From: Frank Li <Frank.li@oss.nxp.com>
To: Linus Walleij <linusw@kernel.org>
Cc: Vinod Koul <vkoul@kernel.org>, Frank Li <Frank.Li@kernel.org>,
linux-arm-kernel@lists.infradead.org, dmaengine@vger.kernel.org,
Ulf Hansson <ulfh@kernel.org>
Subject: Re: [PATCH] dmaengine: ste_dma40: Keep prepared descriptors across termination
Date: Thu, 12 Nov 2026 14:59:38 -0600 [thread overview]
Message-ID: <avYourwIh_YGRjcH@SMW015318> (raw)
In-Reply-To: <20261104-dma40-fixes-v1-1-48c5180f5210@kernel.org>
On Wed, Nov 04, 2026 at 10:28:02PM +0100, Linus Walleij wrote:
> MMCI can prepare the next request while the current DMA transfer is
> running. If the current request fails, its error handler terminates the
> channel before the next descriptor has been submitted.
>
> d40_term_all() frees that descriptor from prepare_queue, although MMCI
> still holds its pointer. Submitting it later accesses freed memory in
> d40_tx_submit(). A reported crash writes to 0x104 through a list pointer
> containing LIST_POISON1, which is consistent with this sequence.
>
> Termination only cancels transfers that have been submitted. Leave
> prepared descriptors on prepare_queue so the client can still submit
> them, and release any remaining prepared descriptors when the channel
> resources are freed. Keep tracking them so abandoned descriptors are
> reclaimed at channel teardown.
>
> Fixes: 82babbb361f2 ("dmaengine/ste_dma40: fix memory leak due to prepared descriptors")
> Signed-off-by: Linus Walleij <linusw@kernel.org>
> ---
Reviewed-by: Frank Li <Frank.Li@nxp.com>
> drivers/dma/ste_dma40.c | 22 +++++++++-------------
> 1 file changed, 9 insertions(+), 13 deletions(-)
>
> diff --git a/drivers/dma/ste_dma40.c b/drivers/dma/ste_dma40.c
> index 74dd4cbdc3c0..f1cbc719c4f0 100644
> --- a/drivers/dma/ste_dma40.c
> +++ b/drivers/dma/ste_dma40.c
> @@ -1166,14 +1166,6 @@ static void d40_term_all(struct d40_chan *d40c)
> d40_desc_free(d40c, d40d);
> }
>
> - /* Release descriptors in prepare queue */
> - if (!list_empty(&d40c->prepare_queue))
> - list_for_each_entry_safe(d40d, _d,
> - &d40c->prepare_queue, node) {
> - d40_desc_remove(d40d);
> - d40_desc_free(d40c, d40d);
> - }
> -
> d40c->pending_tx = 0;
> }
>
> @@ -2222,7 +2214,8 @@ static int d40_config_memcpy(struct d40_chan *d40c)
>
> static int d40_free_dma(struct d40_chan *d40c)
> {
> -
> + struct d40_desc *d40d;
> + struct d40_desc *_d;
> int res = 0;
> u32 event = D40_TYPE_TO_EVENT(d40c->dma_cfg.dev_type);
> struct d40_phy_res *phy = d40c->phy_chan;
> @@ -2252,6 +2245,12 @@ static int d40_free_dma(struct d40_chan *d40c)
> /* Release descriptor state; this does not access DMA40 registers. */
> d40_term_all(d40c);
>
> + /* Release any prepared descriptors the client did not submit. */
> + list_for_each_entry_safe(d40d, _d, &d40c->prepare_queue, node) {
> + d40_desc_remove(d40d);
> + d40_desc_free(d40c, d40d);
> + }
> +
> res = pm_runtime_resume_and_get(d40c->base->dev);
> if (res >= 0) {
> res = d40_channel_execute_command(d40c, D40_DMA_STOP);
> @@ -2508,10 +2507,7 @@ d40_prep_sg(struct dma_chan *dchan, struct scatterlist *sg_src,
> goto free_desc;
> }
>
> - /*
> - * add descriptor to the prepare queue in order to be able
> - * to free them later in terminate_all
> - */
> + /* Track prepared descriptors until submission or channel release. */
> list_add_tail(&desc->node, &chan->prepare_queue);
>
> spin_unlock_irqrestore(&chan->lock, flags);
>
> ---
> base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
> change-id: 20261104-dma40-fixes-d9db6094b7d9
>
> Best regards,
> --
> Linus Walleij <linusw@kernel.org>
>
prev parent reply other threads:[~2026-11-12 21:00 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-04 21:28 [PATCH] dmaengine: ste_dma40: Keep prepared descriptors across termination Linus Walleij
2026-11-12 20:59 ` Frank Li [this message]
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=avYourwIh_YGRjcH@SMW015318 \
--to=frank.li@oss.nxp.com \
--cc=Frank.Li@kernel.org \
--cc=dmaengine@vger.kernel.org \
--cc=linusw@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=ulfh@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