From: Gleb Natapov <gleb@redhat.com>
To: Marcelo Tosatti <mtosatti@redhat.com>
Cc: kvm <kvm@vger.kernel.org>
Subject: Re: KVM: attempt async pf only if address is contained in vma
Date: Tue, 19 Oct 2010 13:05:22 +0200 [thread overview]
Message-ID: <20101019110522.GH10207@redhat.com> (raw)
In-Reply-To: <20101018222402.GA27350@amt.cnet>
On Mon, Oct 18, 2010 at 08:24:03PM -0200, Marcelo Tosatti wrote:
>
> Only attempt async pagefault if address is contained within vma.
>
> Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
>
> diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
> index 5d57ec9..a9cfbd2 100644
> --- a/virt/kvm/kvm_main.c
> +++ b/virt/kvm/kvm_main.c
> @@ -996,8 +996,9 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic,
>
> if (vma == NULL || addr < vma->vm_start ||
> !(vma->vm_flags & VM_PFNMAP)) {
> - if (async && !(vma->vm_flags & VM_PFNMAP) &&
> - (vma->vm_flags & VM_WRITE))
> + if (async && vma && !(vma->vm_flags & VM_PFNMAP) &&
actually we should check for addr < vma->vm_start here too. find_vma()
is very strange function.
What about patch below (only compile tested for now)?
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 796b06e..29e40e5 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -954,6 +954,12 @@ unsigned long gfn_to_hva(struct kvm *kvm, gfn_t gfn)
}
EXPORT_SYMBOL_GPL(gfn_to_hva);
+static pfn_t get_fault_pfn(void)
+{
+ get_page(fault_page);
+ return fault_pfn;
+}
+
static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic,
bool *async)
{
@@ -983,7 +989,7 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic,
struct vm_area_struct *vma;
if (atomic)
- goto return_fault_page;
+ return get_fault_pfn();
down_read(¤t->mm->mmap_sem);
if (is_hwpoison_address(addr)) {
@@ -992,22 +998,20 @@ static pfn_t hva_to_pfn(struct kvm *kvm, unsigned long addr, bool atomic,
return page_to_pfn(hwpoison_page);
}
- vma = find_vma(current->mm, addr);
+ vma = find_vma_intersection(current->mm, addr, addr+1);
- if (vma == NULL || addr < vma->vm_start ||
- !(vma->vm_flags & VM_PFNMAP)) {
- if (async && !(vma->vm_flags & VM_PFNMAP) &&
- (vma->vm_flags & VM_WRITE))
+ if (vma == NULL)
+ return get_fault_pfn();
+ if ((vma->vm_flags & VM_PFNMAP)) {
+ pfn = ((addr - vma->vm_start) >> PAGE_SHIFT) +
+ vma->vm_pgoff;
+ BUG_ON(!kvm_is_mmio_pfn(pfn));
+ } else {
+ if (async && (vma->vm_flags & VM_WRITE))
*async = true;
- up_read(¤t->mm->mmap_sem);
-return_fault_page:
- get_page(fault_page);
- return page_to_pfn(fault_page);
+ pfn = get_fault_pfn();
}
-
- pfn = ((addr - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
up_read(¤t->mm->mmap_sem);
- BUG_ON(!kvm_is_mmio_pfn(pfn));
} else
pfn = page_to_pfn(page[0]);
--
Gleb.
next prev parent reply other threads:[~2010-10-19 11:05 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-10-18 22:24 KVM: attempt async pf only if address is contained in vma Marcelo Tosatti
2010-10-19 11:05 ` Gleb Natapov [this message]
2010-10-19 12:17 ` Gleb Natapov
2010-10-19 13:14 ` Marcelo Tosatti
2010-10-19 13:15 ` Gleb Natapov
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=20101019110522.GH10207@redhat.com \
--to=gleb@redhat.com \
--cc=kvm@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 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.