memory: Fix start offset for bitmap log_clear hook

[PATCH] memory: Fix start offset for bitmap log_clear hook

Posted by Matt Borgerson 5 years, 8 months ago

Currently only the final page offset is being passed to the `log_clear`
hook via `memory_region_clear_dirty_bitmap` after it is used as an
iterator in `cpu_physical_memory_test_and_clear_dirty`. This patch
corrects the start address and size of the region.

Signed-off-by: Matt Borgerson <contact@mborgerson.com>
---
 exec.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/exec.c b/exec.c
index 8e9cc3b47c..f0f7d0be49 100644
--- a/exec.c
+++ b/exec.c
@@ -1314,7 +1314,7 @@ bool cpu_physical_memory_test_and_clear_dirty(ram_addr_t start,
                                               unsigned client)
 {
     DirtyMemoryBlocks *blocks;
-    unsigned long end, page;
+    unsigned long end, page, start_page;
     bool dirty = false;
     RAMBlock *ramblock;
     uint64_t mr_offset, mr_size;
@@ -1324,7 +1324,8 @@ bool cpu_physical_memory_test_and_clear_dirty(ram_addr_t start,
     }
 
     end = TARGET_PAGE_ALIGN(start + length) >> TARGET_PAGE_BITS;
-    page = start >> TARGET_PAGE_BITS;
+    start_page = start >> TARGET_PAGE_BITS;
+    page = start_page;
 
     WITH_RCU_READ_LOCK_GUARD() {
         blocks = atomic_rcu_read(&ram_list.dirty_memory[client]);
@@ -1344,8 +1345,8 @@ bool cpu_physical_memory_test_and_clear_dirty(ram_addr_t start,
             page += num;
         }
 
-        mr_offset = (ram_addr_t)(page << TARGET_PAGE_BITS) - ramblock->offset;
-        mr_size = (end - page) << TARGET_PAGE_BITS;
+        mr_offset = (ram_addr_t)(start_page << TARGET_PAGE_BITS) - ramblock->offset;
+        mr_size = (end - start_page) << TARGET_PAGE_BITS;
         memory_region_clear_dirty_bitmap(ramblock->mr, mr_offset, mr_size);
     }
 
-- 
2.17.1

Re: [PATCH] memory: Fix start offset for bitmap log_clear hook

Posted by Peter Xu 5 years, 8 months ago

On Tue, Feb 18, 2020 at 03:19:10AM -0700, Matt Borgerson wrote:
> Currently only the final page offset is being passed to the `log_clear`
> hook via `memory_region_clear_dirty_bitmap` after it is used as an
> iterator in `cpu_physical_memory_test_and_clear_dirty`. This patch
> corrects the start address and size of the region.
> 
> Signed-off-by: Matt Borgerson <contact@mborgerson.com>

Looks correct, thanks!

Reviewed-by: Peter Xu <peterx@redhat.com>

-- 
Peter Xu

Re: [PATCH] memory: Fix start offset for bitmap log_clear hook

Posted by Matt Borgerson 5 years, 8 months ago

[ ping ]

Hi Paolo, would you mind taking a quick look at this patch for
memory.c to consider
it for merge? This resolves an issue with dirty bits not being cleared
as expected.

Here's the Patchwork link: http://patchwork.ozlabs.org/patch/1240121/

Thanks for your time!

Matt

On Tue, Feb 18, 2020 at 9:53 AM Peter Xu <peterx@redhat.com> wrote:
>
> On Tue, Feb 18, 2020 at 03:19:10AM -0700, Matt Borgerson wrote:
> > Currently only the final page offset is being passed to the `log_clear`
> > hook via `memory_region_clear_dirty_bitmap` after it is used as an
> > iterator in `cpu_physical_memory_test_and_clear_dirty`. This patch
> > corrects the start address and size of the region.
> >
> > Signed-off-by: Matt Borgerson <contact@mborgerson.com>
>
> Looks correct, thanks!
>
> Reviewed-by: Peter Xu <peterx@redhat.com>
>
> --
> Peter Xu
>

Re: [PATCH] memory: Fix start offset for bitmap log_clear hook

Posted by Paolo Bonzini 5 years, 8 months ago

On 25/02/20 21:46, Matt Borgerson wrote:
> [ ping ]
> 
> Hi Paolo, would you mind taking a quick look at this patch for
> memory.c to consider
> it for merge? This resolves an issue with dirty bits not being cleared
> as expected.
> 
> Here's the Patchwork link: http://patchwork.ozlabs.org/patch/1240121/
> 
> Thanks for your time!

Yes, I queued it now.

Thanks!

Paolo