On Sat, Sep 25, 2010 at 06:32,  <[email protected]> wrote:
> Modified: trunk/drivers/char/bfin_sport.c (9170 => 9171)
>
> +     int irq_notfreed;

in general, "no" values are a bad idea.  it can often be confusing
when reading the code when you start negating the negative.  i know
reading things like "if (!irq_nofreed)" makes me re-read it a few
times.  so please change this to "irq_freed" and fix up the logic
accordingly.

then again, why is this even needed ?  why do you need to free the
irqs in NDSO mode when doing a read/write ?  what if someone opens the
device, does some work in NDSO mode, then switches to a different mode
without closing/opening the device in between ?  wont this break ?

if you simply need the IRQs to not fire whilst in NDSO mode and doing
a read/write, then it's better to disable the IRQs while doing the
read/write and then re-enabling them when done.

you should also add a comment as to why you need NDSO mode to muck
with the IRQs like this ...
-mike
_______________________________________________
Linux-kernel-commits mailing list
[email protected]
https://blackfin.uclinux.org/mailman/listinfo/linux-kernel-commits

Reply via email to