On 9/17/26 10:43, Patrice CHOTARD wrote:
>
>
> On 9/11/26 14:38, Patrice CHOTARD wrote:
>>
>>
>> On 9/7/26 11:07, Mattijs Korpershoek wrote:
>>> Hi Patrice,
>>>
>>> Thank you for the patch and sorry for the review delays.
>>>
>>> On Mon, Aug 17, 2026 at 17:41, Patrice Chotard
>>> <[email protected]> wrote:
>>>
>>>> schedule was added in sleep_thread() by commit 4b6a3e860878
>>>> ("usb: gadget: f_mass_storage: Add schedule() in sleep_thread()").
>>>> to ensure that watchdog is still reset periodically even on platform
>>>> that doesn't implement g_dnl_board_usb_cable_connected() and in case USB
>>>> cable is not connected.
>>>>
>>>> Instead of calling schedule() for each for() loop iteration, call
>>>> schedule() only in case g_dnl_board_usb_cable_connected() is not
>>>> overloaded, in this particular case, g_dnl_board_usb_cable_connected()'s
>>>> return value is -EOPNOTSUPP.
>>>>
>>>> Signed-off-by: Patrice Chotard <[email protected]>
>>>> ---
>>>> drivers/usb/gadget/f_mass_storage.c | 7 +++++--
>>>> 1 file changed, 5 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/usb/gadget/f_mass_storage.c
>>>> b/drivers/usb/gadget/f_mass_storage.c
>>>> index 621852080e2..f467693d2cb 100644
>>>> --- a/drivers/usb/gadget/f_mass_storage.c
>>>> +++ b/drivers/usb/gadget/f_mass_storage.c
>>>> @@ -651,6 +651,7 @@ static void busy_indicator(void)
>>>> static int sleep_thread(struct fsg_common *common)
>>>> {
>>>> int i = 0, k = 0;
>>>> + int ret;
>>>>
>>>> /* Wait until a signal arrives or we are woken up */
>>>> for (;;) {
>>>> @@ -673,13 +674,15 @@ static int sleep_thread(struct fsg_common *common)
>>>> return -EPIPE;
>>>>
>>>> /* Check cable connection */
>>>> - if (!g_dnl_board_usb_cable_connected())
>>>> + ret = g_dnl_board_usb_cable_connected();
>>>> + if (!ret)
>>>> return -EIO;
>>>> + if (ret == -EOPNOTSUPP)
>>>> + schedule();
>>>
>>> What happens on boards that override g_dnl_board_usb_cable_connected()
>>> that return a positive value?
>>>
>>> Per my understanding, schedule() will never be called, and the watchdog
>>> will be triggered at some point, no?
>>
>> Hi Mattijs,
>>
>> In case g_dnl_board_usb_cable_connected() returns 1 (it's currently the case
>> on STM32MP157c-DK2 board), we stay inside sleep_thread() for(;;) loop until
>> common->thread_wakeup_needed is set to 1 by wakeup_thread().
>>
>> wakeup_thread() is called periodically (around every second, it was what i
>> observed
>> during testing) by either bulk_out_complete() or bulk_in_complete().
>>
>> Then we came back in fsg_main_thread() which was sleep_thread() caller,
>> and came back in while(1) loop of do_usb_mass_storage() where schedule() is
>> called.
>>
>> So watchdog is reset periodically.
>>
>> Patrice
>>>
>>>>
>>>> k = 0;
>>>> }
>>>>
>>>> - schedule();
>>>> dm_usb_gadget_handle_interrupts(udcdev);
>>>> }
>>>> common->thread_wakeup_needed = 0;
>>>>
>>>> --
>>>> 2.43.0
>>
> Hi Mattijs
>
> We got a discussion with Marek on the previous version of this patch [1].
> Marek's proposal is more simple/efficient than this current patch.
>
> This patch will be abandoned.
>
> Thanks
> Patrice
>
> [1]
> https://patchwork.ozlabs.org/project/uboot/patch/[email protected]/
For information
I submitted Marek's proposal:
https://patchwork.ozlabs.org/project/uboot/patch/20260917-ums_increase_ctrl-c_responsiveness-v1-1-95305f044...@foss.st.com/
Thanks
Patrice
