From: Marcelo Tosatti <mtosatti@redhat.com>
To: kvm <kvm@vger.kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>,
Mike Galbraith <bitbucket@online.de>
Subject: KVM: x86: use dynamic percpu allocations for shared msrs area
Date: Thu, 3 Jan 2013 11:41:39 -0200 [thread overview]
Message-ID: <20130103134139.GA1787@amt.cnet> (raw)
Andy, Mike, can you confirm whether this fixes the percpu allocation
failures when loading kvm.ko? TIA
----
Use dynamic percpu allocations for the shared msrs structure,
to avoid using the limited reserved percpu space.
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 1c9c834..5229a67 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -120,7 +120,7 @@ struct kvm_shared_msrs {
};
static struct kvm_shared_msrs_global __read_mostly shared_msrs_global;
-static DEFINE_PER_CPU(struct kvm_shared_msrs, shared_msrs);
+static struct kvm_shared_msrs __percpu *shared_msrs;
struct kvm_stats_debugfs_item debugfs_entries[] = {
{ "pf_fixed", VCPU_STAT(pf_fixed) },
@@ -191,10 +191,10 @@ static void kvm_on_user_return(struct user_return_notifier *urn)
static void shared_msr_update(unsigned slot, u32 msr)
{
- struct kvm_shared_msrs *smsr;
u64 value;
+ unsigned int cpu = smp_processor_id();
+ struct kvm_shared_msrs *smsr = per_cpu_ptr(shared_msrs, cpu);
- smsr = &__get_cpu_var(shared_msrs);
/* only read, and nobody should modify it at this time,
* so don't need lock */
if (slot >= shared_msrs_global.nr) {
@@ -226,7 +226,8 @@ static void kvm_shared_msr_cpu_online(void)
void kvm_set_shared_msr(unsigned slot, u64 value, u64 mask)
{
- struct kvm_shared_msrs *smsr = &__get_cpu_var(shared_msrs);
+ unsigned int cpu = smp_processor_id();
+ struct kvm_shared_msrs *smsr = per_cpu_ptr(shared_msrs, cpu);
if (((value ^ smsr->values[slot].curr) & mask) == 0)
return;
@@ -242,7 +243,8 @@ EXPORT_SYMBOL_GPL(kvm_set_shared_msr);
static void drop_user_return_notifiers(void *ignore)
{
- struct kvm_shared_msrs *smsr = &__get_cpu_var(shared_msrs);
+ unsigned int cpu = smp_processor_id();
+ struct kvm_shared_msrs *smsr = per_cpu_ptr(shared_msrs, cpu);
if (smsr->registered)
kvm_on_user_return(&smsr->urn);
@@ -5233,9 +5235,16 @@ int kvm_arch_init(void *opaque)
goto out;
}
+ r = -ENOMEM;
+ shared_msrs = alloc_percpu(struct kvm_shared_msrs);
+ if (!shared_msrs) {
+ printk(KERN_ERR "kvm: failed to allocate percpu kvm_shared_msrs\n");
+ goto out;
+ }
+
r = kvm_mmu_module_init();
if (r)
- goto out;
+ goto out_free_percpu;
kvm_set_mmio_spte_mask();
kvm_init_msr_list();
@@ -5258,6 +5267,8 @@ int kvm_arch_init(void *opaque)
return 0;
+out_free_percpu:
+ free_percpu(shared_msrs);
out:
return r;
}
@@ -5275,6 +5286,7 @@ void kvm_arch_exit(void)
#endif
kvm_x86_ops = NULL;
kvm_mmu_module_exit();
+ free_percpu(shared_msrs);
}
int kvm_emulate_halt(struct kvm_vcpu *vcpu)
next reply other threads:[~2013-01-03 15:14 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-01-03 13:41 Marcelo Tosatti [this message]
2013-01-04 5:19 ` KVM: x86: use dynamic percpu allocations for shared msrs area Mike Galbraith
2013-01-08 9:16 ` Gleb Natapov
2013-02-01 22:48 ` Andy Lutomirski
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20130103134139.GA1787@amt.cnet \
--to=mtosatti@redhat.com \
--cc=bitbucket@online.de \
--cc=kvm@vger.kernel.org \
--cc=luto@amacapital.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.