* Re: [PATCH v2] usb: Add a lock when freeing data in usbtmc_disconnect [not found] <91ca594b3f434040a54485339c0e320f@rohde-schwarz.com> @ 2021-04-14 15:10 ` guido 0 siblings, 0 replies; 3+ messages in thread From: guido @ 2021-04-14 15:10 UTC (permalink / raw) To: dpenkler; +Cc: linux-usb, lyl2019, guido.kiener, gregkh Hi Dave, I just found this patch, which does not seem to be a correct solution to solve a race. Maybe there is really an issue with the refcounting of data->kref, but currently I do not have time to check this in home office. At least I see a problem in usbtmc_probe() After calling: /* Protect interrupt in endpoint data until iin_urb is freed */ kref_get(&data->kref); the refcounter is incremented again and if usbtmc_probe() fails, the counter is only decremented with a single kref_put(..). I don't know if this is the reason of Lv Yunglong's problem, but let me know if you have time to track down this issue, and we will work on a correct and tested patch. Regards, Guido > -----Original Message----- > From: Greg KH > Sent: Tuesday, March 23, 2021 10:44 AM > To: Lv Yunlong <lyl2019@mail.ustc.edu.cn> > Cc: linux-usb@vger.kernel.org; linux-kernel@vger.kernel.org > Subject: Re: [PATCH v2] usb: Add a lock when freeing data in > usbtmc_disconnect > > On Tue, Mar 23, 2021 at 02:28:54AM -0700, Lv Yunlong wrote: >> In usbtmc_disconnect, data is got from intf with the initial reference. >> There is no refcount inc operation before usbmc_free_int(data). In >> usbmc_free_int(data), the data may be freed. >> >> But later in usbtmc_disconnect, there is another put function of data. >> It could cause errors in race. >> >> My patch adds a lock to protect kref from changing in race. >> >> Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn> >> --- >> drivers/usb/class/usbtmc.c | 5 +++++ >> 1 file changed, 5 insertions(+) >> >> diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c >> index 74d5a9c5238a..44f1fcabbb1e 100644 >> --- a/drivers/usb/class/usbtmc.c >> +++ b/drivers/usb/class/usbtmc.c >> @@ -2493,8 +2493,13 @@ static void usbtmc_disconnect(struct >> usb_interface *intf) >> usb_scuttle_anchored_urbs(&file_data->in_anchor); >> } >> mutex_unlock(&data->io_mutex); >> + >> + spinlock_t *dev_lock = &data->dev_lock; >> + >> + spin_lock_irq(dev_lock); >> usbtmc_free_int(data); >> kref_put(&data->kref, usbtmc_delete); >> + spin_unlock_irq(dev_lock); >> } >> >> static void usbtmc_draw_down(struct usbtmc_file_data *file_data) >> -- >> 2.25.1 > > You obviously did not even build this patch, let alone test it :( > > Please do not waste maintainer's time by not doing the proper steps > when submitting patches. > > thanks, > > greg k-h ^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH v2] usb: Add a lock when freeing data in usbtmc_disconnect @ 2021-03-23 9:28 Lv Yunlong 2021-03-23 9:44 ` Greg KH 0 siblings, 1 reply; 3+ messages in thread From: Lv Yunlong @ 2021-03-23 9:28 UTC (permalink / raw) To: gregkh, linux-usb; +Cc: linux-kernel, Lv Yunlong In usbtmc_disconnect, data is got from intf with the initial reference. There is no refcount inc operation before usbmc_free_int(data). In usbmc_free_int(data), the data may be freed. But later in usbtmc_disconnect, there is another put function of data. It could cause errors in race. My patch adds a lock to protect kref from changing in race. Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn> --- drivers/usb/class/usbtmc.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c index 74d5a9c5238a..44f1fcabbb1e 100644 --- a/drivers/usb/class/usbtmc.c +++ b/drivers/usb/class/usbtmc.c @@ -2493,8 +2493,13 @@ static void usbtmc_disconnect(struct usb_interface *intf) usb_scuttle_anchored_urbs(&file_data->in_anchor); } mutex_unlock(&data->io_mutex); + + spinlock_t *dev_lock = &data->dev_lock; + + spin_lock_irq(dev_lock); usbtmc_free_int(data); kref_put(&data->kref, usbtmc_delete); + spin_unlock_irq(dev_lock); } static void usbtmc_draw_down(struct usbtmc_file_data *file_data) -- 2.25.1 ^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v2] usb: Add a lock when freeing data in usbtmc_disconnect 2021-03-23 9:28 Lv Yunlong @ 2021-03-23 9:44 ` Greg KH 0 siblings, 0 replies; 3+ messages in thread From: Greg KH @ 2021-03-23 9:44 UTC (permalink / raw) To: Lv Yunlong; +Cc: linux-usb, linux-kernel On Tue, Mar 23, 2021 at 02:28:54AM -0700, Lv Yunlong wrote: > In usbtmc_disconnect, data is got from intf with the initial reference. > There is no refcount inc operation before usbmc_free_int(data). In > usbmc_free_int(data), the data may be freed. > > But later in usbtmc_disconnect, there is another put function of data. > It could cause errors in race. > > My patch adds a lock to protect kref from changing in race. > > Signed-off-by: Lv Yunlong <lyl2019@mail.ustc.edu.cn> > --- > drivers/usb/class/usbtmc.c | 5 +++++ > 1 file changed, 5 insertions(+) > > diff --git a/drivers/usb/class/usbtmc.c b/drivers/usb/class/usbtmc.c > index 74d5a9c5238a..44f1fcabbb1e 100644 > --- a/drivers/usb/class/usbtmc.c > +++ b/drivers/usb/class/usbtmc.c > @@ -2493,8 +2493,13 @@ static void usbtmc_disconnect(struct usb_interface *intf) > usb_scuttle_anchored_urbs(&file_data->in_anchor); > } > mutex_unlock(&data->io_mutex); > + > + spinlock_t *dev_lock = &data->dev_lock; > + > + spin_lock_irq(dev_lock); > usbtmc_free_int(data); > kref_put(&data->kref, usbtmc_delete); > + spin_unlock_irq(dev_lock); > } > > static void usbtmc_draw_down(struct usbtmc_file_data *file_data) > -- > 2.25.1 You obviously did not even build this patch, let alone test it :( Please do not waste maintainer's time by not doing the proper steps when submitting patches. thanks, greg k-h ^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2021-04-14 15:23 UTC | newest] Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- [not found] <91ca594b3f434040a54485339c0e320f@rohde-schwarz.com> 2021-04-14 15:10 ` [PATCH v2] usb: Add a lock when freeing data in usbtmc_disconnect guido 2021-03-23 9:28 Lv Yunlong 2021-03-23 9:44 ` Greg KH
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.