exfat: retain 'VolumeFlags' properly
Subsystems (2)
PATCH 1/1 exfat: retain 'VolumeFlags' properly
Retain ActiveFat, MediaFailure and ClearToZero fields. And, never clear VolumeDirty, if it is dirty at mount.
In '3.1.13.3 Media Failure Field' of exfat specification says ...
If, upon mounting a volume, the value of this field is 1, implementations which scan the entire volume for media failures and record all failures as "bad" clusters in the FAT (or otherwise resolve media failures) may clear the value of this field to 0.
Therefore, should not clear MediaFailure without scanning volume.
In '8.1 Recommended Write Ordering' of exfat specification says ...
Clear the value of the VolumeDirty field to 0, if its value prior to the first step was 0
Therefore, should not clear VolumeDirty when mounted.
Also, rename ERR_MEDIUM to MED_FAILURE.
--- fs/exfat/exfat_fs.h | 5 +++-- fs/exfat/exfat_raw.h | 2 +- fs/exfat/super.c | 22 ++++++++++++++-------- 3 files changed, 18 insertions(+), 11 deletions(-)
fs/exfat/exfat_fs.h +3 -2
@@ -224,7 +224,8 @@ struct exfat_sb_info {unsigned int num_FAT_sectors;/* num of FAT sectors */ unsigned int root_dir;/* root dir cluster */ unsigned int dentries_per_clu;/* num of dentries per cluster */ -unsigned int vol_flag;/* volume dirty flag */ +unsigned int vol_flags;/* volume flags */ +unsigned int vol_flags_noclear;/* volume flags to retain */ struct buffer_head *boot_bh;/* buffer_head of BOOT sector */ unsigned int map_clu;/* allocation bitmap start cluster */
@@ -380,7 +381,7 @@ static inline int exfat_sector_to_cluster(struct exfat_sb_info *sbi, }/* super.c */ -int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flag); +int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flags);/* fatent.c */ #define exfat_get_next_cluster (sb, pclu) exfat_ent_get(sb, *(pclu), pclu)
fs/exfat/exfat_raw.h +1 -1
@@ -16,7 +16,7 @@#define VOL_CLEAN 0x0000#define VOL_DIRTY 0x0002 -#define ERR_MEDIUM 0x0004 +#define MED_FAILURE 0x0004#define EXFAT_EOF_CLUSTER 0xFFFFFFFFu#define EXFAT_BAD_CLUSTER 0xFFFFFFF7u
fs/exfat/super.c +14 -8
@@ -96,17 +96,22 @@ static int exfat_statfs(struct dentry *dentry, struct kstatfs *buf)return 0 ; } -int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flag) +int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flags) {struct exfat_sb_info *sbi =EXFAT_SB (sb);struct boot_sector *p_boot = (struct boot_sector *)sbi->boot_bh->b_data;bool sync; +if (new_flags ==VOL_CLEAN ) + new_flags = (sbi->vol_flags & ~VOL_DIRTY ) | sbi->vol_flags_noclear; +else + new_flags |= sbi->vol_flags; +/* flags are not changed */ -if (sbi->vol_flag == new_flag) +if (sbi->vol_flags == new_flags)return 0 ; - sbi->vol_flag = new_flag; + sbi->vol_flags = new_flags;/* skip updating volume dirty flag, * if this volume has been mounted with read-only
@@ -114,9 +119,9 @@ int exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)if (sb_rdonly (sb))return 0 ; - p_boot->vol_flags =cpu_to_le16 (new_flag); + p_boot->vol_flags =cpu_to_le16 (new_flags); -if (new_flag ==VOL_DIRTY && !buffer_dirty (sbi->boot_bh)) +if ((new_flags &VOL_DIRTY ) && !buffer_dirty (sbi->boot_bh)) sync = true;else sync = false;
@@ -457,7 +462,8 @@ static int exfat_read_boot_sector(struct super_block *sb) sbi->dentries_per_clu =1 << (sbi->cluster_size_bits -DENTRY_SIZE_BITS ); - sbi->vol_flag =le16_to_cpu (p_boot->vol_flags); + sbi->vol_flags =le16_to_cpu (p_boot->vol_flags); + sbi->vol_flags_noclear = sbi->vol_flags & (VOL_DIRTY |MED_FAILURE ); sbi->clu_srch_ptr =EXFAT_FIRST_CLUSTER ; sbi->used_clusters =EXFAT_CLUSTERS_UNTRACKED ;
@@ -472,9 +478,9 @@ static int exfat_read_boot_sector(struct super_block *sb)exfat_err (sb,"bogus data start sector" );return -EINVAL ; } -if (sbi->vol_flag &VOL_DIRTY ) +if (sbi->vol_flags &VOL_DIRTY )exfat_warn (sb,"Volume was not properly unmounted. Some data may be corrupt. Please run fsck." ); -if (sbi->vol_flag &ERR_MEDIUM ) +if (sbi->vol_flags &MED_FAILURE )exfat_warn (sb,"Medium has reported failures. Some data may be lost." );/* exFAT file size is limited by a disk volume size */
Signature · 1 line
2.25.1
Hi Tetsuhiro,
Quoted 9 lines
Retain ActiveFat, MediaFailure and ClearToZero fields. And, never clear VolumeDirty, if it is dirty at mount.
In '3.1.13.3 Media Failure Field' of exfat specification says ...
If, upon mounting a volume, the value of this field is 1, implementations which scan the entire
volume for media failures and record all failures as "bad" clusters in the FAT (or otherwise resolve media failures) may clear the value of this field to 0.
Therefore, should not clear MediaFailure without scanning volume.
Quoted 5 lines
In '8.1 Recommended Write Ordering' of exfat specification says ...
Clear the value of the VolumeDirty field to 0, if its value prior to the first step was 0 Therefore,
should not clear VolumeDirty when mounted.Also, rename
ERR_MEDIUMtoMED_FAILURE.
I think that MEDIA_FAILURE looks better.
Quoted 46 lines
Signed-off-by Tetsuhiro Kohada <[email protected]>--- fs/exfat/exfat_fs.h | 5 +++-- fs/exfat/exfat_raw.h | 2 +- fs/exfat/super.c | 22 ++++++++++++++-------- 3 files changed, 18 insertions(+), 11 deletions(-)fs/exfat/exfat_fs.h +2
-1@@ -224,7 +224,8 @@ struct exfat_sb_info {unsigned int num_FAT_sectors;/* num of FAT sectors */ unsigned int root_dir;/* root dir cluster */ unsigned int dentries_per_clu;/* num of dentries per cluster */ -unsigned int vol_flag;/* volume dirty flag */ +unsigned int vol_flags;/* volume flags */ +unsigned int vol_flags_noclear;/* volume flags to retain */ struct buffer_head *boot_bh;/* buffer_head of BOOT sector */ unsigned int map_clu; /* allocation bitmap start cluster */ @@ -380,7 +381,7 @@ static inlineint exfat_sector_to_cluster(struct
exfat_sb_info*sbi, }/*
super.c*/fs/exfat/exfat_fs.h +2
-1-int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flag); +int exfat_set_vol_flags (struct super_block *sb,unsigned short +new_flags);/* fatent.c */ #define exfat_get_next_cluster (sb, pclu) exfat_ent_get(sb, *(pclu), pclu) diff --git
a/fs/exfat/exfat_raw.hb/fs/exfat/exfat_raw.hindex350ce59cc324..d86a8a6b0601100644fs/exfat/exfat_raw.h +1
-1@@ -16,7 +16,7 @@#define VOL_CLEAN 0x0000#define VOL_DIRTY 0x0002 -#define ERR_MEDIUM 0x0004 +#define MED_FAILURE 0x0004#define EXFAT_EOF_CLUSTER 0xFFFFFFFFu#define EXFAT_BAD_CLUSTER 0xFFFFFFF7u
fs/exfat/super.c static int exfat_statfs(struct dentry *dentry, struct kstatfs *buf) · line 104
Quoted 10 lines
fs/exfat/super.c +2
-1@@ -96,17 +96,22 @@ static int exfat_statfs(struct dentry *dentry, struct kstatfs *buf)return 0 ; } -int exfat_set_vol_flags (struct super_block *sb,unsigned short new_flag) +int exfat_set_vol_flags (struct super_block *sb,unsigned short +new_flags) {struct exfat_sb_info *sbi =EXFAT_SB (sb);struct boot_sector *p_boot = (struct boot_sector *)sbi->boot_bh->b_data;bool sync;
If dirty bit is set in on-disk volume flags, we can just return 0 at the beginning of this function ?
Quoted 17 lines
fs/exfat/super.c +7
-2+if (new_flags ==VOL_CLEAN ) + new_flags = (sbi->vol_flags & ~VOL_DIRTY ) | sbi->vol_flags_noclear; +else + new_flags |= sbi->vol_flags; +/* flags are not changed */ -if (sbi->vol_flag == new_flag) +if (sbi->vol_flags == new_flags)return 0 ; - sbi->vol_flag = new_flag; + sbi->vol_flags = new_flags;/* skip updating volume dirty flag, * if this volume has been mounted with read-only @@ -114,9 +119,9 @@ int
exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)
fs/exfat/super.c int exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag) · line 122
Quoted 5 lines
fs/exfat/super.c +1
-1if (sb_rdonly (sb))return 0 ; - p_boot->vol_flags =cpu_to_le16 (new_flag); + p_boot->vol_flags =cpu_to_le16 (new_flags);
How about set or clear only dirty bit to on-disk volume flags instead of using sbi->vol_flags_noclear ?
if (set)
p_boot->vol_flags |= cpu_to_le16(VOL_DIRTY);
else
p_boot->vol_flags &= cpu_to_le16(~VOL_DIRTY);Quoted 31 lines
fs/exfat/super.c +5
-4-if (new_flag ==VOL_DIRTY && !buffer_dirty (sbi->boot_bh)) +if ((new_flags &VOL_DIRTY ) && !buffer_dirty (sbi->boot_bh)) sync = true;else sync = false;@@ -457,7 +462,8 @@ static int exfat_read_boot_sector(struct super_block *sb) sbi->dentries_per_clu =1 << (sbi->cluster_size_bits -DENTRY_SIZE_BITS ); - sbi->vol_flag =le16_to_cpu (p_boot->vol_flags); + sbi->vol_flags =le16_to_cpu (p_boot->vol_flags); + sbi->vol_flags_noclear = sbi->vol_flags & (VOL_DIRTY |MED_FAILURE ); sbi->clu_srch_ptr =EXFAT_FIRST_CLUSTER ; sbi->used_clusters =EXFAT_CLUSTERS_UNTRACKED ;@@ -472,9 +478,9 @@ static int exfat_read_boot_sector(struct super_block *sb)exfat_err (sb,"bogus data start sector" );return -EINVAL ; } -if (sbi->vol_flag &VOL_DIRTY ) +if (sbi->vol_flags &VOL_DIRTY ) exfat_warn(sb, "Volume was not properly unmounted. Some data may be corrupt. Please run fsck."); -if (sbi->vol_flag &ERR_MEDIUM ) +if (sbi->vol_flags &MED_FAILURE )exfat_warn (sb,"Medium has reported failures. Some data may be lost." );/* exFAT file size is limited by a disk volume size */
--
2.25.1
Thanks for your reply.
Quoted 2 lines
Quoted 1 line
Also, rename
ERR_MEDIUMtoMED_FAILURE.I think that
MEDIA_FAILURElooks better.
I think so too.
If so, should I change VOL_DIRTY to VOLUME_DIRTY?
Quoted 9 lines
fs/exfat/super.c static int exfat_statfs(struct dentry *dentry, struct kstatfs *buf) · line 103
Quoted 8 lines
-int
exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)fs/exfat/super.c +2
-0+int exfat_set_vol_flags (struct super_block *sb,unsigned short +new_flags) {struct exfat_sb_info *sbi =EXFAT_SB (sb);struct boot_sector *p_boot = (struct boot_sector *)sbi->boot_bh->b_data;bool sync;If dirty bit is set in on-disk volume flags, we can just return 0 at the beginning of this function ?
That's right.
Neither 'set VOL_DIRTY' nor 'set VOL_CLEAN' makes a change to volume flags.
Quoted 19 lines
Quoted 19 lines
fs/exfat/super.c +7
-2+if (new_flags ==VOL_CLEAN ) + new_flags = (sbi->vol_flags & ~VOL_DIRTY ) | sbi->vol_flags_noclear; +else + new_flags |= sbi->vol_flags; +/* flags are not changed */ -if (sbi->vol_flag == new_flag) +if (sbi->vol_flags == new_flags)return 0 ; - sbi->vol_flag = new_flag; + sbi->vol_flags = new_flags;/* skip updating volume dirty flag, * if this volume has been mounted with read-only @@ -114,9 +119,9 @@ int
exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)fs/exfat/super.c +0
-0if (sb_rdonly (sb))return 0 ;
Quoted 8 lines
fs/exfat/super.c int exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag) · line 122
Quoted 2 lines
fs/exfat/super.c +1
-1- p_boot->vol_flags =cpu_to_le16 (new_flag); + p_boot->vol_flags =cpu_to_le16 (new_flags);How about set or clear only dirty bit to on-disk volume flags instead of using
sbi->vol_flags_noclear?if (set) p_boot->vol_flags |= cpu_to_le16(VOL_DIRTY); else p_boot->vol_flags &= cpu_to_le16(~VOL_DIRTY);
In this way, the initial VOL_DIRTY cannot be retained.
The purpose of this patch is to avoid losing the initial VOL_DIRTY and MED_FAILURE. Furthermore, I'm also thinking of setting MED_FAILURE.
However, the formula for 'new_flags' may be a bit complicated. Is it better to change to the following?
if (new_flags == VOL_CLEAN) new_flags = sbi->vol_flags & ~VOL_DIRTY else new_flags |= sbi->vol_flags; new_flags |= sbi->vol_flags_noclear;
one more point,
Is there a better name than 'vol_flags_noclear'?
(I can't come up with a good name anymore)
BR
---
Kohada Tetsuhiro <[email protected]>
Quoted 7 lines
Thanks for your reply.
Quoted 2 lines
Quoted 1 line
Also, rename
ERR_MEDIUMtoMED_FAILURE.I think that
MEDIA_FAILURElooks better.I think so too.
If so, should I changeVOL_DIRTYtoVOLUME_DIRTY?
Yes, maybe.
Quoted 44 lines
Quoted 10 lines
fs/exfat/super.c static int exfat_statfs(struct dentry *dentry, struct kstatfs *buf) · line 103
Quoted 9 lines
-int
exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)fs/exfat/super.c +2
-0+int exfat_set_vol_flags (struct super_block *sb,unsigned short +new_flags) {struct exfat_sb_info *sbi =EXFAT_SB (sb);struct boot_sector *p_boot = (struct boot_sector *)sbi->boot_bh->b_data;bool sync;If dirty bit is set in on-disk volume flags, we can just return 0 at the beginning of this function ?
That's right.
Neither 'setVOL_DIRTY' nor 'setVOL_CLEAN' makes a change to volume flags.Quoted 24 lines
Quoted 16 lines
fs/exfat/super.c +7
-2+if (new_flags ==VOL_CLEAN ) + new_flags = (sbi->vol_flags & ~VOL_DIRTY ) | sbi->vol_flags_noclear; +else + new_flags |= sbi->vol_flags; +/* flags are not changed */ -if (sbi->vol_flag == new_flag) +if (sbi->vol_flags == new_flags)return 0 ; - sbi->vol_flag = new_flag; + sbi->vol_flags = new_flags;/* skip updating volume dirty flag, * if this volume has been mounted with read-only @@ -114,9 +119,9 @@ int
exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag)fs/exfat/super.c int exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag) · line 122
Quoted 5 lines
fs/exfat/super.c +1
-1if (sb_rdonly (sb))return 0 ; - p_boot->vol_flags =cpu_to_le16 (new_flag); + p_boot->vol_flags =cpu_to_le16 (new_flags);How about set or clear only dirty bit to on-disk volume flags instead of using
sbi->vol_flags_noclear?Quoted 4 lines
if (set) p_boot->vol_flags |= cpu_to_le16(VOL_DIRTY); else p_boot->vol_flags &= cpu_to_le16(~VOL_DIRTY);
Quoted 3 lines
In this way, the initial
VOL_DIRTYcannot be retained.
The purpose of this patch is to avoid losing the initialVOL_DIRTYandMED_FAILURE. Furthermore, I'm also thinking of settingMED_FAILURE.
I know what you do. I mean this function does not need to be called if volume dirty Is already set on on-disk volume flags as I said earlier.
Quoted 12 lines
However, the formula for '
new_flags' may be a bit complicated. Is it better to change to the following?if (new_flags == VOL_CLEAN) new_flags = sbi->vol_flags & ~VOL_DIRTY else new_flags |= sbi->vol_flags; new_flags |= sbi->vol_flags_noclear;
Quoted 3 lines
one more point,
Is there a better name than 'vol_flags_noclear'?
(I can't come up with a good name anymore)
It looks complicated. It would be better to simply set/clear VOLUME DIRTY bit.
Thanks!
Quoted 4 lines
BR
---
Kohada Tetsuhiro <[email protected]>
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
Quoted 2 lines
Also, rename
ERR_MEDIUMtoMED_FAILURE.I think that
MEDIA_FAILURElooks better.I think so too.
If so, should I changeVOL_DIRTYtoVOLUME_DIRTY?Yes, maybe.
OK.
I'll rename both in v2.
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
fs/exfat/super.c int exfat_set_vol_flags(struct super_block *sb, unsigned short new_flag) · line 122
Quoted 2 lines
fs/exfat/super.c +1
-1- p_boot->vol_flags =cpu_to_le16 (new_flag); + p_boot->vol_flags =cpu_to_le16 (new_flags);How about set or clear only dirty bit to on-disk volume flags instead of using
sbi->vol_flags_noclear?if (set) p_boot->vol_flags |= cpu_to_le16(VOL_DIRTY); else p_boot->vol_flags &= cpu_to_le16(~VOL_DIRTY);
Please let me know about this code.
Does this code assume that 'sbi->vol_flags'(last vol_flag value) is not used?
If you use sbi->vol_flags, I think the original idea is fine.
sbi-> vol_flags = new_flag; p_boot->vol_flags = cpu_to_le16(new_flag);
Quoted 5 lines
Quoted 3 lines
In this way, the initial
VOL_DIRTYcannot be retained.
The purpose of this patch is to avoid losing the initialVOL_DIRTYandMED_FAILURE. Furthermore, I'm also thinking of settingMED_FAILURE.I know what you do. I mean this function does not need to be called if volume dirty Is already set on on-disk volume flags as I said earlier.
Hmm?
Does it mean the caller check that volume was dirty at mount, and caller determine whether to call exfat_set_vol_flags() or not?
If so, MEDIA_FAILUR needs to be set independently of the volume-dirty state.
Quoted 8 lines
Quoted 8 lines
However, the formula for '
new_flags' may be a bit complicated. Is it better to change to the following?if (new_flags == VOL_CLEAN) new_flags = sbi->vol_flags & ~VOL_DIRTY else new_flags |= sbi->vol_flags;
Quoted 7 lines
Quoted 6 lines
new_flags |= sbi->vol_flags_noclear;one more point,
Is there a better name than 'vol_flags_noclear'?
(I can't come up with a good name anymore)It looks complicated. It would be better to simply set/clear VOLUME DIRTY bit.
I think exfat_set_vol_flags() gets a little complicated, because it needs the followings (with bit operation)
a) Set/Clear VOLUME_DIRTY.
b) Set MEDIA_FAILUR.
c) Do not change other flags.
d) Retain VOLUME_DIRTY/MEDIA_FAILUR as it is when mounted.
'vol_flags_noclear' is used for d).
Bit-by-bit operation makes the code redundant. I think it's not a bad way to handle multiple bits.
What do you think?
BR
---
Tetsuhiro Kohada <[email protected]>
Ping..
Quoted 6 lines · Tetsuhiro Kohada
Quoted 1 line
It looks complicated. It would be better to simply set/clear VOLUME DIRTY bit.
I think
exfat_set_vol_flags()gets a little complicated, because it needs the followings (with bit operation)
a) Set/ClearVOLUME_DIRTY.
b) SetMEDIA_FAILUR.
How about splitting these into separate functions as below?
exfat_set_volume_dirty()
exfat_set_vol_flags(sb, sbi->vol_flag | VOLUME_DIRTY);
exfat_clear_volume_dirty()
exfat_set_vol_flags(sb, sbi->vol_flag & ~VOLUME_DIRTY);
exfat_set_media_failure()
exfat_set_vol_flags(sb, sbi->vol_flag | MEDIA_FAILURE);
The implementation is essentially the same for exfat_set_vol_flags(), but I think the intention of the operation will be easier to understand.
BR
---
Tetsuhiro Kohada <[email protected]>
Quoted 1 line
Ping..
Hi Tetsuhiro,
Quoted 9 lines · Tetsuhiro Kohada
Quoted 6 lines
Quoted 1 line
It looks complicated. It would be better to simply set/clear VOLUME DIRTY bit.
I think
exfat_set_vol_flags()gets a little complicated, because it needs the followings (with bit operation)
a) Set/ClearVOLUME_DIRTY.
b) SetMEDIA_FAILUR.
Quoted 8 lines
How about splitting these into separate functions as below?
exfat_set_volume_dirty()exfat_set_vol_flags(sb, sbi->vol_flag | VOLUME_DIRTY);
exfat_clear_volume_dirty()exfat_set_vol_flags(sb, sbi->vol_flag & ~VOLUME_DIRTY);
Looks good.
Quoted 3 lines
exfat_set_media_failure()exfat_set_vol_flags(sb, sbi->vol_flag | MEDIA_FAILURE);
Where will this function be called? We don't need to create unused functions in advance...
Quoted 4 lines
The implementation is essentially the same for
exfat_set_vol_flags(), but I think the intention of the operation will be easier to understand.
Yes.
Thanks!
Quoted 5 lines
BR
---
Tetsuhiro Kohada <[email protected]>
Quoted 2 lines · Namjae Jeon
Quoted 1 line
Ping..
Hi Tetsuhiro,
Thank you for your reply.
Quoted 14 lines
Quoted 14 lines · Tetsuhiro Kohada
Quoted 6 lines
Quoted 1 line
It looks complicated. It would be better to simply set/clear VOLUME DIRTY bit.
I think
exfat_set_vol_flags()gets a little complicated, because it needs the followings (with bit operation)
a) Set/ClearVOLUME_DIRTY.
b) SetMEDIA_FAILUR.How about splitting these into separate functions as below?
exfat_set_volume_dirty()exfat_set_vol_flags(sb, sbi->vol_flag | VOLUME_DIRTY);
Quoted 8 lines
Quoted 2 lines
exfat_clear_volume_dirty()exfat_set_vol_flags(sb, sbi->vol_flag & ~VOLUME_DIRTY);Looks good.
Quoted 3 lines
exfat_set_media_failure()exfat_set_vol_flags(sb, sbi->vol_flag | MEDIA_FAILURE);Where will this function be called? We don't need to create unused functions in advance...
Sorry. I ran ahead without explaining.
I would like to set MEDIA_FAILURE when a format error is detected so that fsck will be run on the next mount. This patch is the basis for setting MEDIA_FAILURE.
But, I have no reason to implement this right now, as you say. I'll add it in a patch that sets MEDIA_FAILURE.
BTW
I would like your opinion on the timing of clearing VOLUME_DIRTY.
https://lore.kernel.org/linux-fsdevel/[email protected]/#t
BR
---
Tetsuhiro Kohada <[email protected]>