Re: [PATCH] Xen: Force non-irq keyhandler to be run in tasklet when receive a debugkey from serial port

["Lan, Tianyu" <tianyu.lan@intel.com>]

On 10/24/2016 6:53 PM, Jan Beulich wrote:
>>>> On 22.10.16 at 13:23, <tianyu.lan@intel.com> wrote:
>> __serial_rx() runs in either irq handler or timer handler and non-irq
>> keyhandler should not run in these contexts. So always force non-irq
>> keyhandler to run in tasklet when receive a debugkey from serial port
>>
>> Signed-off-by: Lan Tianyu <tianyu.lan@intel.com>
>> ---
>>  xen/drivers/char/console.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/xen/drivers/char/console.c b/xen/drivers/char/console.c
>> index b0f74ce..184b523 100644
>> --- a/xen/drivers/char/console.c
>> +++ b/xen/drivers/char/console.c
>> @@ -347,7 +347,7 @@ static void switch_serial_input(void)
>>  static void __serial_rx(char c, struct cpu_user_regs *regs)
>>  {
>>      if ( xen_rx )
>> -        return handle_keypress(c, regs, !in_irq());
>> +        return handle_keypress(c, regs, true);
>
> Together with one of your earlier patches having got reverted, I
> think we need to take a step back here instead of going back to
> what was requested to be changed from v2 of the original patch.
> In particular I assume that the problem you're trying to address is
> not limited to dump_timerq() - at least dump_runq() should be as
> problematic on many-CPU systems.

I think the issue here is that my previous patch commit
610b4eda2c("keyhandler: rework process of nonirq keyhandler") makes
non-irq keyhandler run in irq context. This is caused by input param
"!in_irq()" which is false in irq context. handle_keypress() runs 
keyhandler synchronically. This patch fixes the issue.

>
> I think (and I vaguely recall possibly having said so during earlier
> review) that dump functions the output of which depends on CPU
> count should get modeled after dump_registers(), and it might be
> worth abstracting this in keyhandler.c.

Yes, but this sounds like a new feature or framework rework rather than 
a fix patch.


> In any case quite likely the
> other patch of yours (which the one here basically modifies) may
> then also want to be reverted.

I think patch "timer: process softirq during dumping timer"
does right thing. The issue is triggered by previous patch.


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel