Patch from Kenneth Tan
The get_irqnr_and_base subroutine of ixp4xx does not take interrupt 0 condition into account properly. We should not perform "subs" here. The Z flag will be set when interrupt 0 occur, which resulting "movne r1, sp" in the caller routine (irq_handler) not being executed.
When interrupt 0 occur:
o if CONFIG_CPU_IXP46X is not set, "subs" will set the Z flag and return
o if CONFIG_CPU_IXP46X is set, codes in upper interrupt handling will be trigerred. But since this is not supper interrupt, the "cmp" in the upper interrupt handling portion will set the Z flag and return
Signed-off-by: Kenneth Tan <chong.yin.tan@intel.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
ldr \irqstat, =(IXP4XX_INTC_BASE_VIRT+IXP4XX_ICIP_OFFSET)
ldr \irqstat, [\irqstat] @ get interrupts
cmp \irqstat, #0
- beq 1001f
+ beq 1001f @ upper IRQ?
clz \irqnr, \irqstat
mov \base, #31
- subs \irqnr, \base, \irqnr
+ sub \irqnr, \base, \irqnr
+ b 1002f @ lower IRQ being
+ @ handled
1001:
/*
* IXP465 has an upper IRQ status register
*/
#if defined(CONFIG_CPU_IXP46X)
- bne 1002f
ldr \irqstat, =(IXP4XX_INTC_BASE_VIRT+IXP4XX_ICIP2_OFFSET)
ldr \irqstat, [\irqstat] @ get upper interrupts
mov \irqnr, #63
clz \irqstat, \irqstat
cmp \irqstat, #32
subne \irqnr, \irqnr, \irqstat
-1002:
#endif
+1002:
.endm