Re: [PATCH] QStor SATA/RAID driver for 2.6.9-rc3

[Jeff Garzik <jgarzik@pobox.com>]

Mark Lord wrote:
> It's been there since day one.  The interrupt handling sometimes requires
> more functionality than is available at interrupt time, so it uses
> schedule_work to have a bottom half re-run itself from thread context.
> This is needed in the error-processing and hot plug paths.


ewwww  :)  If you find yourself calling your irq path from 
non-irq-context code, back up, you're going down the wrong path.

The usual way to do what you want is either

1) wrap all code that _might_ be called from inside interrupt handler 
inside spin_lock_irqsave() [except when you are in the interrupt 
handler, of course, which is merely spin_lock()

Any code that checks "if (in_interrupt())" should be shot on sight :)


2) have both interrupt and non-interrupt contexts fire a tasklet using 
tasklet_schedule().  Your tasklet function then does the real work.

	Jeff