Re: [PATCH 0/4] Add support for Kernel WDT

[Jon Hunter <jonathanh@nvidia.com>]

Hi Daniel,

On 12/05/2026 14:56, Jon Hunter wrote:
> 
> On 07/05/2026 16:45, Kartik Rajput wrote:
>> Tegra186 and later SoCs support multiple watchdog timers in the TKE
>> block, the driver so far only uses WDT0. This series extends it to
>> make use of the additional instances and reserve the first available WDT
>> for Kernel.
>>
>> Patch 1 fixes a read-modify-write bug in tegra186_wdt_enable()
>> that would otherwise make registering more than one watchdog unsafe.
>>
>> Patch 2 corrects num_wdts for Tegra186 and Tegra234. WDT2 is connected
>> to the Audio Processing Engine (APE) and cannot be accessed from Linux,
>> so only WDT0 and WDT1 are available.
>>
>> Patch 3 iterates over all WDTs and registers the ones whose TKE SCR
>> firewall grants OS access.
>>
>> Patch 4 reserves the first accessible watchdog as a kernel-only
>> watchdog, arms it at probe with a 120 s timeout, and pets it from
>> the driver's IRQ handler, it is not exposed to userspace, so the
>> system can still reset if the kernel hangs before userspace opens a
>> watchdog.
>>
>> Kartik Rajput (4):
>>    clocksource/drivers/timer-tegra186: Fix support for multiple watchdog
>>      instances
>>    clocksource/drivers/timer-tegra186: Correct num_wdts for Tegra186 and
>>      Tegra234
>>    clocksource/drivers/timer-tegra186: Register all accessible watchdog
>>      timers
>>    clocksource/drivers/timer-tegra186: Reserve and service a kernel
>>      watchdog
>>
>>   drivers/clocksource/timer-tegra186.c | 122 +++++++++++++++++++++++----
>>   1 file changed, 104 insertions(+), 18 deletions(-)
>>
> 
> For the series ...
> 
> Reviewed-by: Jon Hunter <jonathanh@nvidia.com>


Are you OK to pick this up? We would like to get this into -next if 
there are no objections.

Thanks
Jon

-- 
nvpublic