diff options
| author | Marc Zyngier <maz@kernel.org> | 2024-07-05 12:31:53 +0300 |
|---|---|---|
| committer | Thomas Gleixner <tglx@linutronix.de> | 2024-07-15 16:13:55 +0300 |
| commit | 7d2c2048a86477461f7bc75d064579ed349472bc (patch) | |
| tree | 6f4f978783ff9187d2ede67d7eb5f199330d5d24 /drivers | |
| parent | be5e5f3a1120bada0cff1bc84c2a1805da308f6e (diff) | |
| download | linux-7d2c2048a86477461f7bc75d064579ed349472bc.tar.xz | |
irqchip/gic-v4: Always configure affinity on VPE activation
There are currently two paths to set the initial affinity of a VPE:
- at activation time on GICv4 without the stupid VMOVP list, and
on GICv4.1
- at map time for GICv4 with VMOVP list
The latter location may end-up modifying the affinity of VPE that is
currently running, making the results unpredictible.
Instead, unify the two paths, making sure to set the initial affinity only
at activation time.
Reported-by: Nianyao Tang <tangnianyao@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Nianyao Tang <tangnianyao@huawei.com>
Link: https://lore.kernel.org/r/20240705093155.871070-2-maz@kernel.org
Diffstat (limited to 'drivers')
| -rw-r--r-- | drivers/irqchip/irq-gic-v3-its.c | 13 |
1 files changed, 4 insertions, 9 deletions
diff --git a/drivers/irqchip/irq-gic-v3-its.c b/drivers/irqchip/irq-gic-v3-its.c index af5297ef2293..387ac5a31d1d 100644 --- a/drivers/irqchip/irq-gic-v3-its.c +++ b/drivers/irqchip/irq-gic-v3-its.c @@ -1809,13 +1809,9 @@ static void its_map_vm(struct its_node *its, struct its_vm *vm) for (i = 0; i < vm->nr_vpes; i++) { struct its_vpe *vpe = vm->vpes[i]; - struct irq_data *d = irq_get_irq_data(vpe->irq); - /* Map the VPE to the first possible CPU */ - vpe->col_idx = cpumask_first(cpu_online_mask); its_send_vmapp(its, vpe, true); its_send_vinvall(its, vpe); - irq_data_update_effective_affinity(d, cpumask_of(vpe->col_idx)); } } @@ -4582,6 +4578,10 @@ static int its_vpe_irq_domain_activate(struct irq_domain *domain, struct its_vpe *vpe = irq_data_get_irq_chip_data(d); struct its_node *its; + /* Map the VPE to the first possible CPU */ + vpe->col_idx = cpumask_first(cpu_online_mask); + irq_data_update_effective_affinity(d, cpumask_of(vpe->col_idx)); + /* * If we use the list map, we issue VMAPP on demand... Unless * we're on a GICv4.1 and we eagerly map the VPE on all ITSs @@ -4590,9 +4590,6 @@ static int its_vpe_irq_domain_activate(struct irq_domain *domain, if (!gic_requires_eager_mapping()) return 0; - /* Map the VPE to the first possible CPU */ - vpe->col_idx = cpumask_first(cpu_online_mask); - list_for_each_entry(its, &its_nodes, entry) { if (!is_v4(its)) continue; @@ -4601,8 +4598,6 @@ static int its_vpe_irq_domain_activate(struct irq_domain *domain, its_send_vinvall(its, vpe); } - irq_data_update_effective_affinity(d, cpumask_of(vpe->col_idx)); - return 0; } |