[v9,00/61] XArray v9

Commit Message

Andrew Morton March 26, 2018, 10:36 p.m. UTC

On Tue, 13 Mar 2018 06:25:38 -0700 Matthew Wilcox <[e-mail]> wrote:

> This patchset is, I believe, appropriate for merging for 4.17.
> It contains the XArray implementation, to eventually replace the radix
> tree, and converts the page cache to use it.

I looked at this from a for-4.17 POV and ran out of nerve at "[PATCH v9
09/61] xarray: Replace exceptional entries".  It's awfully late.

"[PATCH v9 08/61] page cache: Use xa_lock" looks sufficiently
mechanical to be if-it-compiles-it-works, although perhaps that
shouldn't be in 4.17 either.  Mainly because it commits us to merging
the rest of XArray and there hasn't been a ton of review and test
activity.



It looks like btrfs has changed in -next:

Comments

Matthew Wilcox March 26, 2018, 11:26 p.m. UTC | #1

On Mon, Mar 26, 2018 at 03:36:48PM -0700, Andrew Morton wrote:
> I looked at this from a for-4.17 POV and ran out of nerve at "[PATCH v9
> 09/61] xarray: Replace exceptional entries".  It's awfully late.

I did post v7 five weeks ago ... it wasn't late at the time.

> "[PATCH v9 08/61] page cache: Use xa_lock" looks sufficiently
> mechanical to be if-it-compiles-it-works, although perhaps that
> shouldn't be in 4.17 either.  Mainly because it commits us to merging
> the rest of XArray and there hasn't been a ton of review and test
> activity.

I think we should commit to that.  The API has had a pretty thorough
review, and nobody's stepped up to say "Hey, no, I prefer the old API,
I don't want to see it change".  Merging patch 8 would move us a good
chunk of the way towards getting the IDA in a position where it can
be converted.  Patch 9 would get us even further, but I'm willing to
respin in order to build on just patch 8.

> It looks like btrfs has changed in -next:
> 
> --- a/fs/btrfs/inode.c~page-cache-use-xa_lock-fix
> +++ a/fs/btrfs/inode.c
> @@ -7445,7 +7445,7 @@ out:
>  
>  bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end)
>  {
> -	struct radix_tree_root *root = &inode->i_mapping->page_tree;
> +	struct radix_tree_root *root = &inode->i_mapping->i_pages;
>  	bool found = false;
>  	void **pagep = NULL;
>  	struct page *page = NULL;

btrfs_page_exists_in_range() has been deleted -- David Sterba merged the
patch v8-0006-btrfs-Use-filemap_range_has_page.patch ... which was dropped
from v9 of the patchset, so I'm not sure what you're actually looking at?

Patch

--- a/fs/btrfs/inode.c~page-cache-use-xa_lock-fix
+++ a/fs/btrfs/inode.c
@@ -7445,7 +7445,7 @@  out:
 
 bool btrfs_page_exists_in_range(struct inode *inode, loff_t start, loff_t end)
 {
-	struct radix_tree_root *root = &inode->i_mapping->page_tree;
+	struct radix_tree_root *root = &inode->i_mapping->i_pages;
 	bool found = false;
 	void **pagep = NULL;
 	struct page *page = NULL;