[PATCH v2] fs: btrfs: Change return type to vm_fault_t

[PATCH v2] fs: btrfs: Change return type to vm_fault_t

[Souptick Joarder]

Use new return type vm_fault_t for fault handler. For
now, this is just documenting that the function returns
a VM_FAULT value rather than an errno. Once all instances
are converted, vm_fault_t will become a distinct type.

Ref-> commit 1c8f422059ae ("mm: change return type to vm_fault_t")

vmf_error() is the newly introduce inline function
in 4.17-rc6.

Signed-off-by: Souptick Joarder <jrdr.linux@gmail.com>
---
v2: ret2 is the new temp variable to handle
    temporary return value within btrfs_page_mkwrite

 fs/btrfs/ctree.h |  2 +-
 fs/btrfs/inode.c | 26 ++++++++++++--------------
 2 files changed, 13 insertions(+), 15 deletions(-)

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 0d422c9..1f52d9fd 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -3218,7 +3218,7 @@ int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
 			 size_t size, struct bio *bio,
 			 unsigned long bio_flags);
 void btrfs_set_range_writeback(void *private_data, u64 start, u64 end);
-int btrfs_page_mkwrite(struct vm_fault *vmf);
+vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf);
 int btrfs_readpage(struct file *file, struct page *page);
 void btrfs_evict_inode(struct inode *inode);
 int btrfs_write_inode(struct inode *inode, struct writeback_control *wbc);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 0b86cf1..d9c21e0 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -8873,7 +8873,7 @@ static void btrfs_invalidatepage(struct page *page, unsigned int offset,
  * beyond EOF, then the page is guaranteed safe against truncation until we
  * unlock the page.
  */
-int btrfs_page_mkwrite(struct vm_fault *vmf)
+vm_fault_t btrfs_page_mkwrite(struct vm_fault *vmf)
 {
 	struct page *page = vmf->page;
 	struct inode *inode = file_inode(vmf->vma->vm_file);
@@ -8885,8 +8885,9 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
 	char *kaddr;
 	unsigned long zero_start;
 	loff_t size;
-	int ret;
+	int ret2;
 	int reserved = 0;
+	vm_fault_t ret;
 	u64 reserved_space;
 	u64 page_start;
 	u64 page_end;
@@ -8907,17 +8908,14 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
 	 * end up waiting indefinitely to get a lock on the page currently
 	 * being processed by btrfs_page_mkwrite() function.
 	 */
-	ret = btrfs_delalloc_reserve_space(inode, &data_reserved, page_start,
+	ret2 = btrfs_delalloc_reserve_space(inode, &data_reserved, page_start,
 					   reserved_space);
-	if (!ret) {
-		ret = file_update_time(vmf->vma->vm_file);
+	if (!ret2) {
+		ret2 = file_update_time(vmf->vma->vm_file);
 		reserved = 1;
 	}
-	if (ret) {
-		if (ret == -ENOMEM)
-			ret = VM_FAULT_OOM;
-		else /* -ENOSPC, -EIO, etc */
-			ret = VM_FAULT_SIGBUS;
+	if (ret2) {
+		ret = vmf_error(ret2);
 		if (reserved)
 			goto out;
 		goto out_noreserve;
@@ -8976,15 +8974,15 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
 			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
 			  0, 0, &cached_state);
 
-	ret = btrfs_set_extent_delalloc(inode, page_start, end, 0,
+	ret2 = btrfs_set_extent_delalloc(inode, page_start, end, 0,
 					&cached_state, 0);
-	if (ret) {
+	if (ret2) {
 		unlock_extent_cached(io_tree, page_start, page_end,
 				     &cached_state);
 		ret = VM_FAULT_SIGBUS;
 		goto out_unlock;
 	}
-	ret = 0;
+	ret2 = 0;
 
 	/* page is wholly or partially inside EOF */
 	if (page_start + PAGE_SIZE > size)
@@ -9009,7 +9007,7 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
 	unlock_extent_cached(io_tree, page_start, page_end, &cached_state);
 
 out_unlock:
-	if (!ret) {
+	if (!ret2) {
 		btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, true);
 		sb_end_pagefault(inode->i_sb);
 		extent_changeset_free(data_reserved);
-- 
1.9.1

Re: [PATCH v2] fs: btrfs: Change return type to vm_fault_t

[David Sterba]

On Wed, Jun 06, 2018 at 07:54:44PM +0530, Souptick Joarder wrote:
> @@ -9009,7 +9007,7 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
>  	unlock_extent_cached(io_tree, page_start, page_end, &cached_state);
>  
>  out_unlock:
> -	if (!ret) {
> +	if (!ret2) {
>  		btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, true);
>  		sb_end_pagefault(inode->i_sb);
>  		extent_changeset_free(data_reserved);

9013                 return VM_FAULT_LOCKED;
9014         }
9015         unlock_page(page);
9016 out:
9017         btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, (ret != 0));
9018         btrfs_delalloc_release_space(inode, data_reserved, page_start,
9019                                      reserved_space, (ret != 0));

I've noticed that there's 'ret' used on lines 9017 and 19, comparing to
a raw number. Is this going to be ok once vm_fault_t is it's own type?

There's no corresponding define for 0 among the VM_FAULT_* values, I'd
expect 0 to work interchangeably, similar to the blk_status_t type:

https://elixir.bootlin.com/linux/latest/source/include/linux/blk_types.h#L30

#define	BLK_STS_OK 0
#define BLK_STS_NOTSUPP		((__force blk_status_t)1)
#define BLK_STS_TIMEOUT		((__force blk_status_t)2)
#define BLK_STS_NOSPC		((__force blk_status_t)3)
...

Your patch is otherwise ok, I'm just curious if this is something to
watch for once vmfault type is switched.

Re: [PATCH v2] fs: btrfs: Change return type to vm_fault_t

[Matthew Wilcox]

On Wed, Jun 06, 2018 at 05:53:47PM +0200, David Sterba wrote:
> On Wed, Jun 06, 2018 at 07:54:44PM +0530, Souptick Joarder wrote:
> > @@ -9009,7 +9007,7 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
> >  	unlock_extent_cached(io_tree, page_start, page_end, &cached_state);
> >  
> >  out_unlock:
> > -	if (!ret) {
> > +	if (!ret2) {
> >  		btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, true);
> >  		sb_end_pagefault(inode->i_sb);
> >  		extent_changeset_free(data_reserved);
> 
> 9013                 return VM_FAULT_LOCKED;
> 9014         }
> 9015         unlock_page(page);
> 9016 out:
> 9017         btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, (ret != 0));
> 9018         btrfs_delalloc_release_space(inode, data_reserved, page_start,
> 9019                                      reserved_space, (ret != 0));
> 
> I've noticed that there's 'ret' used on lines 9017 and 19, comparing to
> a raw number. Is this going to be ok once vm_fault_t is it's own type?
> 
> There's no corresponding define for 0 among the VM_FAULT_* values, I'd
> expect 0 to work interchangeably, similar to the blk_status_t type:
> 
> https://elixir.bootlin.com/linux/latest/source/include/linux/blk_types.h#L30
> 
> #define	BLK_STS_OK 0
> #define BLK_STS_NOTSUPP		((__force blk_status_t)1)
> #define BLK_STS_TIMEOUT		((__force blk_status_t)2)
> #define BLK_STS_NOSPC		((__force blk_status_t)3)
> ...
> 
> Your patch is otherwise ok, I'm just curious if this is something to
> watch for once vmfault type is switched.

Yes, sparse treats 0 specially for these kinds of types.  It goes back to
the original use of __bitwise to mean "this is a special-endian type",
but it's also meaningful for types which aren't _numbers_ so much as a
collection of bits.

By the way, do you really think it improves this function to use 'ret' and
'ret2' like this?  It's your code, and you're entitled to adopt whatever
stylistic preferences you like, but I personally find it easier to read
with 'err' being an errno and 'ret' being the vm_fault_t.

Re: [PATCH v2] fs: btrfs: Change return type to vm_fault_t

[David Sterba]

On Wed, Jun 06, 2018 at 10:50:49AM -0700, Matthew Wilcox wrote:
> On Wed, Jun 06, 2018 at 05:53:47PM +0200, David Sterba wrote:
> > On Wed, Jun 06, 2018 at 07:54:44PM +0530, Souptick Joarder wrote:
> > > @@ -9009,7 +9007,7 @@ int btrfs_page_mkwrite(struct vm_fault *vmf)
> > >  	unlock_extent_cached(io_tree, page_start, page_end, &cached_state);
> > >  
> > >  out_unlock:
> > > -	if (!ret) {
> > > +	if (!ret2) {
> > >  		btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, true);
> > >  		sb_end_pagefault(inode->i_sb);
> > >  		extent_changeset_free(data_reserved);
> > 
> > 9013                 return VM_FAULT_LOCKED;
> > 9014         }
> > 9015         unlock_page(page);
> > 9016 out:
> > 9017         btrfs_delalloc_release_extents(BTRFS_I(inode), PAGE_SIZE, (ret != 0));
> > 9018         btrfs_delalloc_release_space(inode, data_reserved, page_start,
> > 9019                                      reserved_space, (ret != 0));
> > 
> > I've noticed that there's 'ret' used on lines 9017 and 19, comparing to
> > a raw number. Is this going to be ok once vm_fault_t is it's own type?
> > 
> > There's no corresponding define for 0 among the VM_FAULT_* values, I'd
> > expect 0 to work interchangeably, similar to the blk_status_t type:
> > 
> > https://elixir.bootlin.com/linux/latest/source/include/linux/blk_types.h#L30
> > 
> > #define	BLK_STS_OK 0
> > #define BLK_STS_NOTSUPP		((__force blk_status_t)1)
> > #define BLK_STS_TIMEOUT		((__force blk_status_t)2)
> > #define BLK_STS_NOSPC		((__force blk_status_t)3)
> > ...
> > 
> > Your patch is otherwise ok, I'm just curious if this is something to
> > watch for once vmfault type is switched.
> 
> Yes, sparse treats 0 specially for these kinds of types.  It goes back to
> the original use of __bitwise to mean "this is a special-endian type",
> but it's also meaningful for types which aren't _numbers_ so much as a
> collection of bits.

Ok, thanks.

> By the way, do you really think it improves this function to use 'ret' and
> 'ret2' like this?  It's your code, and you're entitled to adopt whatever
> stylistic preferences you like, but I personally find it easier to read
> with 'err' being an errno and 'ret' being the vm_fault_t.

The ret/err pattern caused some confusion so we're going to unify that a
bit and use 'ret' for the function scope return.