Re: [PATCH v2 7/7] hw: Simplify using sysbus_init_irqs() [manual]

["Philippe Mathieu-Daudé" <philmd@linaro.org>]

On 1/6/23 12:19, Peter Maydell wrote:
> On Thu, 1 Jun 2023 at 10:48, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>>
>> Audit the sysbus_init_irq() calls and manually convert
>> to sysbus_init_irqs() when a loop is involved.
>>
>> In omap2_intc_init(), the parent_intr[] array contains
>> 2 elements: use ARRAY_SIZE() to iterate over.
>>
>> Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> 
>> diff --git a/hw/timer/renesas_tmr.c b/hw/timer/renesas_tmr.c
>> index c15f654738..dd2929d6e7 100644
>> --- a/hw/timer/renesas_tmr.c
>> +++ b/hw/timer/renesas_tmr.c
>> @@ -428,17 +428,14 @@ static void rtmr_init(Object *obj)
>>   {
>>       SysBusDevice *d = SYS_BUS_DEVICE(obj);
>>       RTMRState *tmr = RTMR(obj);
>> -    int i;
>>
>>       memory_region_init_io(&tmr->memory, OBJECT(tmr), &tmr_ops,
>>                             tmr, "renesas-tmr", 0x10);
>>       sysbus_init_mmio(d, &tmr->memory);
>>
>> -    for (i = 0; i < ARRAY_SIZE(tmr->ovi); i++) {
>> -        sysbus_init_irq(d, &tmr->cmia[i]);
>> -        sysbus_init_irq(d, &tmr->cmib[i]);
>> -        sysbus_init_irq(d, &tmr->ovi[i]);
>> -    }
>> +    sysbus_init_irqs(d, tmr->cmia, ARRAY_SIZE(tmr->cmia));
>> +    sysbus_init_irqs(d, tmr->cmib, ARRAY_SIZE(tmr->cmib));
>> +    sysbus_init_irqs(d, tmr->ovi, ARRAY_SIZE(tmr->ovi));
> 
> Doesn't this change the order of the IRQs? Previously
> we had channel 0 CMIA, channel 0 CMIA, channel 0 OVI,
> channel 1 CMIA, channel 1 CMIB, channel 1 OVI. Now
> we have channel 0 CMIA, channel 1 CMIA, channel 0 CMIB,
> channel 1 CMIB, channel 0 OVI, channel 1 OVI. So they'll
> get miswired in the board code now...

Oops, good catch. Good case to show the sysbus_init_irq() is
fragile. At least named_gpio are clearer.

I'm surprised because v1 series passed CI, but looking closer
the RX tests isn't run. And now I see commit e5d402b28f
("tests/acceptance: disable machine_rx_gdbsim on GitLab").

"Hopefully we can re-enable both
     once the serial timing patches have been added."

Alex, are these patches in the tree now?