Re: [PATCH v2] genirq: bug on inconstent flags and flow handler

[Jason Cooper <jason@lakedaemon.net>]

On Wed, Jul 23, 2014 at 08:49:46PM +0200, Thomas Gleixner wrote:
> On Wed, 23 Jul 2014, Florian Fainelli wrote:
> 
> > It is currently possible for a generic irq chip driver to set IRQ_LEVEL
> > and have its irq flow handler be handle_edge_irq. Setting IRQ_LEVEL in
> > such a case does not make sense, and will actually prevent e.g: the
> > software resend logic from kicking, and potential other problems too.
> > 
> > Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> > ---
> > Changes in v2:
> > - replaced WARN_ON() with BUG_ON() since we really don't want to continue
> >   as suggested by Jason Cooper
> 
> I disagree here. It's not a reason take the machine down. Its good
> enough to WARN. That keeps the machine alive and lets us debug that
> stuff.

Sure, I have no strong opinion either way.  On irc I mentioned that this
condition would most likely only arise when porting to a new SoC, so a
kernel developer that will know what to do.

I know for me personally, I need to be stopped mid-track and forced to
pay attention to something like this.  Otherwise, I'll forget and debug
for hours... :-)

thx,

Jason.