[PATCH 0/2] watchdog: ath79: modernize and broaden build coverage

Rosen Penev posted 2 patches 1 day, 20 hours ago
There is a newer version of this series
drivers/watchdog/Kconfig     |   3 +-
drivers/watchdog/ath79_wdt.c | 248 +++++++++++------------------------
2 files changed, 76 insertions(+), 175 deletions(-)
[PATCH 0/2] watchdog: ath79: modernize and broaden build coverage
Posted by Rosen Penev 1 day, 20 hours ago
Two cleanups for the Atheros AR71XX/AR724X/AR913X watchdog driver.

Patch 1 splits OF out as a hard dependency (the driver has always been
OF-only) and broadens COMPILE_TEST coverage from ARM to all
architectures, so the driver can be built on non-MIPS targets.

Patch 2 moves the driver off the legacy miscdevice/file_operations
interface to the watchdog_device framework.  This drops ~100 lines of
hand-rolled ioctl, open/release, and busy/expect-close bookkeeping in
favor of framework callbacks, fixes get_timeleft() to return seconds
instead of raw clock ticks, and properly propagates -EPROBE_DEFER from
clk_get().  nowayout, the timeout module parameter, and module .owner
are wired through the standard helpers.

Tested on an Archer C7v2; sysupgrade watchdog handoff still works.

Rosen Penev (2):
  watchdog: ath79_wdt: select OF and COMPILE_TEST
  watchdog: ath79: convert to watchdog_device

 drivers/watchdog/Kconfig     |   3 +-
 drivers/watchdog/ath79_wdt.c | 248 +++++++++++------------------------
 2 files changed, 76 insertions(+), 175 deletions(-)

-- 
2.54.0
Re: [PATCH 0/2] watchdog: ath79: modernize and broaden build coverage
Posted by Guenter Roeck 1 day, 19 hours ago
On 5/22/26 16:25, Rosen Penev wrote:
> Two cleanups for the Atheros AR71XX/AR724X/AR913X watchdog driver.
> 
> Patch 1 splits OF out as a hard dependency (the driver has always been
> OF-only) and broadens COMPILE_TEST coverage from ARM to all
> architectures, so the driver can be built on non-MIPS targets.
> 
> Patch 2 moves the driver off the legacy miscdevice/file_operations
> interface to the watchdog_device framework.  This drops ~100 lines of
> hand-rolled ioctl, open/release, and busy/expect-close bookkeeping in
> favor of framework callbacks, fixes get_timeleft() to return seconds
> instead of raw clock ticks, and properly propagates -EPROBE_DEFER from
> clk_get().  nowayout, the timeout module parameter, and module .owner
> are wired through the standard helpers.
> 
> Tested on an Archer C7v2; sysupgrade watchdog handoff still works.
> 
> Rosen Penev (2):
>    watchdog: ath79_wdt: select OF and COMPILE_TEST
>    watchdog: ath79: convert to watchdog_device
> 
>   drivers/watchdog/Kconfig     |   3 +-
>   drivers/watchdog/ath79_wdt.c | 248 +++++++++++------------------------
>   2 files changed, 76 insertions(+), 175 deletions(-)
> 

Is this a resend, ignoring the code review feedback, or v2 without
v2 tag and no change log ?

Since there is neither a v2 tag nor a change log, I assume it is a resend.
I am going to ignore it, assuming that the code review comments have not
been addressed.

Guenter