Message ID | 20221125054119.25135-1-quic_ugoswami@quicinc.com |
---|---|
State | Superseded |
Headers | show |
Series | [v3] usb: gadget: f_fs: Prevent race between functionfs_unbind & ffs_ep0_queue_wait | expand |
Hi All, On 11/25/22 11:11 AM, Udipto Goswami wrote: > While performing fast composition switch, there is a possibility that the > process of ffs_ep0_write/ffs_ep0_read get into a race condition > due to ep0req being freed up from functionfs_unbind. > > Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait > by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't > bounded so it can go ahead and mark the ep0req to NULL, and since there > is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. > > Fix this by making a serialized execution between the two functions using > a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no > other function can use it after the free operation. > > Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") > Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> > --- > v3: Moved dequeue out of mutex to prevent deadlock > > drivers/usb/gadget/function/f_fs.c | 7 +++++++ > 1 file changed, 7 insertions(+) > > diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c > index 73dc10a77cde..523a961b910b 100644 > --- a/drivers/usb/gadget/function/f_fs.c > +++ b/drivers/usb/gadget/function/f_fs.c > @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) > struct usb_request *req = ffs->ep0req; > int ret; > > + if (!req) > + return -EINVAL; > + > req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); > > spin_unlock_irq(&ffs->ev.waitq.lock); > @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) > ENTER(); > > if (!WARN_ON(!ffs->gadget)) { > + /* dequeue before freeing ep0req */ > + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); > + mutex_lock(&ffs->mutex); > usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); > ffs->ep0req = NULL; > ffs->gadget = NULL; > clear_bit(FFS_FL_BOUND, &ffs->flags); > + mutex_unlock(&ffs->mutex); > ffs_data_put(ffs); > } > } Gentle reminder for any comments/suggestions on this patch. Thanks, -Udipto
On Wed, Dec 14, 2022 at 12:47:17PM +0530, Udipto Goswami wrote: > Hi All, > > On 11/25/22 11:11 AM, Udipto Goswami wrote: > > While performing fast composition switch, there is a possibility that the > > process of ffs_ep0_write/ffs_ep0_read get into a race condition > > due to ep0req being freed up from functionfs_unbind. > > > > Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait > > by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't > > bounded so it can go ahead and mark the ep0req to NULL, and since there > > is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. > > > > Fix this by making a serialized execution between the two functions using > > a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no > > other function can use it after the free operation. > > > > Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") > > Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> > > --- > > v3: Moved dequeue out of mutex to prevent deadlock > > > > drivers/usb/gadget/function/f_fs.c | 7 +++++++ > > 1 file changed, 7 insertions(+) > > > > diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c > > index 73dc10a77cde..523a961b910b 100644 > > --- a/drivers/usb/gadget/function/f_fs.c > > +++ b/drivers/usb/gadget/function/f_fs.c > > @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) > > struct usb_request *req = ffs->ep0req; > > int ret; > > + if (!req) > > + return -EINVAL; > > + > > req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); > > spin_unlock_irq(&ffs->ev.waitq.lock); > > @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) > > ENTER(); > > if (!WARN_ON(!ffs->gadget)) { > > + /* dequeue before freeing ep0req */ > > + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); > > + mutex_lock(&ffs->mutex); > > usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); > > ffs->ep0req = NULL; > > ffs->gadget = NULL; > > clear_bit(FFS_FL_BOUND, &ffs->flags); > > + mutex_unlock(&ffs->mutex); > > ffs_data_put(ffs); > > } > > } > > Gentle reminder for any comments/suggestions on this patch. It's the middle of the merge window, and you submitted a patch that has obvious coding style issues, so there's nothing that we can do with it no matter what... Also, you are not explaining how this has been tested, or if it really solves the problem for you or not, and it seems you are mixing two different things into the same patch. thanks, greg k-h
Hi Greg, On 12/14/22 1:53 PM, Greg Kroah-Hartman wrote: > On Wed, Dec 14, 2022 at 12:47:17PM +0530, Udipto Goswami wrote: >> Hi All, >> >> On 11/25/22 11:11 AM, Udipto Goswami wrote: >>> While performing fast composition switch, there is a possibility that the >>> process of ffs_ep0_write/ffs_ep0_read get into a race condition >>> due to ep0req being freed up from functionfs_unbind. >>> >>> Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait >>> by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't >>> bounded so it can go ahead and mark the ep0req to NULL, and since there >>> is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. >>> >>> Fix this by making a serialized execution between the two functions using >>> a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no >>> other function can use it after the free operation. >>> >>> Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") >>> Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> >>> --- >>> v3: Moved dequeue out of mutex to prevent deadlock >>> >>> drivers/usb/gadget/function/f_fs.c | 7 +++++++ >>> 1 file changed, 7 insertions(+) >>> >>> diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c >>> index 73dc10a77cde..523a961b910b 100644 >>> --- a/drivers/usb/gadget/function/f_fs.c >>> +++ b/drivers/usb/gadget/function/f_fs.c >>> @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) >>> struct usb_request *req = ffs->ep0req; >>> int ret; >>> + if (!req) >>> + return -EINVAL; >>> + >>> req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); >>> spin_unlock_irq(&ffs->ev.waitq.lock); >>> @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) >>> ENTER(); >>> if (!WARN_ON(!ffs->gadget)) { >>> + /* dequeue before freeing ep0req */ >>> + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); >>> + mutex_lock(&ffs->mutex); >>> usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); >>> ffs->ep0req = NULL; >>> ffs->gadget = NULL; >>> clear_bit(FFS_FL_BOUND, &ffs->flags); >>> + mutex_unlock(&ffs->mutex); >>> ffs_data_put(ffs); >>> } >>> } >> >> Gentle reminder for any comments/suggestions on this patch. > > It's the middle of the merge window, and you submitted a patch that has > obvious coding style issues, so there's nothing that we can do with it > no matter what... Apologies for this, I had ran checkpatch, I didn't get any error: total: 0 errors, 0 warnings, 23 lines checked 0001-usb-gadget-f_fs-Prevent-race-between-functionfs_unbi.patch has no obvious style problems and is ready for submission. was curious what's the coding style error you are referring here? > > Also, you are not explaining how this has been tested, or if it really > solves the problem for you or not, and it seems you are mixing two > different things into the same patch. Yah it does solves the problem of the race condition. For testing this i was running USB Plug-in Plug-out along with fast composition switch. Thanks for pointing it out I'll make a a separate patch for ep0_dequeue. > > thanks, > > greg k-h
On Wed, Dec 14, 2022 at 03:42:10PM +0530, Udipto Goswami wrote: > Hi Greg, > > On 12/14/22 1:53 PM, Greg Kroah-Hartman wrote: > > On Wed, Dec 14, 2022 at 12:47:17PM +0530, Udipto Goswami wrote: > > > Hi All, > > > > > > On 11/25/22 11:11 AM, Udipto Goswami wrote: > > > > While performing fast composition switch, there is a possibility that the > > > > process of ffs_ep0_write/ffs_ep0_read get into a race condition > > > > due to ep0req being freed up from functionfs_unbind. > > > > > > > > Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait > > > > by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't > > > > bounded so it can go ahead and mark the ep0req to NULL, and since there > > > > is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. > > > > > > > > Fix this by making a serialized execution between the two functions using > > > > a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no > > > > other function can use it after the free operation. > > > > > > > > Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") > > > > Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> > > > > --- > > > > v3: Moved dequeue out of mutex to prevent deadlock > > > > > > > > drivers/usb/gadget/function/f_fs.c | 7 +++++++ > > > > 1 file changed, 7 insertions(+) > > > > > > > > diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c > > > > index 73dc10a77cde..523a961b910b 100644 > > > > --- a/drivers/usb/gadget/function/f_fs.c > > > > +++ b/drivers/usb/gadget/function/f_fs.c > > > > @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) > > > > struct usb_request *req = ffs->ep0req; > > > > int ret; > > > > + if (!req) > > > > + return -EINVAL; > > > > + > > > > req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); > > > > spin_unlock_irq(&ffs->ev.waitq.lock); > > > > @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) > > > > ENTER(); > > > > if (!WARN_ON(!ffs->gadget)) { > > > > + /* dequeue before freeing ep0req */ > > > > + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); > > > > + mutex_lock(&ffs->mutex); > > > > usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); > > > > ffs->ep0req = NULL; > > > > ffs->gadget = NULL; > > > > clear_bit(FFS_FL_BOUND, &ffs->flags); > > > > + mutex_unlock(&ffs->mutex); > > > > ffs_data_put(ffs); > > > > } > > > > } > > > > > > Gentle reminder for any comments/suggestions on this patch. > > > > It's the middle of the merge window, and you submitted a patch that has > > obvious coding style issues, so there's nothing that we can do with it > > no matter what... > > Apologies for this, I had ran checkpatch, I didn't get any error: > > total: 0 errors, 0 warnings, 23 lines checked > > 0001-usb-gadget-f_fs-Prevent-race-between-functionfs_unbi.patch has no > obvious style problems and is ready for submission. > > was curious what's the coding style error you are referring here? No blank line after the variable description and before your new if statement. thanks, greg k-h
On 12/14/22 3:52 PM, Greg Kroah-Hartman wrote: > On Wed, Dec 14, 2022 at 03:42:10PM +0530, Udipto Goswami wrote: >> Hi Greg, >> >> On 12/14/22 1:53 PM, Greg Kroah-Hartman wrote: >>> On Wed, Dec 14, 2022 at 12:47:17PM +0530, Udipto Goswami wrote: >>>> Hi All, >>>> >>>> On 11/25/22 11:11 AM, Udipto Goswami wrote: >>>>> While performing fast composition switch, there is a possibility that the >>>>> process of ffs_ep0_write/ffs_ep0_read get into a race condition >>>>> due to ep0req being freed up from functionfs_unbind. >>>>> >>>>> Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait >>>>> by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't >>>>> bounded so it can go ahead and mark the ep0req to NULL, and since there >>>>> is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. >>>>> >>>>> Fix this by making a serialized execution between the two functions using >>>>> a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no >>>>> other function can use it after the free operation. >>>>> >>>>> Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") >>>>> Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> >>>>> --- >>>>> v3: Moved dequeue out of mutex to prevent deadlock >>>>> >>>>> drivers/usb/gadget/function/f_fs.c | 7 +++++++ >>>>> 1 file changed, 7 insertions(+) >>>>> >>>>> diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c >>>>> index 73dc10a77cde..523a961b910b 100644 >>>>> --- a/drivers/usb/gadget/function/f_fs.c >>>>> +++ b/drivers/usb/gadget/function/f_fs.c >>>>> @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) >>>>> struct usb_request *req = ffs->ep0req; >>>>> int ret; >>>>> + if (!req) >>>>> + return -EINVAL; >>>>> + >>>>> req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); >>>>> spin_unlock_irq(&ffs->ev.waitq.lock); >>>>> @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) >>>>> ENTER(); >>>>> if (!WARN_ON(!ffs->gadget)) { >>>>> + /* dequeue before freeing ep0req */ >>>>> + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); >>>>> + mutex_lock(&ffs->mutex); >>>>> usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); >>>>> ffs->ep0req = NULL; >>>>> ffs->gadget = NULL; >>>>> clear_bit(FFS_FL_BOUND, &ffs->flags); >>>>> + mutex_unlock(&ffs->mutex); >>>>> ffs_data_put(ffs); >>>>> } >>>>> } >>>> >>>> Gentle reminder for any comments/suggestions on this patch. >>> >>> It's the middle of the merge window, and you submitted a patch that has >>> obvious coding style issues, so there's nothing that we can do with it >>> no matter what... >> >> Apologies for this, I had ran checkpatch, I didn't get any error: >> >> total: 0 errors, 0 warnings, 23 lines checked >> >> 0001-usb-gadget-f_fs-Prevent-race-between-functionfs_unbi.patch has no >> obvious style problems and is ready for submission. >> >> was curious what's the coding style error you are referring here? > > No blank line after the variable description and before your new if > statement. Aah, Thanks for pointing out I'll split the change into two and push v4 > > thanks, > > greg k-h
diff --git a/drivers/usb/gadget/function/f_fs.c b/drivers/usb/gadget/function/f_fs.c index 73dc10a77cde..523a961b910b 100644 --- a/drivers/usb/gadget/function/f_fs.c +++ b/drivers/usb/gadget/function/f_fs.c @@ -279,6 +279,9 @@ static int __ffs_ep0_queue_wait(struct ffs_data *ffs, char *data, size_t len) struct usb_request *req = ffs->ep0req; int ret; + if (!req) + return -EINVAL; + req->zero = len < le16_to_cpu(ffs->ev.setup.wLength); spin_unlock_irq(&ffs->ev.waitq.lock); @@ -1892,10 +1895,14 @@ static void functionfs_unbind(struct ffs_data *ffs) ENTER(); if (!WARN_ON(!ffs->gadget)) { + /* dequeue before freeing ep0req */ + usb_ep_dequeue(ffs->gadget->ep0, ffs->ep0req); + mutex_lock(&ffs->mutex); usb_ep_free_request(ffs->gadget->ep0, ffs->ep0req); ffs->ep0req = NULL; ffs->gadget = NULL; clear_bit(FFS_FL_BOUND, &ffs->flags); + mutex_unlock(&ffs->mutex); ffs_data_put(ffs); } }
While performing fast composition switch, there is a possibility that the process of ffs_ep0_write/ffs_ep0_read get into a race condition due to ep0req being freed up from functionfs_unbind. Consider the scenario that the ffs_ep0_write calls the ffs_ep0_queue_wait by taking a lock &ffs->ev.waitq.lock. However, the functionfs_unbind isn't bounded so it can go ahead and mark the ep0req to NULL, and since there is no NULL check in ffs_ep0_queue_wait we will end up in use-after-free. Fix this by making a serialized execution between the two functions using a mutex_lock(ffs->mutex). Also, dequeue the ep0req to ensure that no other function can use it after the free operation. Fixes: ddf8abd25994 ("USB: f_fs: the FunctionFS driver") Signed-off-by: Udipto Goswami <quic_ugoswami@quicinc.com> --- v3: Moved dequeue out of mutex to prevent deadlock drivers/usb/gadget/function/f_fs.c | 7 +++++++ 1 file changed, 7 insertions(+)