Skip to contentSkip to conversation

hfsplus: rework processing of hfs_btree_write() returned error

PATCH In review Vyacheslav Dubeyko 3 participants 5 messages Nov 30, 2012 – Dec 3, 2012

Subsystems (2)
lore
Share & tools
lore

Vyacheslav Dubeyko

author maintainer
#0 lore raw

PATCH 1/1 hfsplus: rework processing of hfs_btree_write() returned error

Hi,

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching.

With the best regards,
Vyacheslav Dubeyko.
--
From: Vyacheslav Dubeyko <[email protected]>
Subject: [PATCH] hfsplus: rework processing of hfs_btree_write() returned error

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching. It is added also logic of processing errors of hfs_btree_write() in hfsplus_system_write_inode() with message about b-tree writing failure.

---
 fs/hfsplus/btree.c      |    5 +++--
 fs/hfsplus/hfsplus_fs.h |    2 +-
 fs/hfsplus/super.c      |   12 ++++++++++--
 3 files changed, 14 insertions(+), 5 deletions(-)
fs/hfsplus/btree.c +3 -2
index 21023d9..685d07d 100644
@@ -159,7 +159,7 @@ void hfs_btree_close(struct hfs_btree *tree)
 	kfree(tree);
 }
 
-void hfs_btree_write(struct hfs_btree *tree)
+int hfs_btree_write(struct hfs_btree *tree)
 {
 	struct hfs_btree_header_rec *head;
 	struct hfs_bnode *node;
@@ -168,7 +168,7 @@ void hfs_btree_write(struct hfs_btree *tree)
 	node = hfs_bnode_find(tree, 0);
 	if (IS_ERR(node))
 		/* panic? */
-		return;
+		return -EIO;
 	/* Load the header */
 	page = node->page[0];
 	head = (struct hfs_btree_header_rec *)(kmap(page) +
@@ -186,6 +186,7 @@ void hfs_btree_write(struct hfs_btree *tree)
 	kunmap(page);
 	set_page_dirty(page);
 	hfs_bnode_put(node);
+	return 0;
 }
 
 static struct hfs_bnode *hfs_bmap_new_bmap(struct hfs_bnode *prev, u32 idx)
fs/hfsplus/hfsplus_fs.h +1 -1
index c571de2..a6da86b 100644
@@ -335,7 +335,7 @@ int hfsplus_block_free(struct super_block *, u32, u32);
 /* btree.c */
 struct hfs_btree *hfs_btree_open(struct super_block *, u32);
 void hfs_btree_close(struct hfs_btree *);
-void hfs_btree_write(struct hfs_btree *);
+int hfs_btree_write(struct hfs_btree *);
 struct hfs_bnode *hfs_bmap_alloc(struct hfs_btree *);
 void hfs_bmap_free(struct hfs_bnode *);
 
fs/hfsplus/super.c +10 -2
index 811a84d..3033b73 100644
@@ -99,6 +99,7 @@ static int hfsplus_system_write_inode(struct inode *inode)
 	struct hfsplus_vh *vhdr = sbi->s_vhdr;
 	struct hfsplus_fork_raw *fork;
 	struct hfs_btree *tree = NULL;
+	int err;
 
 	switch (inode->i_ino) {
 	case HFSPLUS_EXT_CNID:
@@ -127,8 +128,15 @@ static int hfsplus_system_write_inode(struct inode *inode)
 		hfsplus_mark_mdb_dirty(inode->i_sb);
 	}
 	hfsplus_inode_write_fork(inode, fork);
-	if (tree)
-		hfs_btree_write(tree);
+	if (tree) {
+		err = hfs_btree_write(tree);
+		if (err) {
+			printk(KERN_ERR "hfs: unable to write b-tree\n");
+			dprint(DBG_INODE, "hfsplus_system_write_inode: %lu\n",
+				inode->i_ino);
+			return err;
+		}
+	}
 	return 0;
 }
 
Signature · 1 line
1.7.9.5
#1 lore raw
Quoted 10 lines · On Fri, 30 Nov 2012 14:49:26 +0400 Vyacheslav Dubeyko <[email protected]> wrote:

Hi,

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching.

With the best regards,
Vyacheslav Dubeyko.
--
From: Vyacheslav Dubeyko <[email protected]>
Subject: [PATCH] hfsplus: rework processing of hfs_btree_write() returned error

Quoted 3 lines

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching. It is added also logic of processing errors of hfs_btree_write() in hfsplus_system_write_inode() with message about b-tree writing failure.

We don't need two changelogs for each patch ;) The patch's formal changelog text should be sufficient.

Quoted 4 lines
fs/hfsplus/btree.c      |    5 +++--
fs/hfsplus/hfsplus_fs.h |    2 +-
fs/hfsplus/super.c      |   12 ++++++++++--
3 files changed, 14 insertions(+), 5 deletions(-)

I modified a couple of things, please review:

From: Andrew Morton <[email protected]>
Subject: hfsplus-rework-processing-of-hfs_btree_write-returned-error-fix

reduce scope of `err', print errno on error

Cc Al Viro <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Hin-Tak Leung <[email protected]>
Cc Vyacheslav Dubeyko <[email protected]>

---

 fs/hfsplus/super.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
fs/hfsplus/super.c renamed from fs/hfsplus/super.c~hfsplus-rework-processing-of-hfs_btree_write-returned-error-fix +3 -3
@@ -100,7 +100,6 @@ static int hfsplus_system_write_inode(st
 	struct hfsplus_vh *vhdr = sbi->s_vhdr;
 	struct hfsplus_fork_raw *fork;
 	struct hfs_btree *tree = NULL;
-	int err;
 
 	switch (inode->i_ino) {
 	case HFSPLUS_EXT_CNID:
@@ -131,9 +130,10 @@ static int hfsplus_system_write_inode(st
 	}
 	hfsplus_inode_write_fork(inode, fork);
 	if (tree) {
-		err = hfs_btree_write(tree);
+		int err = hfs_btree_write(tree);
 		if (err) {
-			printk(KERN_ERR "hfs: unable to write b-tree\n");
+			printk(KERN_ERR "hfs: unable to write b-tree: %d\n",
+				err);
 			dprint(DBG_INODE, "hfsplus_system_write_inode: %lu\n",
 				inode->i_ino);
 			return err;

_

author maintainer Reviewed-by
#2 lore raw
Quoted 13 lines · Andrew Morton
Quoted 10 lines · On Fri, 30 Nov 2012 14:49:26 +0400 Vyacheslav Dubeyko <[email protected]> wrote:

Hi,

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching.

With the best regards,
Vyacheslav Dubeyko.
--
From: Vyacheslav Dubeyko <[email protected]>
Subject: [PATCH] hfsplus: rework processing of hfs_btree_write() returned error

Quoted 7 lines
Quoted 3 lines

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching. It is added also logic of processing errors of hfs_btree_write() in hfsplus_system_write_inode() with message about b-tree writing failure.

We don't need two changelogs for each patch ;) The patch's formal changelog text should be sufficient.

Ok. :-)

Quoted 7 lines
Quoted 4 lines
fs/hfsplus/btree.c      |    5 +++--
fs/hfsplus/hfsplus_fs.h |    2 +-
fs/hfsplus/super.c      |   12 ++++++++++--
3 files changed, 14 insertions(+), 5 deletions(-)

I modified a couple of things, please review:

I think that it is a good correction. I agree.

With the best regards,
Vyacheslav Dubeyko.

Quoted 44 lines

From: Andrew Morton <[email protected]>
Subject: hfsplus-rework-processing-of-hfs_btree_write-returned-error-fix

reduce scope of `err', print errno on error

Cc Al Viro <[email protected]>
Cc Christoph Hellwig <[email protected]>
Cc Hin-Tak Leung <[email protected]>
Cc Vyacheslav Dubeyko <[email protected]>

---

fs/hfsplus/super.c | 6 +++---

1 file changed, 3 insertions(+), 3 deletions(-)
fs/hfsplus/super.c renamed from fs/hfsplus/super.c~hfsplus-rework-processing-of-hfs_btree_write-returned-error-fix +3 -3
@@ -100,7 +100,6 @@ static int hfsplus_system_write_inode(st
 	struct hfsplus_vh *vhdr = sbi->s_vhdr;
 	struct hfsplus_fork_raw *fork;
 	struct hfs_btree *tree = NULL;
-	int err;
 
 	switch (inode->i_ino) {
 	case HFSPLUS_EXT_CNID:
@@ -131,9 +130,10 @@ static int hfsplus_system_write_inode(st
 	}
 	hfsplus_inode_write_fork(inode, fork);
 	if (tree) {
-		err = hfs_btree_write(tree);
+		int err = hfs_btree_write(tree);
 		if (err) {
-			printk(KERN_ERR "hfs: unable to write b-tree\n");
+			printk(KERN_ERR "hfs: unable to write b-tree: %d\n",
+				err);
 			dprint(DBG_INODE, "hfsplus_system_write_inode: %lu\n",
 				inode->i_ino);
 			return err;

_

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

Acked-by on v1 1/1
#3 lore raw
In reply to Vyacheslav Dubeyko

Acked, and also a comment towards the end (so please keep reading beyond the Ack!).

Quoted 12 lines · Vyacheslav Dubeyko

Hi,

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching.

With the best regards,
Vyacheslav Dubeyko.
--
From: Vyacheslav Dubeyko <[email protected]>
Subject: [PATCH] hfsplus: rework processing of
hfs_btree_write() returned error

Quoted 7 lines

The patch adds in hfs_btree_write() returning of -EIO in the case of failing of b-tree's node searching. It is added also logic of processing errors of hfs_btree_write() in hfsplus_system_write_inode() with message about b-tree writing failure.

Quoted 90 lines
---
 fs/hfsplus/btree.c      |    5
+++--
 fs/hfsplus/hfsplus_fs.h |    2 +-
 fs/hfsplus/super.c     
|   12 ++++++++++--
 3 files changed, 14 insertions(+), 5 deletions(-)
fs/hfsplus/btree.c +1 -1
index 21023d9..685d07d 100644
@@ -159,7 +159,7 @@ void hfs_btree_close(struct hfs_btree
 *tree)
     kfree(tree);
 }
 
-void hfs_btree_write(struct hfs_btree *tree)
+int hfs_btree_write(struct hfs_btree *tree)
 {
     struct hfs_btree_header_rec *head;

    struct hfs_bnode *node;

fs/hfsplus/btree.c +1 -1
@@ -168,7 +168,7 @@ void hfs_btree_write(struct hfs_btree
 *tree)
     node = hfs_bnode_find(tree, 0);
     if (IS_ERR(node))
         /* panic? */
-        return;
+        return -EIO;
     /* Load the header */
     page = node->page[0];

    head = (struct hfs_btree_header_rec
*)(kmap(page) +

fs/hfsplus/btree.c +1 -0
@@ -186,6 +186,7 @@ void hfs_btree_write(struct hfs_btree
 *tree)
     kunmap(page);
     set_page_dirty(page);
     hfs_bnode_put(node);
+    return 0;
 }
 

static struct hfs_bnode *hfs_bmap_new_bmap(struct hfs_bnode
*prev, u32 idx)

fs/hfsplus/hfsplus_fs.h +1 -1
index c571de2..a6da86b 100644
@@ -335,7 +335,7 @@ int hfsplus_block_free(struct
 super_block *, u32, u32);
 /* btree.c */
 struct hfs_btree *hfs_btree_open(struct super_block *,
 u32);
 void hfs_btree_close(struct hfs_btree *);
-void hfs_btree_write(struct hfs_btree *);
+int hfs_btree_write(struct hfs_btree *);
 struct hfs_bnode *hfs_bmap_alloc(struct hfs_btree *);

void hfs_bmap_free(struct hfs_bnode *);

fs/hfsplus/super.c +1 -0
index 811a84d..3033b73 100644
@@ -99,6 +99,7 @@ static int
 hfsplus_system_write_inode(struct inode *inode)
     struct hfsplus_vh *vhdr =
 sbi->s_vhdr;
     struct hfsplus_fork_raw *fork;
     struct hfs_btree *tree = NULL;
+    int err;
 

    switch (inode->i_ino) {
    case HFSPLUS_EXT_CNID:

fs/hfsplus/super.c +2 -2
@@ -127,8 +128,15 @@ static int
 hfsplus_system_write_inode(struct inode *inode)
        
 hfsplus_mark_mdb_dirty(inode->i_sb);
     }
     hfsplus_inode_write_fork(inode, fork);
-    if (tree)
-       
 hfs_btree_write(tree);
+    if (tree) {
+        err =

hfs_btree_write(tree);
+        if (err) {
+
printk(KERN_ERR "hfs: unable to write b-tree\n"); +
dprint(DBG_INODE, "hfsplus_system_write_inode: %lu\n",

Quoted 9 lines

+
    inode->i_ino);
+
return err;
+        }
+    }
    return 0;
}

There is a small issue which crossed my mind when I first noticed with:
"hfsplus: rework processing errors in hfsplus_free_extents()"

It is a printk() followed by dprint() with additional information, on error. Because dprint() needs a compile-time change (or at least, echo'ing into debugfs if switched to dynamic-debug). This means the additional information would not be seen the first time the error occurred. I am wondering whether the additional info should be included in the printk - in a short-hand form, maybe?

Hin-Tak

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

author maintainer
#4 lore raw

On Sat, 2012-12-01 at 12:28 +0000, Hin-Tak Leung wrote:

[snip]

Quoted 15 lines
        if (err) {

fs/hfsplus/super.c static int hfsplus_system_write_inode(struct inode *inode) · line 141

Quoted 13 lines

+
printk(KERN_ERR "hfs: unable to write b-tree\n"); +
dprint(DBG_INODE, "hfsplus_system_write_inode: %lu\n",

(quoted diff) +2 -0
+           
    inode->i_ino);
+           

return err;

fs/hfsplus/super.c +2 -0
+        }
+    }
     return 0;
 }
Quoted 5 lines

There is a small issue which crossed my mind when I first noticed with:
"hfsplus: rework processing errors in hfsplus_free_extents()"

It is a printk() followed by dprint() with additional information, on error. Because dprint() needs a compile-time change (or at least, echo'ing into debugfs if switched to dynamic-debug). This means the additional information would not be seen the first time the error occurred. I am wondering whether the additional info should be included in the printk - in a short-hand form, maybe?

I think that your remark is reasonable. I'll prepare the patch.

Thanks,
Vyacheslav Dubeyko.

Quoted 6 lines

Hin-Tak

--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html