Messages in this thread | | | Date | Fri, 26 Jun 2009 14:21:54 -0700 | Subject | Re: [PATCH 8/9] x86/apic: match destination id with destination mode | From | Yinghai Lu <> |
| |
On Thu, Jun 25, 2009 at 5:15 PM, Pan, Jacob jun<jacob.jun.pan@intel.com> wrote: > From 791a44040dde6670f90a729cf91ed302d84b875a Mon Sep 17 00:00:00 2001 > From: Jacob Pan <jacob.jun.pan@intel.com> > Date: Tue, 12 May 2009 10:51:08 -0700 > Subject: [PATCH] x86/apic: match destination id with destination mode > > current code assigns logical destination IDs regardless of destination modes, this is not a problem since we only use logical delivery so far. but it could be a problem with platforms only supports physical mode > > Signed-off-by: Jacob Pan <jacob.jun.pan@intel.com> > --- > arch/x86/kernel/apic/io_apic.c | 27 +++++++++++++++++++-------- > 1 files changed, 19 insertions(+), 8 deletions(-) > > diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c > index 4d0216f..c84dc3d 100644 > --- a/arch/x86/kernel/apic/io_apic.c > +++ b/arch/x86/kernel/apic/io_apic.c > @@ -1431,7 +1431,10 @@ int setup_ioapic_entry(int apic_id, int irq, > } else { > entry->delivery_mode = apic->irq_delivery_mode; > entry->dest_mode = apic->irq_dest_mode; > - entry->dest = destination; > + if (apic->irq_dest_mode) > + entry->dest = destination; > + else /* physical ID of BSP */ > + entry->dest = boot_cpu_id; > entry->vector = vector; > } > > @@ -1576,7 +1579,10 @@ static void __init setup_timer_IRQ0_pin(unsigned int apic_id, unsigned int pin, > */ > entry.dest_mode = apic->irq_dest_mode; > entry.mask = 0; /* don't mask IRQ for edge */ > - entry.dest = apic->cpu_mask_to_apicid(apic->target_cpus()); > + if (!apic->irq_dest_mode) > + entry.dest = boot_cpu_id; /* physical apic id */ > + else > + entry.dest = apic->cpu_mask_to_apicid(apic->target_cpus()); > entry.delivery_mode = apic->irq_delivery_mode; > entry.polarity = 0; > entry.trigger = 0; > @@ -2343,7 +2349,8 @@ set_ioapic_affinity_irq_desc(struct irq_desc *desc, const struct cpumask *mask) > dest = set_desc_affinity(desc, mask); > if (dest != BAD_APICID) { > /* Only the high 8 bits are valid. */ > - dest = SET_APIC_LOGICAL_ID(dest); > + if (apic->irq_dest_mode) > + dest = SET_APIC_LOGICAL_ID(dest); > __target_IO_APIC_irq(irq, dest, cfg); > ret = 0; > } > @@ -3273,9 +3280,11 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms > err = assign_irq_vector(irq, cfg, apic->target_cpus()); > if (err) > return err; > - > - dest = apic->cpu_mask_to_apicid_and(cfg->domain, apic->target_cpus()); > - > + if (apic->irq_dest_mode) > + dest = apic->cpu_mask_to_apicid_and(cfg->domain, > + apic->target_cpus()); > + else > + dest = boot_cpu_id; > if (irq_remapped(irq)) { > struct irte irte; > int ir_index; > @@ -3339,8 +3348,10 @@ static int set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) > struct irq_cfg *cfg; > struct msi_msg msg; > unsigned int dest; > - > - dest = set_desc_affinity(desc, mask); > + if (apic->irq_dest_mode) > + dest = set_desc_affinity(desc, mask); > + else > + dest = boot_cpu_id; > if (dest == BAD_APICID) > return -1; >
for physical flat mode
apic->cpu_mask_to_apicid() is
static unsigned int physflat_cpu_mask_to_apicid(const struct cpumask *cpumask) { int cpu;
/* * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ cpu = cpumask_first(cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else return BAD_APICID; }
so this patch will put all ioapic routing to boot_cpu_id what is that? how can you assume that is is apic id.? that is NOT apic id.
the patch is totally wrong.
Nacked-by: Yinghai Lu <yinghai@kernel.org>
YH -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |