| Message ID | 20191230161321.2738541-1-bryan.odonoghue@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | usb: dwc3: gadget: Fix failure to detect end of transfer | expand |
On Mon, Dec 30, 2019 at 04:13:21PM +0000, Bryan O'Donoghue wrote: > A recent bugfix 8c7d4b7b3d43 ("usb: dwc3: gadget: Fix logical condition") > correctly fixes a logical error in the gadget driver but, exposes a further > bug in determining when a transfer has completed. > > Prior to 8c7d4b7b3d43 we were calling dwc3_gadget_giveback() when we > shouldn't have been. Afer this change the below test fails to complete on > my hardware. > > Host: > echo "host" > /dev/ttyACM0 > > Device: > cat < /dev/ttyGS0 > > This is caused by the driver incorrectly detecting end of transfer, a > problem that had previous been masked by the continuous calling of > dwc3_gadget_giveback() prior to 8c7d4b7b3d43. > > Remediate by making the test <= instead of == > > Fixes: e0c42ce590fe ("usb: dwc3: gadget: simplify IOC handling") > > Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> > --- > drivers/usb/dwc3/gadget.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) I think this patch: https://lore.kernel.org/linux-usb/ac5a3593a94fdaa3d92e6352356b5f7a01ccdc7c.1576291140.git.thinhn@synopsys.com/ should fix this issue instead, right? If not, do I need to include both of these? thanks, greg k-h
On 30/12/2019 18:57, Greg KH wrote: > On Mon, Dec 30, 2019 at 04:13:21PM +0000, Bryan O'Donoghue wrote: >> A recent bugfix 8c7d4b7b3d43 ("usb: dwc3: gadget: Fix logical condition") >> correctly fixes a logical error in the gadget driver but, exposes a further >> bug in determining when a transfer has completed. >> >> Prior to 8c7d4b7b3d43 we were calling dwc3_gadget_giveback() when we >> shouldn't have been. Afer this change the below test fails to complete on >> my hardware. >> >> Host: >> echo "host" > /dev/ttyACM0 >> >> Device: >> cat < /dev/ttyGS0 >> >> This is caused by the driver incorrectly detecting end of transfer, a >> problem that had previous been masked by the continuous calling of >> dwc3_gadget_giveback() prior to 8c7d4b7b3d43. >> >> Remediate by making the test <= instead of == >> >> Fixes: e0c42ce590fe ("usb: dwc3: gadget: simplify IOC handling") >> >> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> >> --- >> drivers/usb/dwc3/gadget.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) > > I think this patch: > https://lore.kernel.org/linux-usb/ac5a3593a94fdaa3d92e6352356b5f7a01ccdc7c.1576291140.git.thinhn@synopsys.com/ > > should fix this issue instead, right? > > If not, do I need to include both of these? Yep, works fine in isolation.
On 31/12/2019 00:05, Bryan O'Donoghue wrote: > On 30/12/2019 18:57, Greg KH wrote: >> On Mon, Dec 30, 2019 at 04:13:21PM +0000, Bryan O'Donoghue wrote: >>> A recent bugfix 8c7d4b7b3d43 ("usb: dwc3: gadget: Fix logical >>> condition") >>> correctly fixes a logical error in the gadget driver but, exposes a >>> further >>> bug in determining when a transfer has completed. >>> >>> Prior to 8c7d4b7b3d43 we were calling dwc3_gadget_giveback() when we >>> shouldn't have been. Afer this change the below test fails to >>> complete on >>> my hardware. >>> >>> Host: >>> echo "host" > /dev/ttyACM0 >>> >>> Device: >>> cat < /dev/ttyGS0 >>> >>> This is caused by the driver incorrectly detecting end of transfer, a >>> problem that had previous been masked by the continuous calling of >>> dwc3_gadget_giveback() prior to 8c7d4b7b3d43. >>> >>> Remediate by making the test <= instead of == >>> >>> Fixes: e0c42ce590fe ("usb: dwc3: gadget: simplify IOC handling") >>> >>> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> >>> --- >>> drivers/usb/dwc3/gadget.c | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> I think this patch: >> https://lore.kernel.org/linux-usb/ac5a3593a94fdaa3d92e6352356b5f7a01ccdc7c.1576291140.git.thinhn@synopsys.com/ >> >> >> should fix this issue instead, right? >> >> If not, do I need to include both of these? > > Yep, works fine in isolation. > I don't have that in my inbox anywhere - not sure why, would have saved me 1/2 a day of work. Anyway, please feel free to add my Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> to Thinh's patch --- bod
diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index 0c960a97ea02..464c4d9961c7 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -2467,7 +2467,7 @@ static int dwc3_gadget_ep_reclaim_trb_linear(struct dwc3_ep *dep, static bool dwc3_gadget_ep_request_completed(struct dwc3_request *req) { - return req->request.actual == req->request.length; + return req->request.actual <= req->request.length; } static int dwc3_gadget_ep_cleanup_completed_request(struct dwc3_ep *dep,
A recent bugfix 8c7d4b7b3d43 ("usb: dwc3: gadget: Fix logical condition") correctly fixes a logical error in the gadget driver but, exposes a further bug in determining when a transfer has completed. Prior to 8c7d4b7b3d43 we were calling dwc3_gadget_giveback() when we shouldn't have been. Afer this change the below test fails to complete on my hardware. Host: echo "host" > /dev/ttyACM0 Device: cat < /dev/ttyGS0 This is caused by the driver incorrectly detecting end of transfer, a problem that had previous been masked by the continuous calling of dwc3_gadget_giveback() prior to 8c7d4b7b3d43. Remediate by making the test <= instead of == Fixes: e0c42ce590fe ("usb: dwc3: gadget: simplify IOC handling") Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org> --- drivers/usb/dwc3/gadget.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)