[PATCH stable 5.10] usb: dwc3: Remove DWC3 locking during gadget suspend/resume

[PATCH stable 5.10] usb: dwc3: Remove DWC3 locking during gadget suspend/resume

[Selvarasu Ganesan]

Dear stable team,


Patch : usb: dwc3: Remove DWC3 locking during gadget suspend/resume

Commit id:5265397f94424eaea596026fd34dc7acf474dcec

This patch fixes a critical bug in the dwc3 driver that was introduced 
by commit 
(https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24 
<https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24>) 
in the 5.10 kernel series.

The bug causes the below kernel crash (Added usleep in atomic context as 
part of above patch) under dwc3 suspend/resume scenarios.

35.829644] [6: kworker/6:1: 68] BUG: scheduling while atomic: 
kworker/6:1/68/0x00000002

[ 35.829946] [6: kworker/6:1: 68] CPU: 6 PID: 68 Comm: kworker/6:1 
Tainted: G C E 5.10.236-android13-4 #1

[ 35.830010] [6: kworker/6:1: 68] Call trace:

[ 35.830024] [6: kworker/6:1: 68] dump_backtrace.cfi_jt+0x0/0x8

[ 35.830034] [6: kworker/6:1: 68] show_stack+0x1c/0x2c

[ 35.830044] [6: kworker/6:1: 68] dump_stack_lvl+0xd8/0x134

[ 35.830053] [6: kworker/6:1: 68] __schedule_bug+0x80/0xbc

[ 35.830062] [6: kworker/6:1: 68] __schedule+0x55c/0x7e8

[ 35.830068] [6: kworker/6:1: 68] schedule+0x80/0x100

[ 35.830077] [6: kworker/6:1: 68] schedule_hrtimeout_range_clock+0xa8/0x11c

[ 35.830083] [6: kworker/6:1: 68] usleep_range+0x68/0xa4

[ 35.830093] [6: kworker/6:1: 68] dwc3_gadget_run_stop+0x170/0x448

[ 35.830099] [6: kworker/6:1: 68] dwc3_gadget_resume+0x4c/0xdc

[ 35.830108] [6: kworker/6:1: 68] dwc3_resume_common+0x6c/0x23c

[ 35.830115] [6: kworker/6:1: 68] dwc3_runtime_resume+0x40/0xcc

[ 35.830123] [6: kworker/6:1: 68] pm_generic_runtime_resume+0x48/0x88

[ 35.830131] [6: kworker/6:1: 68] __rpm_callback+0x94/0x420

The patch(5265397f9442) for this fix was originally merged in the below 
commit:

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec 
<https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec>

  Please apply this patch to the stable 5.10 kernel to prevent this BUG.

Additionally the below patch also required to avoid dead lock that 
introduced by the abovepatch (5265397f9442) in 5.10 stable kernel.

Patch:usb: dwc3: core: remove lock of otg mode during gadget 
suspend/resume to avoid deadlock

Commit id:7838de15bb700c2898a7d741db9b1f3cbc86c136

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=7838de15bb700c2898a7d741db9b1f3cbc86c136 
<https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=7838de15bb700c2898a7d741db9b1f3cbc86c136>

Thanks,

Selva

Re: [PATCH stable 5.10] usb: dwc3: Remove DWC3 locking during gadget suspend/resume

[Greg KH]

On Wed, Aug 20, 2025 at 07:41:44PM +0530, Selvarasu Ganesan wrote:
> Dear stable team,
> 
> 
> Patch : usb: dwc3: Remove DWC3 locking during gadget suspend/resume
> 
> Commit id:5265397f94424eaea596026fd34dc7acf474dcec
> 
> This patch fixes a critical bug in the dwc3 driver that was introduced 
> by commit 
> (https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24 
> <https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24>) 
> in the 5.10 kernel series.

But that is not what the dwc3 patch says it does, how do we know it is
ok to backport this?

> The bug causes the below kernel crash (Added usleep in atomic context as 
> part of above patch) under dwc3 suspend/resume scenarios.
> 
> 35.829644] [6: kworker/6:1: 68] BUG: scheduling while atomic: 
> kworker/6:1/68/0x00000002
> 
> [ 35.829946] [6: kworker/6:1: 68] CPU: 6 PID: 68 Comm: kworker/6:1 
> Tainted: G C E 5.10.236-android13-4 #1
> 
> [ 35.830010] [6: kworker/6:1: 68] Call trace:
> 
> [ 35.830024] [6: kworker/6:1: 68] dump_backtrace.cfi_jt+0x0/0x8
> 
> [ 35.830034] [6: kworker/6:1: 68] show_stack+0x1c/0x2c
> 
> [ 35.830044] [6: kworker/6:1: 68] dump_stack_lvl+0xd8/0x134
> 
> [ 35.830053] [6: kworker/6:1: 68] __schedule_bug+0x80/0xbc
> 
> [ 35.830062] [6: kworker/6:1: 68] __schedule+0x55c/0x7e8
> 
> [ 35.830068] [6: kworker/6:1: 68] schedule+0x80/0x100
> 
> [ 35.830077] [6: kworker/6:1: 68] schedule_hrtimeout_range_clock+0xa8/0x11c
> 
> [ 35.830083] [6: kworker/6:1: 68] usleep_range+0x68/0xa4
> 
> [ 35.830093] [6: kworker/6:1: 68] dwc3_gadget_run_stop+0x170/0x448
> 
> [ 35.830099] [6: kworker/6:1: 68] dwc3_gadget_resume+0x4c/0xdc
> 
> [ 35.830108] [6: kworker/6:1: 68] dwc3_resume_common+0x6c/0x23c
> 
> [ 35.830115] [6: kworker/6:1: 68] dwc3_runtime_resume+0x40/0xcc
> 
> [ 35.830123] [6: kworker/6:1: 68] pm_generic_runtime_resume+0x48/0x88
> 
> [ 35.830131] [6: kworker/6:1: 68] __rpm_callback+0x94/0x420
> 
> The patch(5265397f9442) for this fix was originally merged in the below 
> commit:
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec 
> <https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec>
> 
>   Please apply this patch to the stable 5.10 kernel to prevent this BUG.
> 
> Additionally the below patch also required to avoid dead lock that 
> introduced by the abovepatch (5265397f9442) in 5.10 stable kernel.
> 
> Patch:usb: dwc3: core: remove lock of otg mode during gadget 
> suspend/resume to avoid deadlock
> 
> Commit id:7838de15bb700c2898a7d741db9b1f3cbc86c136

Can you please submit a backported, and tested, series of patches here
so that we know that we got the correct ones, and we have your
signed-off-by on them to show you tested them?

thanks,

greg k-h

Re: [PATCH stable 5.10] usb: dwc3: Remove DWC3 locking during gadget suspend/resume

[Selvarasu Ganesan]

On 8/20/2025 7:56 PM, Greg KH wrote:
> On Wed, Aug 20, 2025 at 07:41:44PM +0530, Selvarasu Ganesan wrote:
>> Dear stable team,
>>
>>
>> Patch : usb: dwc3: Remove DWC3 locking during gadget suspend/resume
>>
>> Commit id:5265397f94424eaea596026fd34dc7acf474dcec
>>
>> This patch fixes a critical bug in the dwc3 driver that was introduced
>> by commit
>> (https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24
>> <https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/drivers/usb/dwc3/gadget.c?h=v5.10.240&id=90e2820c6c30db2427d020d344dfca7de813bd24>)
>> in the 5.10 kernel series.
> But that is not what the dwc3 patch says it does, how do we know it is
> ok to backport this?
Hi Greg,

We agree that the dwc3 patch (5265397f9442 : usb: dwc3: Remove DWC3 
locking during gadget suspend/resume ) was missing Fixes/stable tags, 
which led to confusion about its backport.
However, these tags should have been included by the author, as the 
patch fixed a critical bug.

Furthermore, a this USB BUG observed easily in the USB reconnect case 
where the device enters suspend/resume mode with current 5.10 stable kernel.


>
>> The bug causes the below kernel crash (Added usleep in atomic context as
>> part of above patch) under dwc3 suspend/resume scenarios.
>>
>> 35.829644] [6: kworker/6:1: 68] BUG: scheduling while atomic:
>> kworker/6:1/68/0x00000002
>>
>> [ 35.829946] [6: kworker/6:1: 68] CPU: 6 PID: 68 Comm: kworker/6:1
>> Tainted: G C E 5.10.236-android13-4 #1
>>
>> [ 35.830010] [6: kworker/6:1: 68] Call trace:
>>
>> [ 35.830024] [6: kworker/6:1: 68] dump_backtrace.cfi_jt+0x0/0x8
>>
>> [ 35.830034] [6: kworker/6:1: 68] show_stack+0x1c/0x2c
>>
>> [ 35.830044] [6: kworker/6:1: 68] dump_stack_lvl+0xd8/0x134
>>
>> [ 35.830053] [6: kworker/6:1: 68] __schedule_bug+0x80/0xbc
>>
>> [ 35.830062] [6: kworker/6:1: 68] __schedule+0x55c/0x7e8
>>
>> [ 35.830068] [6: kworker/6:1: 68] schedule+0x80/0x100
>>
>> [ 35.830077] [6: kworker/6:1: 68] schedule_hrtimeout_range_clock+0xa8/0x11c
>>
>> [ 35.830083] [6: kworker/6:1: 68] usleep_range+0x68/0xa4
>>
>> [ 35.830093] [6: kworker/6:1: 68] dwc3_gadget_run_stop+0x170/0x448
>>
>> [ 35.830099] [6: kworker/6:1: 68] dwc3_gadget_resume+0x4c/0xdc
>>
>> [ 35.830108] [6: kworker/6:1: 68] dwc3_resume_common+0x6c/0x23c
>>
>> [ 35.830115] [6: kworker/6:1: 68] dwc3_runtime_resume+0x40/0xcc
>>
>> [ 35.830123] [6: kworker/6:1: 68] pm_generic_runtime_resume+0x48/0x88
>>
>> [ 35.830131] [6: kworker/6:1: 68] __rpm_callback+0x94/0x420
>>
>> The patch(5265397f9442) for this fix was originally merged in the below
>> commit:
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec
>> <https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/drivers/usb/dwc3?h=next-20250820&id=5265397f94424eaea596026fd34dc7acf474dcec>
>>
>>    Please apply this patch to the stable 5.10 kernel to prevent this BUG.
>>
>> Additionally the below patch also required to avoid dead lock that
>> introduced by the abovepatch (5265397f9442) in 5.10 stable kernel.
>>
>> Patch:usb: dwc3: core: remove lock of otg mode during gadget
>> suspend/resume to avoid deadlock
>>
>> Commit id:7838de15bb700c2898a7d741db9b1f3cbc86c136
> Can you please submit a backported, and tested, series of patches here
> so that we know that we got the correct ones, and we have your
> signed-off-by on them to show you tested them?

Sure, We will post patches that include backported based on the 5.10 
stable kernel.

Thanks,
Selva
>
> thanks,
>
> greg k-h
>