bus errors on nfs writes in latest git

From: J. Bruce Fields
Date: Sat Apr 04 2009 - 12:15:00 EST


My regular connecthon tests are reporting bus errors on the "write/read
30 MB file" on recent upstream kernels. I'm travelling and haven't had
a chance to do anything beyond kick off a dumb git-bisect, which pointed
the blame at the following commit. Any ideas?

--b.


commit c2ec175c39f62949438354f603f4aa170846aabb
Author: Nick Piggin <npiggin@xxxxxxx>
Date: Tue Mar 31 15:23:21 2009 -0700

mm: page_mkwrite change prototype to match fault

Change the page_mkwrite prototype to take a struct vm_fault, and return
VM_FAULT_xxx flags. There should be no functional change.

This makes it possible to return much more detailed error information to
the VM (and also can provide more information eg. virtual_address to the
driver, which might be important in some special cases).

This is required for a subsequent fix. And will also make it easier to
merge page_mkwrite() with fault() in future.

Signed-off-by: Nick Piggin <npiggin@xxxxxxx>
Cc: Chris Mason <chris.mason@xxxxxxxxxx>
Cc: Trond Myklebust <trond.myklebust@xxxxxxxxxx>
Cc: Miklos Szeredi <miklos@xxxxxxxxxx>
Cc: Steven Whitehouse <swhiteho@xxxxxxxxxx>
Cc: Mark Fasheh <mfasheh@xxxxxxxx>
Cc: Joel Becker <joel.becker@xxxxxxxxxx>
Cc: Artem Bityutskiy <dedekind@xxxxxxxxxxxxx>
Cc: Felix Blyakher <felixb@xxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>

diff --git a/Documentation/filesystems/Locking b/Documentation/filesystems/Locking
index 4e78ce6..76efe5b 100644
--- a/Documentation/filesystems/Locking
+++ b/Documentation/filesystems/Locking
@@ -505,7 +505,7 @@ prototypes:
void (*open)(struct vm_area_struct*);
void (*close)(struct vm_area_struct*);
int (*fault)(struct vm_area_struct*, struct vm_fault *);
- int (*page_mkwrite)(struct vm_area_struct *, struct page *);
+ int (*page_mkwrite)(struct vm_area_struct *, struct vm_fault *);
int (*access)(struct vm_area_struct *, unsigned long, void*, int, int);

locking rules:
diff --git a/drivers/video/fb_defio.c b/drivers/video/fb_defio.c
index 0820265..0a7a667 100644
--- a/drivers/video/fb_defio.c
+++ b/drivers/video/fb_defio.c
@@ -85,8 +85,9 @@ EXPORT_SYMBOL_GPL(fb_deferred_io_fsync);

/* vm_ops->page_mkwrite handler */
static int fb_deferred_io_mkwrite(struct vm_area_struct *vma,
- struct page *page)
+ struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct fb_info *info = vma->vm_private_data;
struct fb_deferred_io *fbdefio = info->fbdefio;
struct page *cur;
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 5e1d4e3..7dd1b6d 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2060,7 +2060,7 @@ int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
unsigned long btrfs_force_ra(struct address_space *mapping,
struct file_ra_state *ra, struct file *file,
pgoff_t offset, pgoff_t last_index);
-int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page);
+int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
int btrfs_readpage(struct file *file, struct page *page);
void btrfs_delete_inode(struct inode *inode);
void btrfs_put_inode(struct inode *inode);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 7d4f948..ec54237 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4292,8 +4292,9 @@ static void btrfs_invalidatepage(struct page *page, unsigned long offset)
* beyond EOF, then the page is guaranteed safe against truncation until we
* unlock the page.
*/
-int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+int btrfs_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct inode *inode = fdentry(vma->vm_file)->d_inode;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
@@ -4362,6 +4363,8 @@ again:
out_unlock:
unlock_page(page);
out:
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
return ret;
}

diff --git a/fs/buffer.c b/fs/buffer.c
index 73abe6d..6d51a3d 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -2313,9 +2313,10 @@ int block_commit_write(struct page *page, unsigned from, unsigned to)
* unlock the page.
*/
int
-block_page_mkwrite(struct vm_area_struct *vma, struct page *page,
+block_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf,
get_block_t get_block)
{
+ struct page *page = vmf->page;
struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
unsigned long end;
loff_t size;
@@ -2340,6 +2341,9 @@ block_page_mkwrite(struct vm_area_struct *vma, struct page *page,
ret = block_commit_write(page, 0, end);

out_unlock:
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
+
unlock_page(page);
return ret;
}
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 6083bb3..990c940 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1098,7 +1098,7 @@ extern int ext4_meta_trans_blocks(struct inode *, int nrblocks, int idxblocks);
extern int ext4_chunk_trans_blocks(struct inode *, int nrblocks);
extern int ext4_block_truncate_page(handle_t *handle,
struct address_space *mapping, loff_t from);
-extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page);
+extern int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
extern qsize_t ext4_get_reserved_space(struct inode *inode);

/* ioctl.c */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 71d3ecd..dd82ff3 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -5146,8 +5146,9 @@ static int ext4_bh_unmapped(handle_t *handle, struct buffer_head *bh)
return !buffer_mapped(bh);
}

-int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
loff_t size;
unsigned long len;
int ret = -EINVAL;
@@ -5199,6 +5200,8 @@ int ext4_page_mkwrite(struct vm_area_struct *vma, struct page *page)
goto out_unlock;
ret = 0;
out_unlock:
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
up_read(&inode->i_alloc_sem);
return ret;
}
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index 821d10f..4e340fe 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1234,8 +1234,9 @@ static void fuse_vma_close(struct vm_area_struct *vma)
* - sync(2)
* - try_to_free_pages() with order > PAGE_ALLOC_COSTLY_ORDER
*/
-static int fuse_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+static int fuse_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
/*
* Don't use page->mapping as it may become NULL from a
* concurrent truncate.
diff --git a/fs/gfs2/ops_file.c b/fs/gfs2/ops_file.c
index 3b9e8de..70b9b85 100644
--- a/fs/gfs2/ops_file.c
+++ b/fs/gfs2/ops_file.c
@@ -337,8 +337,9 @@ static int gfs2_allocate_page_backing(struct page *page)
* blocks allocated on disk to back that page.
*/

-static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+static int gfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
struct gfs2_inode *ip = GFS2_I(inode);
struct gfs2_sbd *sdp = GFS2_SB(inode);
@@ -412,6 +413,8 @@ out_unlock:
gfs2_glock_dq(&gh);
out:
gfs2_holder_uninit(&gh);
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
return ret;
}

diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index 90f292b..cec7939 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -451,8 +451,9 @@ const struct address_space_operations nfs_file_aops = {
.launder_page = nfs_launder_page,
};

-static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct file *filp = vma->vm_file;
struct dentry *dentry = filp->f_path.dentry;
unsigned pagelen;
@@ -483,6 +484,8 @@ static int nfs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
ret = pagelen;
out_unlock:
unlock_page(page);
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
return ret;
}

diff --git a/fs/ocfs2/mmap.c b/fs/ocfs2/mmap.c
index eea1d24..b606496 100644
--- a/fs/ocfs2/mmap.c
+++ b/fs/ocfs2/mmap.c
@@ -154,8 +154,9 @@ out:
return ret;
}

-static int ocfs2_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+static int ocfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
struct buffer_head *di_bh = NULL;
sigset_t blocked, oldset;
@@ -196,7 +197,8 @@ out:
ret2 = ocfs2_vm_op_unblock_sigs(&oldset);
if (ret2 < 0)
mlog_errno(ret2);
-
+ if (ret)
+ ret = VM_FAULT_SIGBUS;
return ret;
}

diff --git a/fs/ubifs/file.c b/fs/ubifs/file.c
index 93b6de5..0ff89fe 100644
--- a/fs/ubifs/file.c
+++ b/fs/ubifs/file.c
@@ -1434,8 +1434,9 @@ static int ubifs_releasepage(struct page *page, gfp_t unused_gfp_flags)
* mmap()d file has taken write protection fault and is being made
* writable. UBIFS must ensure page is budgeted for.
*/
-static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
+static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
+ struct page *page = vmf->page;
struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
struct ubifs_info *c = inode->i_sb->s_fs_info;
struct timespec now = ubifs_current_time(inode);
@@ -1447,7 +1448,7 @@ static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
ubifs_assert(!(inode->i_sb->s_flags & MS_RDONLY));

if (unlikely(c->ro_media))
- return -EROFS;
+ return VM_FAULT_SIGBUS; /* -EROFS */

/*
* We have not locked @page so far so we may budget for changing the
@@ -1480,7 +1481,7 @@ static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
if (err == -ENOSPC)
ubifs_warn("out of space for mmapped file "
"(inode number %lu)", inode->i_ino);
- return err;
+ return VM_FAULT_SIGBUS;
}

lock_page(page);
@@ -1520,6 +1521,8 @@ static int ubifs_vm_page_mkwrite(struct vm_area_struct *vma, struct page *page)
out_unlock:
unlock_page(page);
ubifs_release_budget(c, &req);
+ if (err)
+ err = VM_FAULT_SIGBUS;
return err;
}

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index e14c4e3..f4e2554 100644
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -234,9 +234,9 @@ xfs_file_mmap(
STATIC int
xfs_vm_page_mkwrite(
struct vm_area_struct *vma,
- struct page *page)
+ struct vm_fault *vmf)
{
- return block_page_mkwrite(vma, page, xfs_get_blocks);
+ return block_page_mkwrite(vma, vmf, xfs_get_blocks);
}

const struct file_operations xfs_file_operations = {
diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h
index f19fd90..3d7bcde 100644
--- a/include/linux/buffer_head.h
+++ b/include/linux/buffer_head.h
@@ -216,7 +216,7 @@ int cont_write_begin(struct file *, struct address_space *, loff_t,
get_block_t *, loff_t *);
int generic_cont_expand_simple(struct inode *inode, loff_t size);
int block_commit_write(struct page *page, unsigned from, unsigned to);
-int block_page_mkwrite(struct vm_area_struct *vma, struct page *page,
+int block_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf,
get_block_t get_block);
void block_sync_page(struct page *);
sector_t generic_block_bmap(struct address_space *, sector_t, get_block_t *);
diff --git a/include/linux/mm.h b/include/linux/mm.h
index 2223f8d..aeabe95 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -135,6 +135,7 @@ extern pgprot_t protection_map[16];

#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
+#define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */

/*
* This interface is used by x86 PAT code to identify a pfn mapping that is
@@ -187,7 +188,7 @@ struct vm_operations_struct {

/* notification that a previously read-only page is about to become
* writable, if an error is returned it will cause a SIGBUS */
- int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
+ int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);

/* called by access_process_vm when get_user_pages() fails, typically
* for use by special VMAs that can switch between memory and hardware
diff --git a/mm/memory.c b/mm/memory.c
index 5b4ad5e..cf6873e 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -1945,6 +1945,15 @@ static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
* get_user_pages(.write=1, .force=1).
*/
if (vma->vm_ops && vma->vm_ops->page_mkwrite) {
+ struct vm_fault vmf;
+ int tmp;
+
+ vmf.virtual_address = (void __user *)(address &
+ PAGE_MASK);
+ vmf.pgoff = old_page->index;
+ vmf.flags = FAULT_FLAG_WRITE|FAULT_FLAG_MKWRITE;
+ vmf.page = old_page;
+
/*
* Notify the address space that the page is about to
* become writable so that it can prohibit this or wait
@@ -1956,8 +1965,12 @@ static int do_wp_page(struct mm_struct *mm, struct vm_area_struct *vma,
page_cache_get(old_page);
pte_unmap_unlock(page_table, ptl);

- if (vma->vm_ops->page_mkwrite(vma, old_page) < 0)
+ tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
+ if (unlikely(tmp &
+ (VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
+ ret = tmp;
goto unwritable_page;
+ }

/*
* Since we dropped the lock we need to revalidate
@@ -2106,7 +2119,7 @@ oom:

unwritable_page:
page_cache_release(old_page);
- return VM_FAULT_SIGBUS;
+ return ret;
}

/*
@@ -2648,9 +2661,14 @@ static int __do_fault(struct mm_struct *mm, struct vm_area_struct *vma,
* to become writable
*/
if (vma->vm_ops->page_mkwrite) {
+ int tmp;
+
unlock_page(page);
- if (vma->vm_ops->page_mkwrite(vma, page) < 0) {
- ret = VM_FAULT_SIGBUS;
+ vmf.flags |= FAULT_FLAG_MKWRITE;
+ tmp = vma->vm_ops->page_mkwrite(vma, &vmf);
+ if (unlikely(tmp &
+ (VM_FAULT_ERROR | VM_FAULT_NOPAGE))) {
+ ret = tmp;
anon = 1; /* no anon but release vmf.page */
goto out_unlocked;
}
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/