mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE
@ 2026-09-05 12:35 朱 天浩
  2026-09-05 14:08 ` liubaolin
  2026-09-07  8:54 ` Hyunchul Lee
  0 siblings, 2 replies; 6+ messages in thread
From: 朱 天浩 @ 2026-09-05 12:35 UTC (permalink / raw)
  To: linkinjeon, hyc.lee; +Cc: ntfs, linux-kernel

When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
which is indeed what mkfs arranges.  However, the folio rewrite kept a
leftover from the old buffer-head/runlist based implementation:
vol->cluster_size_mask is still applied to the mirror write offset.
In fact it is no longer needed -- and in some configurations it is
actively wrong.

For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
written to the location of record 0.  $MFTMirr therefore gets out of
sync with $MFT and the following warning is printed on remount:

	ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
	      record 0 do not match.  Run chkdsk.

Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
to the base LCN address instead of applying the cluster_size_mask
truncation.  This is the standard file-offset calculation and is
correct for every combination of cluster size, page size and MFT
record size, provided $MFTMirr data is contiguous from mftmirr_lcn
(which mkfs always arranges).

While here, verify that the folio returned by read_mapping_folio() is
large enough to hold the whole mirror record (folio_ofs +
mft_record_size <= folio_size(folio)), guarding against records that
straddle a folio boundary when mft_record_size exceeds PAGE_SIZE,
although this cannot happen in practice.

Tested on the volume above: before the change mirror records 1-3 are
all written to record 0's sector; after the change the write-back
probe reports:

	ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
	ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628

which matches the expected sectors computed as:

	(NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT

	record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
	record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640

A 13 MB file write succeeds on this volume and the file is byte-for-byte
identical after a umount/mount cycle.

Fixes: 115380f9a2f9 ("ntfs: update mft operations")
Assisted-by: UOS-AI
Assisted-by: CodeBuddy:Hy4 preview
Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
---
 fs/ntfs/mft.c | 26 ++++++++++++++++----------
 1 file changed, 16 insertions(+), 10 deletions(-)

diff --git a/fs/ntfs/mft.c b/fs/ntfs/mft.c
index 7e58c99f1728..a62b1d41697c 100644
--- a/fs/ntfs/mft.c
+++ b/fs/ntfs/mft.c
@@ -462,7 +462,7 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
 {
 	u8 *kmirr;
 	struct folio *folio;
-	unsigned int folio_ofs, lcn_folio_off = 0;
+	unsigned int folio_ofs;
 	int err = 0;
 	struct bio *bio;
 
@@ -482,25 +482,31 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
 		goto err_out;
 	}
 
+	/* Offset of the mft mirror record inside the folio. */
+	folio_ofs = NTFS_MFT_NR_TO_POFS(vol, mft_no);
+
+	/* The mirror record must fit entirely within this folio. */
+	if (folio_ofs + vol->mft_record_size > folio_size(folio)) {
+		ntfs_error(vol->sb, "Mft mirror record 0x%llx does not fit in folio.",
+			   mft_no);
+		folio_put(folio);
+		err = -EIO;
+		goto err_out;
+	}
+
 	folio_lock(folio);
 	folio_clear_uptodate(folio);
-	/* Offset of the mft mirror record inside the page. */
-	folio_ofs = NTFS_MFT_NR_TO_POFS(vol, mft_no);
-	/* The address in the page of the mirror copy of the mft record @m. */
+	/* The address in the folio of the mirror copy of the mft record @m. */
 	kmirr = kmap_local_folio(folio, 0) + folio_ofs;
 	/* Copy the mst protected mft record to the mirror. */
 	memcpy(kmirr, m, vol->mft_record_size);
 	kunmap_local(kmirr);
 
-	if (vol->cluster_size_bits > PAGE_SHIFT) {
-		lcn_folio_off = folio->index << PAGE_SHIFT;
-		lcn_folio_off &= vol->cluster_size_mask;
-	}
-
 	bio = bio_alloc(vol->sb->s_bdev, 1, REQ_OP_WRITE, GFP_NOIO);
 	bio->bi_iter.bi_sector =
 		ntfs_bytes_to_bio_sector(NTFS_CLU_TO_B(vol, vol->mftmirr_lcn) +
-					 lcn_folio_off + folio_ofs);
+					 ((u64)folio->index << PAGE_SHIFT) +
+					 folio_ofs);
 
 	if (bio_add_folio(bio, folio, vol->mft_record_size, folio_ofs))
 		err = submit_bio_wait(bio);
-- 
2.51.0

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE
  2026-09-05 12:35 [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE 朱 天浩
@ 2026-09-05 14:08 ` liubaolin
  2026-09-07  8:54 ` Hyunchul Lee
  1 sibling, 0 replies; 6+ messages in thread
From: liubaolin @ 2026-09-05 14:08 UTC (permalink / raw)
  To: 朱 天浩, linkinjeon, hyc.lee; +Cc: ntfs, linux-kernel



在 2026/9/5 20:35, 朱 天浩 写道:
> When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
> MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
> which is indeed what mkfs arranges.  However, the folio rewrite kept a
> leftover from the old buffer-head/runlist based implementation:
> vol->cluster_size_mask is still applied to the mirror write offset.
> In fact it is no longer needed -- and in some configurations it is
> actively wrong.
> 
> For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
> MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
> written to the location of record 0.  $MFTMirr therefore gets out of
> sync with $MFT and the following warning is printed on remount:
> 
> 	ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
> 	      record 0 do not match.  Run chkdsk.
> 
> Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
> to the base LCN address instead of applying the cluster_size_mask
> truncation.  This is the standard file-offset calculation and is
> correct for every combination of cluster size, page size and MFT
> record size, provided $MFTMirr data is contiguous from mftmirr_lcn
> (which mkfs always arranges).
> 
> While here, verify that the folio returned by read_mapping_folio() is
> large enough to hold the whole mirror record (folio_ofs +
> mft_record_size <= folio_size(folio)), guarding against records that
> straddle a folio boundary when mft_record_size exceeds PAGE_SIZE,
> although this cannot happen in practice.
> 
> Tested on the volume above: before the change mirror records 1-3 are
> all written to record 0's sector; after the change the write-back
> probe reports:
> 
> 	ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
> 	ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628
> 
> which matches the expected sectors computed as:
> 
> 	(NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT
> 
> 	record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
> 	record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640
> 
> A 13 MB file write succeeds on this volume and the file is byte-for-byte
> identical after a umount/mount cycle.
> 
> Fixes: 115380f9a2f9 ("ntfs: update mft operations")
> Assisted-by: UOS-AI
> Assisted-by: CodeBuddy:Hy4 preview
> Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
> ---
>   fs/ntfs/mft.c | 26 ++++++++++++++++----------
>   1 file changed, 16 insertions(+), 10 deletions(-)
> 
> diff --git a/fs/ntfs/mft.c b/fs/ntfs/mft.c
> index 7e58c99f1728..a62b1d41697c 100644
> --- a/fs/ntfs/mft.c
> +++ b/fs/ntfs/mft.c
> @@ -462,7 +462,7 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>   {
>   	u8 *kmirr;
>   	struct folio *folio;
> -	unsigned int folio_ofs, lcn_folio_off = 0;
> +	unsigned int folio_ofs;
>   	int err = 0;
>   	struct bio *bio;
>   
> @@ -482,25 +482,31 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>   		goto err_out;
>   	}
>   
> +	/* Offset of the mft mirror record inside the folio. */
> +	folio_ofs = NTFS_MFT_NR_TO_POFS(vol, mft_no);
> +
> +	/* The mirror record must fit entirely within this folio. */
> +	if (folio_ofs + vol->mft_record_size > folio_size(folio)) {
> +		ntfs_error(vol->sb, "Mft mirror record 0x%llx does not fit in folio.",
> +			   mft_no);
> +		folio_put(folio);
> +		err = -EIO;
> +		goto err_out;
> +	}
> +
>   	folio_lock(folio);
>   	folio_clear_uptodate(folio);
> -	/* Offset of the mft mirror record inside the page. */
> -	folio_ofs = NTFS_MFT_NR_TO_POFS(vol, mft_no);
> -	/* The address in the page of the mirror copy of the mft record @m. */
> +	/* The address in the folio of the mirror copy of the mft record @m. */
>   	kmirr = kmap_local_folio(folio, 0) + folio_ofs;
>   	/* Copy the mst protected mft record to the mirror. */
>   	memcpy(kmirr, m, vol->mft_record_size);
>   	kunmap_local(kmirr);
>   
> -	if (vol->cluster_size_bits > PAGE_SHIFT) {
> -		lcn_folio_off = folio->index << PAGE_SHIFT;
> -		lcn_folio_off &= vol->cluster_size_mask;
> -	}
> -
>   	bio = bio_alloc(vol->sb->s_bdev, 1, REQ_OP_WRITE, GFP_NOIO);
>   	bio->bi_iter.bi_sector =
>   		ntfs_bytes_to_bio_sector(NTFS_CLU_TO_B(vol, vol->mftmirr_lcn) +
> -					 lcn_folio_off + folio_ofs);
> +					 ((u64)folio->index << PAGE_SHIFT) +
> +					 folio_ofs);
>   
>   	if (bio_add_folio(bio, folio, vol->mft_record_size, folio_ofs))
>   		err = submit_bio_wait(bio);

Hi Tianhao,
   Thanks for submitting this patch.
   You identified a real issue where the write offset for $MFTMirr 
records can be calculated incorrectly on some configurations, which may 
leave $MFTMirr inconsistent with $MFT.

   I have reviewed the patch carefully, and I believe the proposed fix 
is  effective.

Reviewed-by: Baolin Liu <liubaolin@kylinos.cn>

Thanks,
Baolin.


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE
  2026-09-05 12:35 [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE 朱 天浩
  2026-09-05 14:08 ` liubaolin
@ 2026-09-07  8:54 ` Hyunchul Lee
  2026-09-07 20:41   ` [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios 朱 天浩
  1 sibling, 1 reply; 6+ messages in thread
From: Hyunchul Lee @ 2026-09-07  8:54 UTC (permalink / raw)
  To: 朱 天浩; +Cc: linkinjeon, ntfs, linux-kernel

Hi Tianhao,

2026년 9월 5일 (토) 오후 9:35, 朱 天浩 <zhutianhao75@hotmail.com>님이 작성:
>
> When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
> MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
> which is indeed what mkfs arranges.  However, the folio rewrite kept a
> leftover from the old buffer-head/runlist based implementation:
> vol->cluster_size_mask is still applied to the mirror write offset.
> In fact it is no longer needed -- and in some configurations it is
> actively wrong.
>
> For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
> MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
> written to the location of record 0.  $MFTMirr therefore gets out of
> sync with $MFT and the following warning is printed on remount:
>
>         ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
>               record 0 do not match.  Run chkdsk.
>
> Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
> to the base LCN address instead of applying the cluster_size_mask
> truncation.  This is the standard file-offset calculation and is
> correct for every combination of cluster size, page size and MFT
> record size, provided $MFTMirr data is contiguous from mftmirr_lcn
> (which mkfs always arranges).
>
> While here, verify that the folio returned by read_mapping_folio() is
> large enough to hold the whole mirror record (folio_ofs +
> mft_record_size <= folio_size(folio)), guarding against records that
> straddle a folio boundary when mft_record_size exceeds PAGE_SIZE,
> although this cannot happen in practice.
>
> Tested on the volume above: before the change mirror records 1-3 are
> all written to record 0's sector; after the change the write-back
> probe reports:
>
>         ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
>         ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628
>
> which matches the expected sectors computed as:
>
>         (NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT
>
>         record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
>         record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640
>
> A 13 MB file write succeeds on this volume and the file is byte-for-byte
> identical after a umount/mount cycle.
>
> Fixes: 115380f9a2f9 ("ntfs: update mft operations")
> Assisted-by: UOS-AI
> Assisted-by: CodeBuddy:Hy4 preview
> Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
> ---
>  fs/ntfs/mft.c | 26 ++++++++++++++++----------
>  1 file changed, 16 insertions(+), 10 deletions(-)
>
> diff --git a/fs/ntfs/mft.c b/fs/ntfs/mft.c
> index 7e58c99f1728..a62b1d41697c 100644
> --- a/fs/ntfs/mft.c
> +++ b/fs/ntfs/mft.c
> @@ -462,7 +462,7 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>  {
>         u8 *kmirr;
>         struct folio *folio;
> -       unsigned int folio_ofs, lcn_folio_off = 0;
> +       unsigned int folio_ofs;
>         int err = 0;
>         struct bio *bio;
>
> @@ -482,25 +482,31 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>                 goto err_out;
>         }
>
> +       /* Offset of the mft mirror record inside the folio. */
> +       folio_ofs = NTFS_MFT_NR_TO_POFS(vol, mft_no);
> +
> +       /* The mirror record must fit entirely within this folio. */
> +       if (folio_ofs + vol->mft_record_size > folio_size(folio)) {
> +               ntfs_error(vol->sb, "Mft mirror record 0x%llx does not fit in folio.",
> +                          mft_no);
> +               folio_put(folio);
> +               err = -EIO;
> +               goto err_out;
> +       }

vol->mft_record_size < PAGE_SIZE is guaranteed by ntfs_fill_super().
Is this check necessary?

Thanks,
Hyunchul

^ permalink raw reply	[flat|nested] 6+ messages in thread

* [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios
  2026-09-07  8:54 ` Hyunchul Lee
@ 2026-09-07 20:41   ` 朱 天浩
  2026-09-08  5:51     ` liubaolin
  2026-09-08  9:41     ` Namjae Jeon
  0 siblings, 2 replies; 6+ messages in thread
From: 朱 天浩 @ 2026-09-07 20:41 UTC (permalink / raw)
  To: Hyunchul Lee; +Cc: linkinjeon, ntfs, linux-kernel

When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
which is indeed what mkfs arranges.  However, the folio rewrite kept a
leftover from the old buffer-head/runlist based implementation:
vol->cluster_size_mask is still applied to the mirror write offset.
In fact it is no longer needed -- and in some configurations it is
actively wrong.

The bug triggers whenever the four mirror records do not fit in a
single folio, i.e. when mft_record_size exceeds PAGE_SIZE / 4 (for
example mft_record_size > 1KiB on 4KiB pages).

For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
written to the location of record 0.  $MFTMirr therefore gets out of
sync with $MFT and the following warning is printed on remount:

	ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
	      record 0 do not match.  Run chkdsk.

Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
to the base LCN address instead of applying the cluster_size_mask
truncation.  This is the standard file-offset calculation and is
correct for every combination of cluster size, page size and MFT
record size, provided $MFTMirr data is contiguous from mftmirr_lcn
(which mkfs always arranges).

Tested on the volume above: before the change mirror records 1-3 are
all written to record 0's sector; after the change the write-back
probe reports:

	ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
	ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628

which matches the expected sectors computed as:

	(NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT

	record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
	record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640

A 13 MB file write succeeds on this volume and the file is byte-for-byte
identical after a umount/mount cycle.

Fixes: 115380f9a2f9 ("ntfs: update mft operations")
Assisted-by: UOS-AI
Assisted-by: CodeBuddy:Hy4 preview
Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
---
v2:
- Removed the unreachable folio-size guard (thanks Hyunchul for the review)
- Dropped the "While here" paragraph from the commit message
- Clarified the trigger condition and retitled the subject

fs/ntfs/mft.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)

diff --git a/fs/ntfs/mft.c b/fs/ntfs/mft.c
index 7e58c99f1728..8399f07d45d6 100644
--- a/fs/ntfs/mft.c
+++ b/fs/ntfs/mft.c
@@ -462,7 +462,7 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
 {
 	u8 *kmirr;
 	struct folio *folio;
-	unsigned int folio_ofs, lcn_folio_off = 0;
+	unsigned int folio_ofs;
 	int err = 0;
 	struct bio *bio;
 
@@ -492,15 +492,11 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
 	memcpy(kmirr, m, vol->mft_record_size);
 	kunmap_local(kmirr);
 
-	if (vol->cluster_size_bits > PAGE_SHIFT) {
-		lcn_folio_off = folio->index << PAGE_SHIFT;
-		lcn_folio_off &= vol->cluster_size_mask;
-	}
-
 	bio = bio_alloc(vol->sb->s_bdev, 1, REQ_OP_WRITE, GFP_NOIO);
 	bio->bi_iter.bi_sector =
 		ntfs_bytes_to_bio_sector(NTFS_CLU_TO_B(vol, vol->mftmirr_lcn) +
-					 lcn_folio_off + folio_ofs);
+					 ((u64)folio->index << PAGE_SHIFT) +
+					 folio_ofs);
 
 	if (bio_add_folio(bio, folio, vol->mft_record_size, folio_ofs))
 		err = submit_bio_wait(bio);
-- 
2.51.0

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios
  2026-09-07 20:41   ` [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios 朱 天浩
@ 2026-09-08  5:51     ` liubaolin
  2026-09-08  9:41     ` Namjae Jeon
  1 sibling, 0 replies; 6+ messages in thread
From: liubaolin @ 2026-09-08  5:51 UTC (permalink / raw)
  To: 朱 天浩, Hyunchul Lee; +Cc: linkinjeon, ntfs, linux-kernel



在 2026/9/8 04:41, 朱 天浩 写道:
> When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
> MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
> which is indeed what mkfs arranges.  However, the folio rewrite kept a
> leftover from the old buffer-head/runlist based implementation:
> vol->cluster_size_mask is still applied to the mirror write offset.
> In fact it is no longer needed -- and in some configurations it is
> actively wrong.
> 
> The bug triggers whenever the four mirror records do not fit in a
> single folio, i.e. when mft_record_size exceeds PAGE_SIZE / 4 (for
> example mft_record_size > 1KiB on 4KiB pages).
> 
> For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
> MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
> written to the location of record 0.  $MFTMirr therefore gets out of
> sync with $MFT and the following warning is printed on remount:
> 
> 	ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
> 	      record 0 do not match.  Run chkdsk.
> 
> Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
> to the base LCN address instead of applying the cluster_size_mask
> truncation.  This is the standard file-offset calculation and is
> correct for every combination of cluster size, page size and MFT
> record size, provided $MFTMirr data is contiguous from mftmirr_lcn
> (which mkfs always arranges).
> 
> Tested on the volume above: before the change mirror records 1-3 are
> all written to record 0's sector; after the change the write-back
> probe reports:
> 
> 	ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
> 	ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628
> 
> which matches the expected sectors computed as:
> 
> 	(NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT
> 
> 	record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
> 	record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640
> 
> A 13 MB file write succeeds on this volume and the file is byte-for-byte
> identical after a umount/mount cycle.
> 
> Fixes: 115380f9a2f9 ("ntfs: update mft operations")
> Assisted-by: UOS-AI
> Assisted-by: CodeBuddy:Hy4 preview
> Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
> ---
> v2:
> - Removed the unreachable folio-size guard (thanks Hyunchul for the review)
> - Dropped the "While here" paragraph from the commit message
> - Clarified the trigger condition and retitled the subject
> 
> fs/ntfs/mft.c | 10 +++-------
>   1 file changed, 3 insertions(+), 7 deletions(-)
> 
> diff --git a/fs/ntfs/mft.c b/fs/ntfs/mft.c
> index 7e58c99f1728..8399f07d45d6 100644
> --- a/fs/ntfs/mft.c
> +++ b/fs/ntfs/mft.c
> @@ -462,7 +462,7 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>   {
>   	u8 *kmirr;
>   	struct folio *folio;
> -	unsigned int folio_ofs, lcn_folio_off = 0;
> +	unsigned int folio_ofs;
>   	int err = 0;
>   	struct bio *bio;
>   
> @@ -492,15 +492,11 @@ int ntfs_sync_mft_mirror(struct ntfs_volume *vol, const u64 mft_no,
>   	memcpy(kmirr, m, vol->mft_record_size);
>   	kunmap_local(kmirr);
>   
> -	if (vol->cluster_size_bits > PAGE_SHIFT) {
> -		lcn_folio_off = folio->index << PAGE_SHIFT;
> -		lcn_folio_off &= vol->cluster_size_mask;
> -	}
> -
>   	bio = bio_alloc(vol->sb->s_bdev, 1, REQ_OP_WRITE, GFP_NOIO);
>   	bio->bi_iter.bi_sector =
>   		ntfs_bytes_to_bio_sector(NTFS_CLU_TO_B(vol, vol->mftmirr_lcn) +
> -					 lcn_folio_off + folio_ofs);
> +					 ((u64)folio->index << PAGE_SHIFT) +
> +					 folio_ofs);
>   
>   	if (bio_add_folio(bio, folio, vol->mft_record_size, folio_ofs))
>   		err = submit_bio_wait(bio);


Looks good to me.

Reviewed-by: Baolin Liu <liubaolin@kylinos.cn>


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios
  2026-09-07 20:41   ` [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios 朱 天浩
  2026-09-08  5:51     ` liubaolin
@ 2026-09-08  9:41     ` Namjae Jeon
  1 sibling, 0 replies; 6+ messages in thread
From: Namjae Jeon @ 2026-09-08  9:41 UTC (permalink / raw)
  To: 朱 天浩; +Cc: Hyunchul Lee, ntfs, linux-kernel

On Tue, Sep 8, 2026 at 5:41 AM 朱 天浩 <zhutianhao75@hotmail.com> wrote:
>
> When commit 115380f9a2f9 ("ntfs: update mft operations") refactored the
> MFT code to use folios, it assumed $MFTMirr is allocated contiguously,
> which is indeed what mkfs arranges.  However, the folio rewrite kept a
> leftover from the old buffer-head/runlist based implementation:
> vol->cluster_size_mask is still applied to the mirror write offset.
> In fact it is no longer needed -- and in some configurations it is
> actively wrong.
>
> The bug triggers whenever the four mirror records do not fit in a
> single folio, i.e. when mft_record_size exceeds PAGE_SIZE / 4 (for
> example mft_record_size > 1KiB on 4KiB pages).
>
> For example, on the built-in 4Kn SSD (Apple SSD AP0256J) of a
> MacBookPro14,1 with 4096-byte MFT records, mirror record 3 is still
> written to the location of record 0.  $MFTMirr therefore gets out of
> sync with $MFT and the following warning is printed on remount:
>
>         ntfs: (device nvme0n1p3): check_mft_mirror(): $MFT and $MFTMirr
>               record 0 do not match.  Run chkdsk.
>
> Fix it by always adding the folio offset (folio->index << PAGE_SHIFT)
> to the base LCN address instead of applying the cluster_size_mask
> truncation.  This is the standard file-offset calculation and is
> correct for every combination of cluster size, page size and MFT
> record size, provided $MFTMirr data is contiguous from mftmirr_lcn
> (which mkfs always arranges).
>
> Tested on the volume above: before the change mirror records 1-3 are
> all written to record 0's sector; after the change the write-back
> probe reports:
>
>         ntfs: NTFSDBG pre mft_no=0x3 folio_idx=0x3 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0640
>         ntfs: NTFSDBG pre mft_no=0x0 folio_idx=0x0 ofs=0x0 mftmirr_lcn=0x2540c5 clu_bits=12 rec_bits=12 PAGE_SHIFT=12 sect=0x12a0628
>
> which matches the expected sectors computed as:
>
>         (NTFS_CLU_TO_B(mftmirr_lcn) + (folio->index << PAGE_SHIFT)) >> SECTOR_SHIFT
>
>         record 0: (0x2540c5 << 12) + 0x0000 = 0x2540c5000 >> 9 = 0x12a0628
>         record 3: (0x2540c5 << 12) + 0x3000 = 0x2540c8000 >> 9 = 0x12a0640
>
> A 13 MB file write succeeds on this volume and the file is byte-for-byte
> identical after a umount/mount cycle.
>
> Fixes: 115380f9a2f9 ("ntfs: update mft operations")
> Assisted-by: UOS-AI
> Assisted-by: CodeBuddy:Hy4 preview
> Signed-off-by: Zhu Tianhao <zhutianhao75@hotmail.com>
Applied it to #ntfs-next.
Thanks!

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2026-09-08  9:41 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-09-05 12:35 [PATCH] ntfs: fix $MFTMirr write offset for MFT record sizes larger than PAGE_SIZE 朱 天浩
2026-09-05 14:08 ` liubaolin
2026-09-07  8:54 ` Hyunchul Lee
2026-09-07 20:41   ` [PATCH v2] ntfs: fix $MFTMirr write offset when it spans multiple folios 朱 天浩
2026-09-08  5:51     ` liubaolin
2026-09-08  9:41     ` Namjae Jeon

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®