From: Baochen Qiang <baochen.qiang@oss.qualcomm.com>
To: ZhaoJinming <zhaojinming@uniontech.com>,
Jeff Johnson <jjohnson@kernel.org>
Cc: linux-wireless@vger.kernel.org, ath11k@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2] wifi: ath11k: fix resource leak on error in ext IRQ setup
Date: Wed, 15 Jul 2026 11:15:51 +0800 [thread overview]
Message-ID: <3c67115f-5aea-402a-987b-c3092e0c3386@oss.qualcomm.com> (raw)
In-Reply-To: <20260713105912.20095-1-zhaojinming@uniontech.com>
On 7/13/2026 6:59 PM, ZhaoJinming wrote:
> In ath11k_ahb_config_irq(), when a CE request_irq() fails, the function
> returns the error immediately without freeing the CE IRQs that were
> successfully registered in previous loop iterations. The probe error
> path does not call ath11k_ahb_free_irq() either, so the previously
> registered CE IRQ handlers remain attached to the interrupt lines and
> are never released.
>
> In ath11k_ahb_config_ext_irq(), when an external request_irq() fails,
> the error is only logged and the loop continues. The function then
> returns 0 indicating success, leaving the device in a partially
> configured state where some external IRQs are not registered. This
> causes enable_irq()/disable_irq()/free_irq() to be called on
> unregistered IRQs during runtime and remove/shutdown, triggering
> WARN_ON(!desc->action), and missing interrupt handlers lead to data
> loss.
>
> Additionally, if alloc_netdev_dummy() fails for a later IRQ group, the
> function returns -ENOMEM without freeing the ext IRQs and napi_ndev
> that were successfully set up for earlier groups.
>
> Fix all three issues: propagate the error up to the caller and unwind
> all successfully registered IRQs and allocated resources on failure.
>
> Signed-off-by: ZhaoJinming <zhaojinming@uniontech.com>
> ---
>
> Changes in v2:
> - Move `irq_grp` declaration from for-loop body to function scope to fix
> compile error in err_request_irq error path.
> - Set ret = -ENOMEM before goto err_request_irq on alloc_netdev_dummy()
> failure to avoid returning uninitialized value.
> - When ath11k_ahb_config_ext_irq() fails, unwind the already-registered
> CE IRQs via a shared err_ce_irq cleanup path to avoid leaking them.
>
> drivers/net/wireless/ath/ath11k/ahb.c | 48 ++++++++++++++++++++++++++--
> 1 file changed, 44 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/wireless/ath/ath11k/ahb.c b/drivers/net/wireless/ath/ath11k/ahb.c
> index f566d699d074..11bafbd94aff 100644
> --- a/drivers/net/wireless/ath/ath11k/ahb.c
> +++ b/drivers/net/wireless/ath/ath11k/ahb.c
> @@ -524,20 +524,24 @@ static irqreturn_t ath11k_ahb_ext_interrupt_handler(int irq, void *arg)
> static int ath11k_ahb_config_ext_irq(struct ath11k_base *ab)
> {
> struct ath11k_hw_params *hw = &ab->hw_params;
> + struct ath11k_ext_irq_grp *irq_grp;
> int i, j;
> int irq;
> int ret;
>
> for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++) {
> - struct ath11k_ext_irq_grp *irq_grp = &ab->ext_irq_grp[i];
> + irq_grp = &ab->ext_irq_grp[i];
> u32 num_irq = 0;
declaration after statement. better to have num_irq declaration at the beginning of the
for loop.
>
> irq_grp->ab = ab;
> irq_grp->grp_id = i;
>
> irq_grp->napi_ndev = alloc_netdev_dummy(0);
> - if (!irq_grp->napi_ndev)
> - return -ENOMEM;
> + if (!irq_grp->napi_ndev) {
> + ret = -ENOMEM;
> + irq_grp->num_irq = 0;
the ath12k_base is allocated by kzalloc, so the ext_irq_grp array, as a member of
ath12k_base, is already zero'ed, hence this re-initialization is redundant. But I am OK
with it since it improves code readability.
> + goto err_request_irq;
> + }
>
> netif_napi_add(irq_grp->napi_ndev, &irq_grp->napi,
> ath11k_ahb_ext_grp_napi_poll);
> @@ -600,11 +604,25 @@ static int ath11k_ahb_config_ext_irq(struct ath11k_base *ab)
> if (ret) {
> ath11k_err(ab, "failed request_irq for %d\n",
> irq);
> + irq_grp->num_irq = j;
with this forcing to the number of successful irq request ...
> + goto err_request_irq;
> }
> }
> }
>
> return 0;
> +
> +err_request_irq:
> + for ( ; i >= 0; i--) {
then with the index starting from the failed loop ...
> + irq_grp = &ab->ext_irq_grp[i];
> + for (j = irq_grp->num_irq - 1; j >= 0; j--)
> + free_irq(ab->irq_num[irq_grp->irqs[j]], irq_grp);
> + if (irq_grp->napi_ndev) {
and with the guard ...
> + netif_napi_del(&irq_grp->napi);
> + free_netdev(irq_grp->napi_ndev);
> + }
> + }
at the cost of above three, the centralized err handling is working. Yeah this simplifies
the code, but makes it hard to understand.
what about:
1. refactor the existing ath11k_ahb_free_ext_irq():
ath11k_ahb_free_ext_irq_grp()
{
for (j = 0; j < irq_grp->num_irq; j++)
free_irq(ab->irq_num[irq_grp->irqs[j]], irq_grp);
netif_napi_del(&irq_grp->napi);
free_netdev(irq_grp->napi_ndev);
}
then we can have
ath11k_ahb_free_ext_irq()
{
for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++)
ath11k_ahb_free_ext_irq_grp(irq_grp)
}
and the complete err handling of ext irq can be
ath11k_ahb_config_ext_irq()
{
for (i = 0; i < ATH11K_EXT_IRQ_GRP_NUM_MAX; i++) {
irq_grp->napi_ndev = alloc_netdev_dummy(0);
if (!irq_grp->napi_ndev) {
ret = -ENOMEM;
irq_grp->num_irq = 0;
goto err_request_irq;
}
...
for (j = 0; j < num_irq; j++) {
ret = request_irq();
if (ret) {
irq_grp->num_irq = j;
ath11k_ahb_free_ext_irq_grp(irq_grp);
goto err_request_irq;
}
}
irq_grp->num_irq = num_irq;
}
err_request_irq:
for ( i--; i >= 0; i--) {
irq_grp = &ab->ext_irq_grp[i];
ath11k_ahb_free_ext_irq_grp(irq_grp);
}
return ret;
}
> + return ret;
> }
>
> static int ath11k_ahb_config_irq(struct ath11k_base *ab)
> @@ -629,15 +647,30 @@ static int ath11k_ahb_config_irq(struct ath11k_base *ab)
> ret = request_irq(irq, ath11k_ahb_ce_interrupt_handler,
> IRQF_TRIGGER_RISING, irq_name[irq_idx],
> ce_pipe);
> - if (ret)
> - return ret;
> + if (ret) {
> + ath11k_err(ab, "failed request_irq for %d\n", irq);
> + goto err_ce_irq;
> + }
>
> ab->irq_num[irq_idx] = irq;
> }
>
> /* Configure external interrupts */
> ret = ath11k_ahb_config_ext_irq(ab);
> + if (ret) {
> + ath11k_err(ab, "failed to configure ext irq: %d\n", ret);
> + goto err_ce_irq;
> + }
> +
> + return 0;
>
> +err_ce_irq:
> + for (i--; i >= 0; i--) {
> + if (ath11k_ce_get_attr_flags(ab, i) & CE_ATTR_DIS_INTR)
> + continue;
> + free_irq(ab->irq_num[ATH11K_IRQ_CE0_OFFSET + i],
> + &ab->ce.ce_pipe[i]);
> + }
this is also a duplication of the CE part of ath11k_ahb_free_irq(), consider refactoring
the exsiting helper and reuse it.
> return ret;
> }
>
next prev parent reply other threads:[~2026-07-15 3:15 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-22 2:56 [PATCH] " ZhaoJinming
2026-07-13 5:33 ` Baochen Qiang
2026-07-13 10:59 ` [PATCH v2] " ZhaoJinming
2026-07-15 3:15 ` Baochen Qiang [this message]
2026-07-16 11:46 ` [PATCH v3] " ZhaoJinming
2026-07-17 3:47 ` Baochen Qiang
2026-07-21 7:21 ` [PATCH v4] " ZhaoJinming
2026-07-21 7:40 ` Baochen Qiang
2026-07-21 14:05 ` Jeff Johnson
2026-07-23 1:12 ` 赵金明
2026-08-05 18:29 ` [PATCH] " kernel test robot
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=3c67115f-5aea-402a-987b-c3092e0c3386@oss.qualcomm.com \
--to=baochen.qiang@oss.qualcomm.com \
--cc=ath11k@lists.infradead.org \
--cc=jjohnson@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-wireless@vger.kernel.org \
--cc=zhaojinming@uniontech.com \
/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
all inboxes | Powered by JetHome®