From: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
To: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Cc: Avi Kivity <avi@redhat.com>,
Marcelo Tosatti <mtosatti@redhat.com>,
LKML <linux-kernel@vger.kernel.org>, KVM <kvm@vger.kernel.org>
Subject: [PATCH v2 13/16] KVM: MMU: break sptes write-protect if gfn is writable
Date: Fri, 13 Apr 2012 18:15:55 +0800 [thread overview]
Message-ID: <4F87FCDB.3050008@linux.vnet.ibm.com> (raw)
In-Reply-To: <4F87FA69.5060106@linux.vnet.ibm.com>
Make all sptes to be writable if the gfn become write-free to reduce
the later page fault
The idea is from Avi
Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
---
arch/x86/kvm/mmu.c | 34 +++++++++++++++++++++++++++++++++-
1 files changed, 33 insertions(+), 1 deletions(-)
diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 578a1e2..efa5d59 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -2323,15 +2323,45 @@ static void kvm_unsync_pages(struct kvm_vcpu *vcpu, gfn_t gfn)
}
}
+/*
+ * If the gfn become write-free, we make all sptes which point to this
+ * gfn to be writable.
+ * Note: we should call mark_page_dirty for the gfn later.
+ */
+static void rmap_break_page_table_wp(struct kvm_memory_slot *slot, gfn_t gfn)
+{
+ struct spte_iterator iter;
+ u64 *sptep;
+ int i;
+
+ for (i = PT_PAGE_TABLE_LEVEL;
+ i < PT_PAGE_TABLE_LEVEL + KVM_NR_PAGE_SIZES; i++) {
+ unsigned long *rmap = __gfn_to_rmap(gfn, i, slot);
+
+ for_each_rmap_spte(rmap, &iter, sptep) {
+ u64 spte = *sptep;
+
+ if (!is_writable_pte(spte) &&
+ (spte & SPTE_ALLOW_WRITE)) {
+ spte &= ~SPTE_WRITE_PROTECT;
+ spte |= PT_WRITABLE_MASK;
+ mmu_spte_update(sptep, spte);
+ }
+ }
+ }
+}
+
static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
bool can_unsync)
{
+ struct kvm_memory_slot *slot;
struct kvm_mmu_page *s;
struct hlist_node *node;
unsigned long *rmap;
bool need_unsync = false;
- rmap = gfn_to_rmap(vcpu->kvm, gfn, PT_PAGE_TABLE_LEVEL);
+ slot = gfn_to_memslot(vcpu->kvm, gfn);
+ rmap = __gfn_to_rmap(gfn, PT_PAGE_TABLE_LEVEL, slot);
if (!vcpu->kvm->arch.indirect_shadow_pages)
goto write_free;
@@ -2353,6 +2383,8 @@ static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
if (need_unsync)
kvm_unsync_pages(vcpu, gfn);
+ rmap_break_page_table_wp(slot, gfn);
+
write_free:
__clear_bit(PTE_LIST_WP_BIT, rmap);
return 0;
--
1.7.7.6
next prev parent reply other threads:[~2012-04-13 10:16 UTC|newest]
Thread overview: 49+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-04-13 10:05 [PATCH v2 00/16] KVM: MMU: fast page fault Xiao Guangrong
2012-04-13 10:09 ` [PATCH v2 01/16] KVM: MMU: cleanup __direct_map Xiao Guangrong
2012-04-13 10:10 ` [PATCH v2 02/16] KVM: MMU: introduce mmu_spte_establish Xiao Guangrong
2012-04-13 10:10 ` [PATCH v2 03/16] KVM: MMU: properly assert spte on rmap walking path Xiao Guangrong
2012-04-14 2:15 ` Takuya Yoshikawa
2012-04-16 3:26 ` Xiao Guangrong
2012-04-13 10:11 ` [PATCH v2 04/16] KVM: MMU: return bool in __rmap_write_protect Xiao Guangrong
2012-04-14 2:00 ` Takuya Yoshikawa
2012-04-15 11:25 ` Avi Kivity
2012-04-16 14:14 ` Takuya Yoshikawa
2012-04-16 14:28 ` Avi Kivity
2012-04-16 15:54 ` Takuya Yoshikawa
2012-04-13 10:11 ` [PATCH v2 05/16] KVM: MMU: abstract spte write-protect Xiao Guangrong
2012-04-14 2:26 ` Takuya Yoshikawa
2012-04-16 3:27 ` Xiao Guangrong
2012-04-13 10:12 ` [PATCH v2 06/16] KVM: VMX: export PFEC.P bit on ept Xiao Guangrong
2012-04-13 10:12 ` [PATCH v2 07/16] KVM: MMU: introduce for_each_pte_list_spte Xiao Guangrong
2012-04-14 2:44 ` Takuya Yoshikawa
2012-04-16 3:36 ` Xiao Guangrong
2012-04-17 14:47 ` Takuya Yoshikawa
2012-04-18 4:01 ` Xiao Guangrong
2012-04-21 1:01 ` Takuya Yoshikawa
2012-04-21 4:36 ` Xiao Guangrong
2012-04-18 10:03 ` Xiao Guangrong
2012-04-21 1:03 ` Takuya Yoshikawa
2012-04-13 10:13 ` [PATCH v2 08/16] KVM: MMU: store more bits in rmap Xiao Guangrong
2012-04-13 10:13 ` [PATCH v2 09/16] KVM: MMU: fast mmu_need_write_protect path for hard mmu Xiao Guangrong
2012-04-13 10:14 ` [PATCH v2 10/16] KVM: MMU: fask check whether page is writable Xiao Guangrong
2012-04-14 3:01 ` Takuya Yoshikawa
2012-04-16 3:38 ` Xiao Guangrong
2012-04-15 15:16 ` Avi Kivity
2012-04-16 3:25 ` Xiao Guangrong
2012-04-16 10:02 ` Avi Kivity
2012-04-16 10:20 ` Xiao Guangrong
2012-04-16 11:47 ` Avi Kivity
2012-04-17 3:55 ` Xiao Guangrong
2012-04-17 7:41 ` Avi Kivity
2012-04-17 12:10 ` Xiao Guangrong
2012-04-13 10:14 ` [PATCH v2 11/16] KVM: MMU: introduce SPTE_ALLOW_WRITE bit Xiao Guangrong
2012-04-13 10:15 ` [PATCH v2 12/16] KVM: MMU: introduce SPTE_WRITE_PROTECT bit Xiao Guangrong
2012-04-13 10:15 ` Xiao Guangrong [this message]
2012-04-13 10:16 ` [PATCH v2 14/16] KVM: MMU: fast path of handling guest page fault Xiao Guangrong
2012-04-18 1:47 ` Marcelo Tosatti
2012-04-18 3:53 ` Xiao Guangrong
2012-04-18 23:08 ` Marcelo Tosatti
2012-04-13 10:17 ` [PATCH v2 15/16] KVM: MMU: trace fast " Xiao Guangrong
2012-04-13 10:17 ` [PATCH v2 16/16] KVM: MMU: fix kvm_mmu_pagetable_walk tracepoint Xiao Guangrong
2012-04-14 3:37 ` [PATCH v2 00/16] KVM: MMU: fast page fault Takuya Yoshikawa
2012-04-16 3:50 ` Xiao Guangrong
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=4F87FCDB.3050008@linux.vnet.ibm.com \
--to=xiaoguangrong@linux.vnet.ibm.com \
--cc=avi@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mtosatti@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