[RFC PATCH v9 1/4] ARM: gic: consolidate PPI handling

[Marc Zyngier]

On 22/07/11 10:42, Russell King - ARM Linux wrote:
> On Thu, Jul 21, 2011 at 05:57:25PM +0100, Marc Zyngier wrote:
>> @@ -256,12 +260,33 @@ void __init gic_cascade_irq(unsigned int gic_nr, unsigned int irq)
>>  	irq_set_chained_handler(irq, gic_handle_cascade_irq);
>>  }
>>  
>> +static unsigned int gic_nr_ppis, gic_ppi_base;
>> +
>> +#define PPI_IRQACT(nr)						\
>> +	{							\
>> +		.handler	= percpu_timer_handler,		\
> 
> Won't this break on non-SMP non-localtimer builds?

Good point. I'd probably need to #ifdef that (as it was #ifdef'ed in the
original code), and remove it in patch #2.

>> +		.flags		= IRQF_PERCPU | IRQF_TIMER,	\
>> +		.irq		= nr,				\
>> +		.name		= "PPI-" # nr,			\
>> +	}
>> +
>> +static struct irqaction ppi_irqaction_template[16] __initdata = {
>> +	PPI_IRQACT(0),  PPI_IRQACT(1),  PPI_IRQACT(2),  PPI_IRQACT(3),
>> +	PPI_IRQACT(4),  PPI_IRQACT(5),  PPI_IRQACT(6),  PPI_IRQACT(7),
>> +	PPI_IRQACT(8),  PPI_IRQACT(9),  PPI_IRQACT(10), PPI_IRQACT(11),
>> +	PPI_IRQACT(12), PPI_IRQACT(13), PPI_IRQACT(14), PPI_IRQACT(15),
>> +};
>> +
>> +static struct irqaction *ppi_irqaction;
>> +
>>  static void __init gic_dist_init(struct gic_chip_data *gic,
>>  	unsigned int irq_start)
>>  {
>>  	unsigned int gic_irqs, irq_limit, i;
>>  	void __iomem *base = gic->dist_base;
>>  	u32 cpumask = 1 << smp_processor_id();
>> +	u32 dist_ctr, nrcpus;
> 
> nrcpus doesn't seem to be used.  With that eliminated, dist_ctr doesn't
> seem to have much purpose.
> 
>> +	u32 nrppis = 0, ppi_base = 0;
> 
> Might be better to move this inside the "if (gic == &gic_data[0]) {" block,
> along with the printk too.
> 
>>  
>>  	cpumask |= cpumask << 8;
>>  	cpumask |= cpumask << 16;
>> @@ -272,11 +297,38 @@ static void __init gic_dist_init(struct gic_chip_data *gic,
>>  	 * Find out how many interrupts are supported.
>>  	 * The GIC only supports up to 1020 interrupt sources.
>>  	 */
>> -	gic_irqs = readl_relaxed(base + GIC_DIST_CTR) & 0x1f;
>> -	gic_irqs = (gic_irqs + 1) * 32;
>> +	dist_ctr = readl_relaxed(base + GIC_DIST_CTR);
>> +	gic_irqs = ((dist_ctr & 0x1f) + 1) * 32;
>>  	if (gic_irqs > 1020)
>>  		gic_irqs = 1020;
>>  
>> +	/* Find out how many CPUs are supported (8 max). */
>> +	nrcpus = ((dist_ctr >> 5) & 7) + 1;
> 
> As mentioned above, the above change can be killed because it doesn't
> alter anything which is used.

Actually, all this really belongs to patch #2. Will move it there.

>> +
>> +	/*
>> +	 * Nobody would be insane enough to use PPIs on a secondary
>> +	 * GIC, right?
>> +	 */
>> +	if (gic == &gic_data[0]) {
>> +		nrppis = 16 - (irq_start & 15);
>> +		ppi_base = gic->irq_offset + 32 - nrppis;
>> +		ppi_irqaction = kzalloc(sizeof(*ppi_irqaction) * nrppis,
>> +					 GFP_KERNEL);
>> +		if (!ppi_irqaction) {
>> +			pr_err("GIC: Can't allocate PPI memory");
>> +			nrppis = 0;
>> +			ppi_base = 0;
>> +		}
>> +
>> +		for (i = 0; i < nrppis; i++)
>> +			ppi_irqaction[i] = ppi_irqaction_template[i + (ppi_base & 15)];
>> +		gic_nr_ppis = nrppis;
>> +		gic_ppi_base = ppi_base;
> 
> Would:
> 		ppi_irqaction = kmemdup(ppi_irqaction_template,
> 					sizeof(*ppi_irqaction) * nrppis,
> 					GFP_KERNEL);
> 		if (ppi_irqaction) {
> 			gic_nr_ppis = nrppis;
> 			gic_ppi_base = ppi_base;
> 		}
> 
> be a shorter way to write what you have above?

Indeed. Will fix that in the next version.

Thanks for reviewing.

	M.
-- 
Jazz is not dead. It just smells funny...