Re: [PATCH v4 08/11] exfat: add iomap buffered I/O support

[Yuezhang.Mo@xxxxxxxx]

> @@ -749,28 +750,56 @@ static ssize_t exfat_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
> 
>  static vm_fault_t exfat_page_mkwrite(struct vm_fault *vmf)
>  {
> -       int err;
>         struct inode *inode = file_inode(vmf->vma->vm_file);
>         struct exfat_inode_info *ei = EXFAT_I(inode);
> -       loff_t new_valid_size;
> +       vm_fault_t ret;
> +       loff_t new_valid_size, mmap_valid_size;
> 
>         if (!inode_trylock(inode))
>                 return VM_FAULT_RETRY;
> 
> -       new_valid_size = ((loff_t)vmf->pgoff + 1) << PAGE_SHIFT;
> -       new_valid_size = min(new_valid_size, i_size_read(inode));
> +       mmap_valid_size = ((loff_t)vmf->pgoff + 1) << PAGE_SHIFT;
> +       new_valid_size = min(mmap_valid_size, i_size_read(inode));
> 
>         if (ei->valid_size < new_valid_size) {
> -               err = exfat_extend_valid_size(inode, new_valid_size);
> -               if (err < 0) {
> -                       inode_unlock(inode);
> -                       return vmf_fs_error(err);
> +               if (ei->zeroed_size < mmap_valid_size) {
> +                       int err;
> +
> +                       /*
> +                        * Only zero the range that hasn't been zeroed yet for
> +                        * this mmap write path. zeroed_size tracks the largest
> +                        * page-aligned offset that has already been zeroed.
> +                        *
> +                        * This prevents unnecessarily zeroing out the entire
> +                        * tail page on every page fault when userspace writes
> +                        * data byte-by-byte through mmap (after a small
> +                        * fallocate). It fixes data corruption in the tail page
> +                        * while preserving the existing valid_size semantics.
> +                        */
> +                       err = iomap_zero_range(inode, ei->zeroed_size,
> +                                       mmap_valid_size - ei->zeroed_size, NULL,
> +                                       &exfat_iomap_ops, NULL, NULL);

For volumes with clusters smaller than 4K, mmap_valid_size may be greater
than exfat_on_disksize(), causing iomap_zero_range() to allocate new clusters
for the file. We need to ensure that mmap_valid_size is not greater than
exfat_on_disksize() to avoid this situation.

> +                       if (err < 0) {
> +                               inode_unlock(inode);
> +                               return vmf_fs_error(err);
> +                       }
> +                       ei->zeroed_size = mmap_valid_size;