)
+{
+ list_splice_tail_init(&vc->desc_allocated, head);
+ vchan_get_all_descriptors(vc, head);
+}
+
static inline void vchan_free_chan_resources(struct virt_dma_chan *vc)
{
struct virt_dma_desc *vd;
@@ -191,7 +207,7 @@ static inline void vchan_free_chan_resources(struct virt_dma_chan *vc)
LIST_HEAD(head);
spin_lock_irqsave(&vc->lock, flags);
- vchan_get_all_descriptors(vc, &head);
+ vchan_get_all_allocated_descs(vc, &head);
list_for_each_entry(vd, &head, node)
dmaengine_desc_clear_reuse(&vd->tx);
spin_unlock_irqrestore(&vc->lock, flags);
--
2.33.0
next reply other threads:[~2023-07-20 11:45 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-20 11:42 Jie Hai [this message]
2023-09-25 1:09 ` [PATCH] dmaengine: virt-dma : fix vchan error on multi-thread Jie Hai
2023-12-08 1:52 ` Jie Hai
2024-01-31 1:32 ` Jie Hai
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230720114212.51224-1-haijie1@huawei.com \
--to=haijie1@huawei.com \
--cc=dmaengine@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=vkoul@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).