[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] reduce useless output from 'd' consolecommand

On 12/6/08 16:50, "Jan Beulich" <jbeulich@xxxxxxxxxx> wrote:

>>>> Keir Fraser <keir.fraser@xxxxxxxxxxxxx> 12.06.08 17:31 >>>
>> I quite like it as it is. The output for a CPU in guest mode is quite short,
>> and outputting printk("CPU%d blah") before doing the IPI also lets us see
>> very clearly if a CPU is probably not responding to IPI. I think changing
>> console_start_log_everything() to console_start_sync() would be a good idea,
>> at least for key 'd'.
> The part about placement of the printk() is certainly reasonable, but
> the guest mode output being short I have to question - its stack trace
> can be as long as the hypervisor's. And clearly printing hypervisor
> context when guest execution was interrupted is pointless.

I added console_start/end_sync() to the key handler, so that at least
reduces this from an issue resulting in annoying loss of critical debug data
down to a minor quibble. And now you're *guaranteed* not to lose any debug
data (so long as all CPUs are alive) which is even better than thinning down
the output and hoping for the best.

 -- Keir

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.