diff options
| author | Geert Uytterhoeven <geert@linux-m68k.org> | 2013-11-15 14:15:07 -0800 |
|---|---|---|
| committer | Thomas Gleixner <tglx@linutronix.de> | 2013-11-19 19:06:41 +0100 |
| commit | 6bab2c613d7fa70bb8514f89ab7455ede717142b (patch) | |
| tree | 444873d38abef1005be025a7ae2962d0ba52818e /kernel/irq | |
| parent | 00d1a39e69d5afa7523dad515a05b21abd17c389 (diff) | |
| download | linux-6bab2c613d7fa70bb8514f89ab7455ede717142b.tar.bz2 | |
genirq: Correct fuzzy and fragile IRQ_RETVAL() definition
commit bedd30d986a0 ("genirq: make irqreturn_t an enum") blindly replaced
"0" by "IRQ_NONE" in the "IRQ_RETVAL(x)" macro definition.
However, as "x" is a condition, "0" meant "boolean false", not an
irqreturn_t value.
All of this worked, and kept working after the addition of IRQ_WAKE_THREAD,
as
- both "boolean false" and "IRQ_NONE" are "0" (for the comparison),
- "boolean true" and "boolean false" nicely map to the correct values of
"IRQ_HANDLED" and "IRQ_NONE" (for the return value).
Correct the macro definition for clarity and future-proofness.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Diffstat (limited to 'kernel/irq')
0 files changed, 0 insertions, 0 deletions