From: David Hildenbrand <david@redhat.com>
To: kvm@vger.kernel.org
Cc: Paolo Bonzini <pbonzini@redhat.com>,
rkrcmar@redhat.com, Peter Xu <peterx@redhat.com>,
david@redhat.com
Subject: [PATCH v3 02/24] KVM: x86: new irqchip mode KVM_IRQCHIP_INIT_IN_PROGRESS
Date: Fri, 7 Apr 2017 10:50:19 +0200 [thread overview]
Message-ID: <20170407085041.5257-3-david@redhat.com> (raw)
In-Reply-To: <20170407085041.5257-1-david@redhat.com>
Let's add a new mode and set it while we create the irqchip via
KVM_CREATE_IRQCHIP and KVM_CAP_SPLIT_IRQCHIP.
This mode will be used later to test if adding routes
(in kvm_set_routing_entry()) is already allowed.
Signed-off-by: David Hildenbrand <david@redhat.com>
---
arch/x86/include/asm/kvm_host.h | 1 +
arch/x86/kvm/irq.h | 18 +++++++++++++-----
arch/x86/kvm/x86.c | 11 ++++++++++-
3 files changed, 24 insertions(+), 6 deletions(-)
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index 74ef58c..164e544 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -727,6 +727,7 @@ struct kvm_hv {
enum kvm_irqchip_mode {
KVM_IRQCHIP_NONE,
+ KVM_IRQCHIP_INIT_IN_PROGRESS, /* temporarily set during creation */
KVM_IRQCHIP_KERNEL, /* created with KVM_CREATE_IRQCHIP */
KVM_IRQCHIP_SPLIT, /* created with KVM_CAP_SPLIT_IRQCHIP */
};
diff --git a/arch/x86/kvm/irq.h b/arch/x86/kvm/irq.h
index 40d5b2c..59e05fe 100644
--- a/arch/x86/kvm/irq.h
+++ b/arch/x86/kvm/irq.h
@@ -93,21 +93,29 @@ static inline int pic_in_kernel(struct kvm *kvm)
static inline int irqchip_split(struct kvm *kvm)
{
- return kvm->arch.irqchip_mode == KVM_IRQCHIP_SPLIT;
+ int mode = kvm->arch.irqchip_mode;
+
+ /* Matches smp_wmb() when setting irqchip_mode */
+ smp_rmb();
+ return mode == KVM_IRQCHIP_SPLIT;
}
static inline int irqchip_kernel(struct kvm *kvm)
{
- return kvm->arch.irqchip_mode == KVM_IRQCHIP_KERNEL;
+ int mode = kvm->arch.irqchip_mode;
+
+ /* Matches smp_wmb() when setting irqchip_mode */
+ smp_rmb();
+ return mode == KVM_IRQCHIP_KERNEL;
}
static inline int irqchip_in_kernel(struct kvm *kvm)
{
- bool ret = kvm->arch.irqchip_mode != KVM_IRQCHIP_NONE;
+ int mode = kvm->arch.irqchip_mode;
- /* Matches with wmb after initializing kvm->irq_routing. */
+ /* Matches smp_wmb() when setting irqchip_mode */
smp_rmb();
- return ret;
+ return mode > KVM_IRQCHIP_INIT_IN_PROGRESS;
}
void kvm_pic_reset(struct kvm_kpic_state *s);
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index ccbd45e..8f776f0 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -3934,9 +3934,14 @@ static int kvm_vm_ioctl_enable_cap(struct kvm *kvm,
goto split_irqchip_unlock;
if (kvm->created_vcpus)
goto split_irqchip_unlock;
+ kvm->arch.irqchip_mode = KVM_IRQCHIP_INIT_IN_PROGRESS;
r = kvm_setup_empty_irq_routing(kvm);
- if (r)
+ if (r) {
+ kvm->arch.irqchip_mode = KVM_IRQCHIP_NONE;
+ /* Pairs with smp_rmb() when reading irqchip_mode */
+ smp_wmb();
goto split_irqchip_unlock;
+ }
/* Pairs with irqchip_in_kernel. */
smp_wmb();
kvm->arch.irqchip_mode = KVM_IRQCHIP_SPLIT;
@@ -4024,8 +4029,12 @@ long kvm_arch_vm_ioctl(struct file *filp,
goto create_irqchip_unlock;
}
+ kvm->arch.irqchip_mode = KVM_IRQCHIP_INIT_IN_PROGRESS;
r = kvm_setup_default_irq_routing(kvm);
if (r) {
+ kvm->arch.irqchip_mode = KVM_IRQCHIP_NONE;
+ /* Pairs with smp_rmb() when reading irqchip_mode */
+ smp_wmb();
mutex_lock(&kvm->slots_lock);
mutex_lock(&kvm->irq_lock);
kvm_ioapic_destroy(kvm);
--
2.9.3
next prev parent reply other threads:[~2017-04-07 8:50 UTC|newest]
Thread overview: 43+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-04-07 8:50 [PATCH v3 00/24] pic/ioapic/irqchip cleanups + minor fixes David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 01/24] KVM: x86: race between KVM_SET_GSI_ROUTING and KVM_CREATE_IRQCHIP David Hildenbrand
2017-04-24 22:39 ` Steve Rutherford
2017-04-25 12:34 ` David Hildenbrand
2017-04-25 14:58 ` Radim Krčmář
2017-04-25 19:03 ` [PATCH] KVM: x86: don't hold kvm->lock in KVM_SET_GSI_ROUTING David Hildenbrand
2017-04-25 19:38 ` Steve Rutherford
2017-04-25 19:59 ` Radim Krčmář
2017-04-26 8:40 ` David Hildenbrand
2017-04-26 14:27 ` Radim Krčmář
2017-04-26 15:21 ` David Hildenbrand
2017-04-26 20:47 ` Radim Krčmář
2017-04-26 12:33 ` kbuild test robot
2017-04-26 14:24 ` Radim Krčmář
2017-04-07 8:50 ` David Hildenbrand [this message]
2017-04-12 18:26 ` [PATCH v3 02/24] KVM: x86: new irqchip mode KVM_IRQCHIP_INIT_IN_PROGRESS Radim Krčmář
2017-04-12 19:55 ` David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 03/24] KVM: x86: check against irqchip_mode in kvm_set_routing_entry() David Hildenbrand
2017-04-12 18:36 ` Radim Krčmář
2017-04-12 19:56 ` David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 04/24] KVM: x86: check against irqchip_mode in pic_in_kernel() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 05/24] KVM: x86: check against irqchip_mode in ioapic_in_kernel() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 06/24] KVM: x86: get rid of pic_irqchip() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 07/24] KVM: x86: get rid of ioapic_irqchip() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 08/24] KVM: x86: use ioapic_in_kernel() to check for ioapic existence David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 09/24] KVM: x86: remove duplicate checks for ioapic David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 10/24] KVM: x86: convert kvm_(set|get)_ioapic() into void David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 11/24] KVM: x86: don't take kvm->irq_lock when creating IRQCHIP David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 12/24] KVM: x86: push usage of slots_lock down David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 13/24] KVM: x86: KVM_IRQCHIP_PIC_MASTER only has 8 pins David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 14/24] KVM: x86: remove all-vcpu request from kvm_ioapic_init() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 15/24] KVM: x86: directly call kvm_make_scan_ioapic_request() in ioapic.c David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 16/24] KVM: x86: rename kvm_vcpu_request_scan_ioapic() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 17/24] KVM: x86: drop goto label in kvm_set_routing_entry() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 18/24] KVM: x86: cleanup return handling in setup_routing_entry() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 19/24] KVM: x86: simplify pic_unlock() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 20/24] KVM: x86: make kvm_pic_reset() static David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 21/24] KVM: x86: drop picdev_in_range() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 22/24] KVM: x86: set data directly in picdev_read() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 23/24] KVM: x86: simplify pic_ioport_read() David Hildenbrand
2017-04-07 8:50 ` [PATCH v3 24/24] KVM: x86: use irqchip_kernel() to check for pic+ioapic David Hildenbrand
2017-04-12 18:58 ` [PATCH v3 00/24] pic/ioapic/irqchip cleanups + minor fixes Radim Krčmář
2017-04-12 19:59 ` David Hildenbrand
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=20170407085041.5257-3-david@redhat.com \
--to=david@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=peterx@redhat.com \
--cc=rkrcmar@redhat.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).