Re: [PATCH] irq: provide debug_poll_all_shared_irqs() method under CONFIG_DEBUG_SHIRQ

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

 



* Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Fri, 16 Jan 2009, Ingo Molnar wrote:
> > 
> > The sooner we start seeing these kinds of problems the better - 
> > delaying debug patches to after feature work is never a good idea - 
> > delaying feature changes works much better ;-)
> 
> The problem with this particular debug feature is that when it finds 
> something, the machine usually just hangs because the interrupt handler 
> is just looping over reading some status register that returns 0xff, and 
> thinks that means "interrupt pending".
> 
> Or maybe it's not looping, it's using some value as an index, or just 
> isn't ready to be called at all because it shut down things and as a 
> result it actually crashes with a NULL pointer dereference or something.
> 
> And if you are in the middle of a resume, the result is not pretty. The 
> crash is often totally undebuggable. You can test it by using the 
> 'no_console_suspend' and 'acpi_sleep=s3_bios' kernel parameters, and 
> _maybe_ it will get things working. You're probably more likely to get a 
> working setup if you use VGA text mode rather than FBCON, though.
> 
> In other words: the actual problems you hit tend to be very similar to 
> the nasty problems that DEBUG_SHIRQ can uncover - but with the _normal_ 
> DEBUG_SHIRQ you at least usually have a console, so you can add printk's 
> to see where it happened, and you get oopses etc.
> 
> So I do think this is worth doing as a "make it easier to _see_ all the 
> nasty suspend/resume problems", but at the same time it's going to be 
> really really painful to debug, and it's not an option we can likely 
> ever enable in some distro kernel - because a regular user can't even 
> send a bug-report (just a "resume used to work, now it stopped working" 
> thing).

Yeah, indeed. Perhaps a sysctl that turns it off by default might help? 
Under CONFIG_PM_DEBUG or so.

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-pci" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux