| Message ID | 20230925091741.797863-1-baijiaju@buaa.edu.cn (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | usb: dwc2: fix possible NULL pointer dereference caused by driver concurrency | expand |
On Mon, Sep 25, 2023 at 05:17:41PM +0800, Jia-Ju Bai wrote: > In _dwc2_hcd_urb_enqueue(), "urb->hcpriv = NULL" is executed without > holding the lock "hsotg->lock". In _dwc2_hcd_urb_dequeue(): > > spin_lock_irqsave(&hsotg->lock, flags); > ... > if (!urb->hcpriv) { > dev_dbg(hsotg->dev, "## urb->hcpriv is NULL ##\n"); > goto out; > } > rc = dwc2_hcd_urb_dequeue(hsotg, urb->hcpriv); // Use urb->hcpriv > ... > out: > spin_unlock_irqrestore(&hsotg->lock, flags); > > When _dwc2_hcd_urb_enqueue() and _dwc2_hcd_urb_dequeue() are > concurrently executed, the NULL check of "urb->hcpriv" can be executed > before "urb->hcpriv = NULL". After urb->hcpriv is NULL, it can be used > in the function call to dwc2_hcd_urb_dequeue(), which can cause a NULL > pointer dereference. Odd trailing spaces in your changelog text, is that intentional? > > To fix this possible bug, "urb->hcpriv = NULL" should be executed with > holding the lock "hsotg->lock". > > Signed-off-by: Jia-Ju Bai <baijiaju@buaa.edu.cn> > --- > drivers/usb/dwc2/hcd.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/usb/dwc2/hcd.c b/drivers/usb/dwc2/hcd.c > index 657f1f659ffa..35c7a4df8e71 100644 > --- a/drivers/usb/dwc2/hcd.c > +++ b/drivers/usb/dwc2/hcd.c > @@ -4769,8 +4769,8 @@ static int _dwc2_hcd_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, > if (qh_allocated && qh->channel && qh->channel->qh == qh) > qh->channel->qh = NULL; > fail2: > - spin_unlock_irqrestore(&hsotg->lock, flags); > urb->hcpriv = NULL; > + spin_unlock_irqrestore(&hsotg->lock, flags); > kfree(qtd); > fail1: > if (qh_allocated) { > -- > 2.34.1 > What commit id does this fix? And how did you test this to verify it works properly? And how was it found? thanks, greg k-h
Thanks for the reply :) On 2023/9/25 17:38, Greg KH wrote: > On Mon, Sep 25, 2023 at 05:17:41PM +0800, Jia-Ju Bai wrote: >> In _dwc2_hcd_urb_enqueue(), "urb->hcpriv = NULL" is executed without >> holding the lock "hsotg->lock". In _dwc2_hcd_urb_dequeue(): >> >> spin_lock_irqsave(&hsotg->lock, flags); >> ... >> if (!urb->hcpriv) { >> dev_dbg(hsotg->dev, "## urb->hcpriv is NULL ##\n"); >> goto out; >> } >> rc = dwc2_hcd_urb_dequeue(hsotg, urb->hcpriv); // Use urb->hcpriv >> ... >> out: >> spin_unlock_irqrestore(&hsotg->lock, flags); >> >> When _dwc2_hcd_urb_enqueue() and _dwc2_hcd_urb_dequeue() are >> concurrently executed, the NULL check of "urb->hcpriv" can be executed >> before "urb->hcpriv = NULL". After urb->hcpriv is NULL, it can be used >> in the function call to dwc2_hcd_urb_dequeue(), which can cause a NULL >> pointer dereference. > Odd trailing spaces in your changelog text, is that intentional? They are caused by the configuration of my file editor, and I changed it just now. > >> To fix this possible bug, "urb->hcpriv = NULL" should be executed with >> holding the lock "hsotg->lock". >> >> Signed-off-by: Jia-Ju Bai <baijiaju@buaa.edu.cn> >> --- >> drivers/usb/dwc2/hcd.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/usb/dwc2/hcd.c b/drivers/usb/dwc2/hcd.c >> index 657f1f659ffa..35c7a4df8e71 100644 >> --- a/drivers/usb/dwc2/hcd.c >> +++ b/drivers/usb/dwc2/hcd.c >> @@ -4769,8 +4769,8 @@ static int _dwc2_hcd_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, >> if (qh_allocated && qh->channel && qh->channel->qh == qh) >> qh->channel->qh = NULL; >> fail2: >> - spin_unlock_irqrestore(&hsotg->lock, flags); >> urb->hcpriv = NULL; >> + spin_unlock_irqrestore(&hsotg->lock, flags); >> kfree(qtd); >> fail1: >> if (qh_allocated) { >> -- >> 2.34.1 >> > What commit id does this fix? > > And how did you test this to verify it works properly? > > And how was it found? I sent a v2 patch to add these details and fix the mistakes about trailing spaces. Thanks, Jia-Ju Bai
diff --git a/drivers/usb/dwc2/hcd.c b/drivers/usb/dwc2/hcd.c index 657f1f659ffa..35c7a4df8e71 100644 --- a/drivers/usb/dwc2/hcd.c +++ b/drivers/usb/dwc2/hcd.c @@ -4769,8 +4769,8 @@ static int _dwc2_hcd_urb_enqueue(struct usb_hcd *hcd, struct urb *urb, if (qh_allocated && qh->channel && qh->channel->qh == qh) qh->channel->qh = NULL; fail2: - spin_unlock_irqrestore(&hsotg->lock, flags); urb->hcpriv = NULL; + spin_unlock_irqrestore(&hsotg->lock, flags); kfree(qtd); fail1: if (qh_allocated) {
In _dwc2_hcd_urb_enqueue(), "urb->hcpriv = NULL" is executed without holding the lock "hsotg->lock". In _dwc2_hcd_urb_dequeue(): spin_lock_irqsave(&hsotg->lock, flags); ... if (!urb->hcpriv) { dev_dbg(hsotg->dev, "## urb->hcpriv is NULL ##\n"); goto out; } rc = dwc2_hcd_urb_dequeue(hsotg, urb->hcpriv); // Use urb->hcpriv ... out: spin_unlock_irqrestore(&hsotg->lock, flags); When _dwc2_hcd_urb_enqueue() and _dwc2_hcd_urb_dequeue() are concurrently executed, the NULL check of "urb->hcpriv" can be executed before "urb->hcpriv = NULL". After urb->hcpriv is NULL, it can be used in the function call to dwc2_hcd_urb_dequeue(), which can cause a NULL pointer dereference. To fix this possible bug, "urb->hcpriv = NULL" should be executed with holding the lock "hsotg->lock". Signed-off-by: Jia-Ju Bai <baijiaju@buaa.edu.cn> --- drivers/usb/dwc2/hcd.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)