In commit 92af4fc6ec33 ("usb: musb: Fix suspend with devices
connected for a64"), the logic to support the
MUSB_QUIRK_B_DISCONNECT_99 quirk was modified to only conditionally
schedule the musb->irq_work delayed work.
This commit badly breaks ECM Gadget on AM335X. Indeed, with this
commit, one can observe massive packet loss:
$ ping 192.168.0.100
...
15 packets transmitted, 3 received, 80% packet loss, time 14316ms
Reverting this commit brings back a properly functioning ECM
Gadget. An analysis of the commit seems to indicate that a mistake was
made: the previous code was not falling through into the
MUSB_QUIRK_B_INVALID_VBUS_91, but now it is, unless the condition is
taken.
Changing the logic to be as it was before the problematic commit *and*
only conditionally scheduling musb->irq_work resolves the regression:
$ ping 192.168.0.100
...
64 packets transmitted, 64 received, 0% packet loss, time 64475ms
Fixes: 92af4fc6ec33 ("usb: musb: Fix suspend with devices connected for a64")
Signed-off-by: Thomas Petazzoni <[email protected]>
Cc: [email protected]
---
drivers/usb/musb/musb_core.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
index 8f09a387b773..4c8f0112481f 100644
--- a/drivers/usb/musb/musb_core.c
+++ b/drivers/usb/musb/musb_core.c
@@ -2009,9 +2009,8 @@ static void musb_pm_runtime_check_session(struct musb *musb)
schedule_delayed_work(&musb->irq_work,
msecs_to_jiffies(1000));
musb->quirk_retries--;
- break;
}
- fallthrough;
+ break;
case MUSB_QUIRK_B_INVALID_VBUS_91:
if (musb->quirk_retries && !musb->flush_irq_work) {
musb_dbg(musb,
--
2.31.1
On 28/05/2021 16:04:46+0200, Thomas Petazzoni wrote:
> In commit 92af4fc6ec33 ("usb: musb: Fix suspend with devices
> connected for a64"), the logic to support the
> MUSB_QUIRK_B_DISCONNECT_99 quirk was modified to only conditionally
> schedule the musb->irq_work delayed work.
>
> This commit badly breaks ECM Gadget on AM335X. Indeed, with this
> commit, one can observe massive packet loss:
>
> $ ping 192.168.0.100
> ...
> 15 packets transmitted, 3 received, 80% packet loss, time 14316ms
>
> Reverting this commit brings back a properly functioning ECM
> Gadget. An analysis of the commit seems to indicate that a mistake was
> made: the previous code was not falling through into the
> MUSB_QUIRK_B_INVALID_VBUS_91, but now it is, unless the condition is
> taken.
>
> Changing the logic to be as it was before the problematic commit *and*
> only conditionally scheduling musb->irq_work resolves the regression:
>
> $ ping 192.168.0.100
> ...
> 64 packets transmitted, 64 received, 0% packet loss, time 64475ms
>
> Fixes: 92af4fc6ec33 ("usb: musb: Fix suspend with devices connected for a64")
> Signed-off-by: Thomas Petazzoni <[email protected]>
Tested-by: Alexandre Belloni <[email protected]>
> Cc: [email protected]
> ---
> drivers/usb/musb/musb_core.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
> index 8f09a387b773..4c8f0112481f 100644
> --- a/drivers/usb/musb/musb_core.c
> +++ b/drivers/usb/musb/musb_core.c
> @@ -2009,9 +2009,8 @@ static void musb_pm_runtime_check_session(struct musb *musb)
> schedule_delayed_work(&musb->irq_work,
> msecs_to_jiffies(1000));
> musb->quirk_retries--;
> - break;
> }
> - fallthrough;
> + break;
> case MUSB_QUIRK_B_INVALID_VBUS_91:
> if (musb->quirk_retries && !musb->flush_irq_work) {
> musb_dbg(musb,
> --
> 2.31.1
>
--
Alexandre Belloni, co-owner and COO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com
* Alexandre Belloni <[email protected]> [210528 15:35]:
> On 28/05/2021 16:04:46+0200, Thomas Petazzoni wrote:
> > In commit 92af4fc6ec33 ("usb: musb: Fix suspend with devices
> > connected for a64"), the logic to support the
> > MUSB_QUIRK_B_DISCONNECT_99 quirk was modified to only conditionally
> > schedule the musb->irq_work delayed work.
> >
> > This commit badly breaks ECM Gadget on AM335X. Indeed, with this
> > commit, one can observe massive packet loss:
> >
> > $ ping 192.168.0.100
> > ...
> > 15 packets transmitted, 3 received, 80% packet loss, time 14316ms
> >
> > Reverting this commit brings back a properly functioning ECM
> > Gadget. An analysis of the commit seems to indicate that a mistake was
> > made: the previous code was not falling through into the
> > MUSB_QUIRK_B_INVALID_VBUS_91, but now it is, unless the condition is
> > taken.
> >
> > Changing the logic to be as it was before the problematic commit *and*
> > only conditionally scheduling musb->irq_work resolves the regression:
> >
> > $ ping 192.168.0.100
> > ...
> > 64 packets transmitted, 64 received, 0% packet loss, time 64475ms
> >
> > Fixes: 92af4fc6ec33 ("usb: musb: Fix suspend with devices connected for a64")
> > Signed-off-by: Thomas Petazzoni <[email protected]>
> Tested-by: Alexandre Belloni <[email protected]>
Ouch, sorry about that one. And thanks for fixing it:
Acked-by: Tony Lindgren <[email protected]>
On Fri, May 28, 2021 at 04:04:46PM +0200, Thomas Petazzoni wrote:
> In commit 92af4fc6ec33 ("usb: musb: Fix suspend with devices
> connected for a64"), the logic to support the
> MUSB_QUIRK_B_DISCONNECT_99 quirk was modified to only conditionally
> schedule the musb->irq_work delayed work.
>
> This commit badly breaks ECM Gadget on AM335X. Indeed, with this
> commit, one can observe massive packet loss:
>
> $ ping 192.168.0.100
> ...
> 15 packets transmitted, 3 received, 80% packet loss, time 14316ms
>
> Reverting this commit brings back a properly functioning ECM
> Gadget. An analysis of the commit seems to indicate that a mistake was
> made: the previous code was not falling through into the
> MUSB_QUIRK_B_INVALID_VBUS_91, but now it is, unless the condition is
> taken.
>
> Changing the logic to be as it was before the problematic commit *and*
> only conditionally scheduling musb->irq_work resolves the regression:
>
> $ ping 192.168.0.100
> ...
> 64 packets transmitted, 64 received, 0% packet loss, time 64475ms
>
> Fixes: 92af4fc6ec33 ("usb: musb: Fix suspend with devices connected for a64")
> Signed-off-by: Thomas Petazzoni <[email protected]>
> Cc: [email protected]
> ---
> drivers/usb/musb/musb_core.c | 3 +--
> 1 file changed, 1 insertion(+), 2 deletions(-)
>
> diff --git a/drivers/usb/musb/musb_core.c b/drivers/usb/musb/musb_core.c
> index 8f09a387b773..4c8f0112481f 100644
> --- a/drivers/usb/musb/musb_core.c
> +++ b/drivers/usb/musb/musb_core.c
> @@ -2009,9 +2009,8 @@ static void musb_pm_runtime_check_session(struct musb *musb)
> schedule_delayed_work(&musb->irq_work,
> msecs_to_jiffies(1000));
> musb->quirk_retries--;
> - break;
> }
> - fallthrough;
> + break;
> case MUSB_QUIRK_B_INVALID_VBUS_91:
> if (musb->quirk_retries && !musb->flush_irq_work) {
> musb_dbg(musb,
> --
> 2.31.1
Tested-by: Drew Fustini <[email protected]>
This patches fixes the problem on the BeagleBone Black (AM3358) where
the USB gadget interfaces would frequently reset. For example:
configfs-gadget gadget: init rndis
configfs-gadget gadget: RNDIS RX/TX early activation ...
usb0: qlen 10
configfs-gadget gadget: rndis_open
rndis_set_param_medium: 0 4259840
usb0: eth_start
rndis_set_param_dev:
configfs-gadget gadget: set_config: interface 2 (Mass Storage Function) requested delayed status
configfs-gadget gadget: delayed_status count 1
configfs-gadget gadget: reset ncm control 3
configfs-gadget gadget: init ncm ctrl 3
configfs-gadget gadget: notify speed 425984000
configfs-gadget gadget: reset acm ttyGS0
configfs-gadget gadget: activate acm ttyGS0
configfs-gadget gadget: acm ttyGS0 serial state 0000
configfs-gadget gadget: usb_composite_setup_continue
configfs-gadget gadget: usb_composite_setup_continue: Completing delayed status
configfs-gadget gadget: rndis req21.00 v0000 i0000 l24
rndis_msg_parser: RNDIS_MSG_INIT
configfs-gadget gadget: rndis reqa1.01 v0000 i0000 l1025
configfs-gadget gadget: rndis req21.00 v0000 i0000 l32
gen_ndis_query_resp: RNDIS_OID_GEN_PHYSICAL_MEDIUM
configfs-gadget gadget: rndis reqa1.01 v0000 i0000 l1025
configfs-gadget gadget: rndis req21.00 v0000 i0000 l76
gen_ndis_query_resp: RNDIS_OID_802_3_PERMANENT_ADDRESS
configfs-gadget gadget: rndis reqa1.01 v0000 i0000 l1025
configfs-gadget gadget: rndis req21.00 v0000 i0000 l32
gen_ndis_set_resp: RNDIS_OID_GEN_CURRENT_PACKET_FILTER 0000002d
configfs-gadget gadget: rndis reqa1.01 v0000 i0000 l1025
configfs-gadget gadget: init ncm
configfs-gadget gadget: activate ncm
usb1: qlen 10
configfs-gadget gadget: ncm_open
usb1: eth_start
configfs-gadget gadget: reset ncm
usb1: gether_disconnect
configfs-gadget gadget: ncm reqa1.80 v0000 i0003 l28
configfs-gadget gadget: non-CRC mode selected
configfs-gadget gadget: ncm req21.8a v0000 i0003 l0
configfs-gadget gadget: NCM16 selected
configfs-gadget gadget: ncm req21.84 v0000 i0003 l0
configfs-gadget gadget: init ncm
configfs-gadget gadget: activate ncm
usb1: qlen 10
configfs-gadget gadget: ncm_open
usb1: eth_start
configfs-gadget gadget: acm ttyGS0 req21.20 v0000 i0005 l7
configfs-gadget gadget: notify speed 425984000
configfs-gadget gadget: notify connect true
I have posted dmesg log without the patch [1] where this happens often
and the dmesg log with the patch [2] where it does happen at all.
Thank you for fixing this!
Drew
[1] https://gist.github.com/pdp7/bd61e5f78545de182605992254d3eeee
[2] https://gist.github.com/pdp7/95b0f34fa1d423d4764984b400c562cf