Hi,
just some remarks regarding the new comment.
Bodo
On 17.04.22 07:26, Xiaoguang Wang wrote:
... snip ...
static const struct vm_operations_struct tcmu_vm_ops = {
@@ -3205,12 +3228,19 @@ static void find_free_blocks(void)
udev->dbi_max = block;
}
+ /*
+ * Release the block pages.
+ * Also note that since tcmu_vma_fault() gets one extra page
+ * refcount, tcmu_blocks_release() won't free pages if pages
+ * are in page fault procedure, which means it's safe to
s/in page fault procedure/mapped/
+ * call tcmu_blocks_release() before unmap_mapping_range().
"... before unmap_mapping_range(), which drops the refcount of pages it
unmaps and thus releases those pages."
Please feel free to find an even better wording.
+ */
+ pages_freed = tcmu_blocks_release(udev, start, end - 1);
+
/* Here will truncate the data area from off */
off = udev->data_off + (loff_t)start * udev->data_blk_size;
unmap_mapping_range(udev->inode->i_mapping, off, 0, 1);
- /* Release the block pages */
- pages_freed = tcmu_blocks_release(udev, start, end - 1);
mutex_unlock(&udev->cmdr_lock);
total_pages_freed += pages_freed;
