As noticed by Xiao, since soft-dirty clear command modifies page
tables we have to flush tlbs and call mmu notifiers. While the
former is done by the clear_refs engine itself, the latter is to
be done.
One thing to note about this -- in order not to call per-page
invalidate notifier (_all_ address space is about to be changed),
the _invalidate_range_start and _end are used. But for those start
and end are not known exactly. To address this, the same trick as
in exit_mmap() is used -- start is 0 and end is (unsigned long)-1.
Signed-off-by: Pavel Emelyanov <[email protected]>
---
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 27453c0..dbf61f6 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -11,6 +11,7 @@
#include <linux/rmap.h>
#include <linux/swap.h>
#include <linux/swapops.h>
+#include <linux/mmu_notifier.h>
#include <asm/elf.h>
#include <asm/uaccess.h>
@@ -815,6 +816,8 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
.private = &cp,
};
down_read(&mm->mmap_sem);
+ if (type == CLEAR_REFS_SOFT_DIRTY)
+ mmu_notifier_invalidate_range_start(mm, 0, -1);
for (vma = mm->mmap; vma; vma = vma->vm_next) {
cp.vma = vma;
if (is_vm_hugetlb_page(vma))
@@ -835,6 +838,8 @@ static ssize_t clear_refs_write(struct file *file, const char __user *buf,
walk_page_range(vma->vm_start, vma->vm_end,
&clear_refs_walk);
}
+ if (type == CLEAR_REFS_SOFT_DIRTY)
+ mmu_notifier_invalidate_range_end(mm, 0, -1);
flush_tlb_mm(mm);
up_read(&mm->mmap_sem);
mmput(mm);