public inbox for kvm@vger.kernel.org
 help / color / mirror / Atom feed
From: Marcelo Tosatti <mtosatti@redhat.com>
To: kvm@vger.kernel.org
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Subject: [patch 2/2] qemu: kvm: mark guest mapping as MADV_DONTFORK
Date: Fri, 22 Aug 2008 20:03:35 -0300	[thread overview]
Message-ID: <20080822230510.199157861@localhost.localdomain> (raw)
In-Reply-To: 20080822230333.323535558@localhost.localdomain

[-- Attachment #1: kvm-setup-mem --]
[-- Type: text/plain, Size: 2720 bytes --]

When qemu fork's (ssh migration, qemu-nbd, slirp), the guest memory
mapping becomes shared and write-protected by parent and child, until
execve switches to a new mm.

Since kernel commit 7885a4fbebcbae1f8594445a46aefb9d97353594,
get_user_pages with force=1 parameter will break COW during this window,
leaving stale shadow mappings that point to the previously shared page.

Fix this by madvising the range as MADV_DONTFORK, if mmu notifiers are
disabled.

Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>

Index: kvm-userspace.tip/qemu/qemu-kvm.c
===================================================================
--- kvm-userspace.tip.orig/qemu/qemu-kvm.c
+++ kvm-userspace.tip/qemu/qemu-kvm.c
@@ -25,6 +25,7 @@ int kvm_pit = 1;
 #include <pthread.h>
 #include <sys/utsname.h>
 #include <sys/syscall.h>
+#include <sys/mman.h>
 
 #define bool _Bool
 #define false 0
@@ -812,6 +813,19 @@ void kvm_cpu_register_physical_memory(ta
     }
 }
 
+int kvm_setup_guest_memory(void *area, unsigned long size)
+{
+    int ret = 0;
+
+    if (kvm_enabled() && !kvm_has_sync_mmu(kvm_context))
+        ret = madvise(area, size, MADV_DONTFORK);
+
+    if (ret)
+        perror ("madvise");
+
+    return ret;
+}
+
 int kvm_qemu_check_extension(int ext)
 {
     return kvm_check_extension(kvm_context, ext);
Index: kvm-userspace.tip/qemu/qemu-kvm.h
===================================================================
--- kvm-userspace.tip.orig/qemu/qemu-kvm.h
+++ kvm-userspace.tip/qemu/qemu-kvm.h
@@ -45,6 +45,7 @@ void *kvm_cpu_create_phys_mem(target_phy
 
 void kvm_cpu_destroy_phys_mem(target_phys_addr_t start_addr,
 			      unsigned long size);
+int kvm_setup_guest_memory(void *area, unsigned long size);
 
 int kvm_arch_qemu_create_context(void);
 
Index: kvm-userspace.tip/qemu/vl.c
===================================================================
--- kvm-userspace.tip.orig/qemu/vl.c
+++ kvm-userspace.tip/qemu/vl.c
@@ -8890,7 +8890,7 @@ static int gethugepagesize(void)
     return hugepagesize;
 }
 
-void *alloc_mem_area(unsigned long memory, const char *path)
+void *alloc_mem_area(size_t memory, unsigned long *len, const char *path)
 {
     char *filename;
     void *area;
@@ -8929,18 +8929,23 @@ void *alloc_mem_area(unsigned long memor
 	return NULL;
     }
 
+    *len = memory;
     return area;
 }
 
 void *qemu_alloc_physram(unsigned long memory)
 {
     void *area = NULL;
+    unsigned long map_len = memory;
 
     if (mem_path)
-	area = alloc_mem_area(memory, mem_path);
+	area = alloc_mem_area(memory, &map_len, mem_path);
     if (!area)
 	area = qemu_vmalloc(memory);
-
+#ifdef USE_KVM
+    if (kvm_setup_guest_memory(area, map_len))
+        area = NULL;
+#endif
     return area;
 }
 

-- 


  parent reply	other threads:[~2008-08-22 23:07 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-08-22 23:03 [patch 0/2] madvise(MADV_DONTFORK) guest mem if !mmu_notifiers Marcelo Tosatti
2008-08-22 23:03 ` [patch 1/2] libkvm: add kvm_has_sync_mmu Marcelo Tosatti
2008-08-22 23:03 ` Marcelo Tosatti [this message]
2008-08-25 10:20 ` [patch 0/2] madvise(MADV_DONTFORK) guest mem if !mmu_notifiers Avi Kivity

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=20080822230510.199157861@localhost.localdomain \
    --to=mtosatti@redhat.com \
    --cc=kvm@vger.kernel.org \
    /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