| Message ID | 20200318151651.2510-1-rasmus.villemoes@prevas.dk |
|---|---|
| State | New |
| Headers | show |
| Series | [v2] powerpc: allow opting out of WATCHDOG_RESET() from timer interrupt | expand |
Dear Rasmus Villemoes, In message <20200318151651.2510-1-rasmus.villemoes at prevas.dk> you wrote: > When using CONFIG_(SPL_)WDT, the watchdog_reset function is a lot more > complicated than just poking a few SOC-specific registers - it > involves accessing all kinds of global data, and if the interrupt > happens at the wrong time (say, in the middle of an WATCHDOG_RESET() > call from ordinary code), that can end up corrupting said global data. > > Also, having WATCHDOG_RESET() called automatically from the timer > interrupt runs counter to the idea of a watchdog device - if the board > runs into an infinite loops with interrupts still enabled, the > watchdog will never fire. > > Allow the board to opt out of this behaviour by setting > CONFIG_SYS_WATCHDOG_FREQ to 0 - as that setting is currently > nonsensical, it cannot affect any existing boards. > > Signed-off-by: Rasmus Villemoes <rasmus.villemoes at prevas.dk> If such a change gets implemented, it should at least be made consistently, i. e. not only in arch/powerpc/lib/interrupts.c but also in the timer_interrupt() handler in drivers/timer/mpc83xx_timer.c and in dtimer_interrupt() in arch/m68k/lib/time.c and wherever else similar code may exist. Only changing one plase is not a good idea. And please add some documentation also to the README. Thanks. Best regards, Wolfgang Denk
diff --git a/arch/powerpc/lib/interrupts.c b/arch/powerpc/lib/interrupts.c index 64ee0cc210..b213e40578 100644 --- a/arch/powerpc/lib/interrupts.c +++ b/arch/powerpc/lib/interrupts.c @@ -79,7 +79,12 @@ void timer_interrupt(struct pt_regs *regs) timestamp++; #if defined(CONFIG_WATCHDOG) || defined (CONFIG_HW_WATCHDOG) - if ((timestamp % (CONFIG_SYS_WATCHDOG_FREQ)) == 0) + /* + * Setting CONFIG_SYS_WATCHDOG_FREQ to 0 can be used to + * disable calling WATCHDOG_RESET() from the timer interrupt + * function. + */ + if (CONFIG_SYS_WATCHDOG_FREQ && (timestamp % (CONFIG_SYS_WATCHDOG_FREQ)) == 0) WATCHDOG_RESET (); #endif /* CONFIG_WATCHDOG || CONFIG_HW_WATCHDOG */
When using CONFIG_(SPL_)WDT, the watchdog_reset function is a lot more complicated than just poking a few SOC-specific registers - it involves accessing all kinds of global data, and if the interrupt happens at the wrong time (say, in the middle of an WATCHDOG_RESET() call from ordinary code), that can end up corrupting said global data. Also, having WATCHDOG_RESET() called automatically from the timer interrupt runs counter to the idea of a watchdog device - if the board runs into an infinite loops with interrupts still enabled, the watchdog will never fire. Allow the board to opt out of this behaviour by setting CONFIG_SYS_WATCHDOG_FREQ to 0 - as that setting is currently nonsensical, it cannot affect any existing boards. Signed-off-by: Rasmus Villemoes <rasmus.villemoes at prevas.dk> --- v2: add comment explaining the semantics of setting CONFIG_SYS_WATCHDOG_FREQ to 0. arch/powerpc/lib/interrupts.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-)