Skip to contentSkip to conversation

kvm_vm_ioctl_get_dirty_log doesn't need any, probably doesn't want any

PATCH In review Rusty Russell 2 participants 6 messages Jul 30, 2007 – Jul 31, 2007 5 messages not in the archive KVM

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 kvm_vm_ioctl_get_dirty_log doesn't need any, probably doesn't want any

I'm not sure what "any" was for: some optimization which got trimmed, I imagine. But neither "any" nor "i" are used later in the function, so I've excised it.

drivers/kvm/kvm_main.c +2 -8
diff -r a17d7fe2b240 drivers/kvm/kvm_main.c
@@ -830,9 +830,7 @@ static int kvm_vm_ioctl_get_dirty_log(st
 				      struct kvm_dirty_log *log)
 {
 	struct kvm_memory_slot *memslot;
-	int r, i;
-	int n;
-	unsigned long any = 0;
+	int r, n;
 
 	spin_lock(&kvm->lock);
 
@@ -851,11 +849,7 @@ static int kvm_vm_ioctl_get_dirty_log(st
 	if (!memslot->dirty_bitmap)
 		goto out;
 
-	n = ALIGN(memslot->npages, BITS_PER_LONG) / 8;
-
-	for (i = 0; !any && i < n/sizeof(long); ++i)
-		any = memslot->dirty_bitmap[i];
-
+	n = BITS_TO_LONGS(memslot->npages) * sizeof(long);
 	r = -EFAULT;
 	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
 		goto out;

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/

A message that is not in the archive (sent off-list or not yet received): 1185798696.6131.44.camel-bi+AKbBUZKY6gyzm1THtWbp2dZbC/[email protected]

#2 lore raw
Quoted 5 lines · Rusty Russell

I'm not sure what "any" was for: some optimization which got trimmed, I imagine. But neither "any" nor "i" are used later in the function, so I've excised it.

Looks like it was used to guard kvm_mmu_slot_remove_write_access(); optimizing the case where the guest just leaves the screen alone (which it usually does, especially in benchmarks).

I'd rather reinstate that optimization. See 66d8a4e4d4bd470216028daabb9d887b73259c96 where the damage was done.

Quoted 32 lines

diff -r a17d7fe2b240 drivers/kvm/kvm_main.c

drivers/kvm/kvm_main.c +2 -8
@@ -830,9 +830,7 @@ static int kvm_vm_ioctl_get_dirty_log(st
 				      struct kvm_dirty_log *log)
 {
 	struct kvm_memory_slot *memslot;
-	int r, i;
-	int n;
-	unsigned long any = 0;
+	int r, n;
 
 	spin_lock(&kvm->lock);
 
@@ -851,11 +849,7 @@ static int kvm_vm_ioctl_get_dirty_log(st
 	if (!memslot->dirty_bitmap)
 		goto out;
 
-	n = ALIGN(memslot->npages, BITS_PER_LONG) / 8;
-
-	for (i = 0; !any && i < n/sizeof(long); ++i)
-		any = memslot->dirty_bitmap[i];
-
+	n = BITS_TO_LONGS(memslot->npages) * sizeof(long);
 	r = -EFAULT;
 	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
 		goto out;
Signature · 8 lines
error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/

A message that is not in the archive (sent off-list or not yet received): 46ADDC16.2090801-atKUWr5tajBWk0Htik3J/[email protected]

#4 lore raw

PATCH 1/1 kvm_vm_ioctl_get_dirty_log restore "nothing dirty" optimization

kvm_vm_ioctl_get_dirty_log scans bitmap to see it it's all zero, but doesn't use that information.

Avi says:
Looks like it was used to guard kvm_mmu_slot_remove_write_access();
optimizing the case where the guest just leaves the screen alone (which it usually does, especially in benchmarks).

I'd rather reinstate that optimization. See 66d8a4e4d4bd470216028daabb9d887b73259c96 where the damage was done.

It's pretty simple: if the bitmap is all zero, we don't need to do anything to clean it.

drivers/kvm/kvm_main.c +7 -5
diff -r 66566cf6b576 drivers/kvm/kvm_main.c
@@ -838,11 +838,13 @@ static int kvm_vm_ioctl_get_dirty_log(st
 	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
 		goto out;
 
-	spin_lock(&kvm->lock);
-	kvm_mmu_slot_remove_write_access(kvm, log->slot);
-	kvm_flush_remote_tlbs(kvm);
-	memset(memslot->dirty_bitmap, 0, n);
-	spin_unlock(&kvm->lock);
+	if (any) {
+		spin_lock(&kvm->lock);
+		kvm_mmu_slot_remove_write_access(kvm, log->slot);
+		kvm_flush_remote_tlbs(kvm);
+		memset(memslot->dirty_bitmap, 0, n);
+		spin_unlock(&kvm->lock);
+	}
 
 	r = 0;
 

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/

A message that is not in the archive (sent off-list or not yet received): 1185801694.6131.46.camel-bi+AKbBUZKY6gyzm1THtWbp2dZbC/[email protected]

#6 lore raw

Re: [PATCH] kvm_vm_ioctl_get_dirty_log restore "nothing dirty" optimization

Quoted 26 lines · Rusty Russell

kvm_vm_ioctl_get_dirty_log scans bitmap to see it it's all zero, but doesn't use that information.

Avi says:
Looks like it was used to guard kvm_mmu_slot_remove_write_access();
optimizing the case where the guest just leaves the screen alone (which it usually does, especially in benchmarks).

I'd rather reinstate that optimization. See 66d8a4e4d4bd470216028daabb9d887b73259c96 where the damage was done.

It's pretty simple: if the bitmap is all zero, we don't need to do anything to clean it.

diff -r 66566cf6b576 drivers/kvm/kvm_main.c

drivers/kvm/kvm_main.c +0 -3
@@ -838,11 +838,13 @@ static int kvm_vm_ioctl_get_dirty_log(st
 	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
 		goto out;
 
-	spin_lock(&kvm->lock);
-	kvm_mmu_slot_remove_write_access(kvm, log->slot);
-	kvm_flush_remote_tlbs(kvm);

drivers/kvm/kvm_main.c static int kvm_vm_ioctl_get_dirty_log(st · line 848

Quoted 10 lines
drivers/kvm/kvm_main.c +7 -2
-	memset(memslot->dirty_bitmap, 0, n);
-	spin_unlock(&kvm->lock);
+	if (any) {
+		spin_lock(&kvm->lock);
+		kvm_mmu_slot_remove_write_access(kvm, log->slot);
+		kvm_flush_remote_tlbs(kvm);
+		memset(memslot->dirty_bitmap, 0, n);
+		spin_unlock(&kvm->lock);
+	}
 

spin_lock()? my, that's so last week. kvm->lock is a mutex.

Signature · 8 lines
error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/

A message that is not in the archive (sent off-list or not yet received): 46ADE78A.7080406-atKUWr5tajBWk0Htik3J/[email protected]

#8 lore raw

PATCH 1/1 kvm_vm_ioctl_get_dirty_log restore "nothing dirty" optimization

kvm_vm_ioctl_get_dirty_log scans bitmap to see it it's all zero, but doesn't use that information.

Avi says:
Looks like it was used to guard kvm_mmu_slot_remove_write_access();
optimizing the case where the guest just leaves the screen alone (which it usually does, especially in benchmarks).

I'd rather reinstate that optimization. See 66d8a4e4d4bd470216028daabb9d887b73259c96 where the damage was done.

It's pretty simple: if the bitmap is all zero, we don't need to do anything to clean it.

drivers/kvm/kvm_main.c +8 -5
diff -r 37f0dd1eaa0d drivers/kvm/kvm_main.c
@@ -803,11 +803,14 @@ static int kvm_vm_ioctl_get_dirty_log(st
 	if (copy_to_user(log->dirty_bitmap, memslot->dirty_bitmap, n))
 		goto out;
 
-	mutex_lock(&kvm->lock);
-	kvm_mmu_slot_remove_write_access(kvm, log->slot);
-	kvm_flush_remote_tlbs(kvm);
-	memset(memslot->dirty_bitmap, 0, n);
-	mutex_unlock(&kvm->lock);
+	/* If nothing is dirty, don't bother messing with page tables. */
+	if (any) {
+		mutex_lock(&kvm->lock);
+		kvm_mmu_slot_remove_write_access(kvm, log->slot);
+		kvm_flush_remote_tlbs(kvm);
+		memset(memslot->dirty_bitmap, 0, n);
+		mutex_unlock(&kvm->lock);
+	}
 
 	r = 0;
 

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/

A message that is not in the archive (sent off-list or not yet received): 1185875867.6131.76.camel-bi+AKbBUZKY6gyzm1THtWbp2dZbC/[email protected]

Applied
#10 lore raw

Re: [PATCH] kvm_vm_ioctl_get_dirty_log restore "nothing dirty" optimization

Quoted 8 lines · Rusty Russell

kvm_vm_ioctl_get_dirty_log scans bitmap to see it it's all zero, but doesn't use that information.

Avi says:
Looks like it was used to guard kvm_mmu_slot_remove_write_access();
optimizing the case where the guest just leaves the screen alone (which it usually does, especially in benchmarks).

Quoted 7 lines

I'd rather reinstate that optimization. See 66d8a4e4d4bd470216028daabb9d887b73259c96 where the damage was done.

It's pretty simple: if the bitmap is all zero, we don't need to do anything to clean it.

Applied; thanks.

Signature · 8 lines
error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/