tmakin/ack-tegra
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

FROMLIST: KVM: arm64: vgic: Explicitly implement vgic_dist::ready ordering

Browse Source 
In preparation to remove synchronize_srcu() from MMIO registration,
remove the distributor's dependency on this implicit barrier by
direct acquire-release synchronization on the flag write and its
lock-free check.

Bug: 395485007
Bug: 357781595
Link: https://lore.kernel.org/all/20250624092256.1105524-3-keirf@google.com/

Change-Id: Id23675e275d78a387d5cb884c4c7da9a5aefe7fe
Signed-off-by: Keir Fraser <keirf@google.com>
This commit is contained in:
Keir Fraser
2025-06-30 09:37:03 +00:00
parent d6045efc66
commit 4be05c6524
1 changed files with 2 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
arch/arm64/kvm/vgic/vgic-init.c
+2 -9
View File
@@ -521,7 +521,7 @@ int kvm_vgic_map_resources(struct kvm *kvm)
gpa_t dist_base; gpa_t dist_base;
int ret = 0; int ret = 0;
if (likely(dist->ready)) if (likely(smp_load_acquire(&dist->ready)))
return 0; return 0;
mutex_lock(&kvm->slots_lock); mutex_lock(&kvm->slots_lock);
@@ -552,14 +552,7 @@ int kvm_vgic_map_resources(struct kvm *kvm)
goto out_slots; goto out_slots;
} }
/* smp_store_release(&dist->ready, true);
* kvm_io_bus_register_dev() guarantees all readers see the new MMIO
* registration before returning through synchronize_srcu(), which also
* implies a full memory barrier. As such, marking the distributor as
* 'ready' here is guaranteed to be ordered after all vCPUs having seen
* a completely configured distributor.
*/
dist->ready = true;
goto out_slots; goto out_slots;
out: out:
mutex_unlock(&kvm->arch.config_lock); mutex_unlock(&kvm->arch.config_lock);