diff mbox series

netfs, ceph: Partially revert "netfs: Replace PG_fscache by setting folio->private and marking dirty"

Message ID 2181767.1723665003@warthog.procyon.org.uk (mailing list archive)
State New, archived
Headers show
Series netfs, ceph: Partially revert "netfs: Replace PG_fscache by setting folio->private and marking dirty" | expand

Commit Message

David Howells Aug. 14, 2024, 7:50 p.m. UTC
This partially reverts commit 2ff1e97587f4d398686f52c07afde3faf3da4e5c.

In addition to reverting the removal of PG_private_2 wrangling from the
buffered read code[1][2], the removal of the waits for PG_private_2 from
netfs_release_folio() and netfs_invalidate_folio() need reverting too.

It also adds a wait into ceph_evict_inode() to wait for netfs read and
copy-to-cache ops to complete.

Fixes: 2ff1e97587f4 ("netfs: Replace PG_fscache by setting folio->private and marking dirty")
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Max Kellermann <max.kellermann@ionos.com>
cc: Ilya Dryomov <idryomov@gmail.com>
cc: Xiubo Li <xiubli@redhat.com>
cc: Jeff Layton <jlayton@kernel.org>
cc: Matthew Wilcox <willy@infradead.org>
cc: ceph-devel@vger.kernel.org
cc: netfs@lists.linux.dev
cc: linux-fsdevel@vger.kernel.org
cc: linux-mm@kvack.org
Link: https://lore.kernel.org/r/3575457.1722355300@warthog.procyon.org.uk [1]
Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8e5ced7804cb9184c4a23f8054551240562a8eda [2]
---
 fs/ceph/inode.c |    1 +
 fs/netfs/misc.c |    7 +++++++
 2 files changed, 8 insertions(+)

Comments

Christian Brauner Aug. 15, 2024, 12:51 p.m. UTC | #1
On Wed, Aug 14, 2024 at 08:50:03PM GMT, David Howells wrote:
>     
> This partially reverts commit 2ff1e97587f4d398686f52c07afde3faf3da4e5c.
> 
> In addition to reverting the removal of PG_private_2 wrangling from the
> buffered read code[1][2], the removal of the waits for PG_private_2 from
> netfs_release_folio() and netfs_invalidate_folio() need reverting too.
> 
> It also adds a wait into ceph_evict_inode() to wait for netfs read and
> copy-to-cache ops to complete.
> 
> Fixes: 2ff1e97587f4 ("netfs: Replace PG_fscache by setting folio->private and marking dirty")
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Max Kellermann <max.kellermann@ionos.com>
> cc: Ilya Dryomov <idryomov@gmail.com>
> cc: Xiubo Li <xiubli@redhat.com>
> cc: Jeff Layton <jlayton@kernel.org>
> cc: Matthew Wilcox <willy@infradead.org>
> cc: ceph-devel@vger.kernel.org
> cc: netfs@lists.linux.dev
> cc: linux-fsdevel@vger.kernel.org
> cc: linux-mm@kvack.org
> Link: https://lore.kernel.org/r/3575457.1722355300@warthog.procyon.org.uk [1]
> Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8e5ced7804cb9184c4a23f8054551240562a8eda [2]
> ---

Now in vfs.fixes.

@Konstantin, for some reason I keep having issues with this patch
series. It confuses b4 to no end. When I try to shazam this single patch
here using 2181767.1723665003@warthog.procyon.org.uk it pull down a 26
patch series. And with -P _ it complains that it can't find the messag
id.
Konstantin Ryabitsev Aug. 15, 2024, 2:04 p.m. UTC | #2
On Thu, Aug 15, 2024 at 02:51:58PM GMT, Christian Brauner wrote:
> @Konstantin, for some reason I keep having issues with this patch
> series. It confuses b4 to no end. When I try to shazam this single patch
> here using 2181767.1723665003@warthog.procyon.org.uk it pull down a 26
> patch series. And with -P _ it complains that it can't find the messag
> id.

I cannot reproduce this, which suggests to me that there's something going on
with cached data specific to your local content. Can you please run the
following:

    b4 -d shazam 2181767.1723665003@warthog.procyon.org.uk > /tmp/b4-debug.out 2>&1

then gzip and send that to me?

-K
Xiubo Li Aug. 19, 2024, 1:31 a.m. UTC | #3
On 8/15/24 03:50, David Howells wrote:
>      
> This partially reverts commit 2ff1e97587f4d398686f52c07afde3faf3da4e5c.
>
> In addition to reverting the removal of PG_private_2 wrangling from the
> buffered read code[1][2], the removal of the waits for PG_private_2 from
> netfs_release_folio() and netfs_invalidate_folio() need reverting too.
>
> It also adds a wait into ceph_evict_inode() to wait for netfs read and
> copy-to-cache ops to complete.
>
> Fixes: 2ff1e97587f4 ("netfs: Replace PG_fscache by setting folio->private and marking dirty")
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Max Kellermann <max.kellermann@ionos.com>
> cc: Ilya Dryomov <idryomov@gmail.com>
> cc: Xiubo Li <xiubli@redhat.com>
> cc: Jeff Layton <jlayton@kernel.org>
> cc: Matthew Wilcox <willy@infradead.org>
> cc: ceph-devel@vger.kernel.org
> cc: netfs@lists.linux.dev
> cc: linux-fsdevel@vger.kernel.org
> cc: linux-mm@kvack.org
> Link: https://lore.kernel.org/r/3575457.1722355300@warthog.procyon.org.uk [1]
> Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8e5ced7804cb9184c4a23f8054551240562a8eda [2]
> ---
>   fs/ceph/inode.c |    1 +
>   fs/netfs/misc.c |    7 +++++++
>   2 files changed, 8 insertions(+)
>
> diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
> index 71cd70514efa..4a8eec46254b 100644
> --- a/fs/ceph/inode.c
> +++ b/fs/ceph/inode.c
> @@ -695,6 +695,7 @@ void ceph_evict_inode(struct inode *inode)
>   
>   	percpu_counter_dec(&mdsc->metric.total_inodes);
>   
> +	netfs_wait_for_outstanding_io(inode);
>   	truncate_inode_pages_final(&inode->i_data);
>   	if (inode->i_state & I_PINNING_NETFS_WB)
>   		ceph_fscache_unuse_cookie(inode, true);
> diff --git a/fs/netfs/misc.c b/fs/netfs/misc.c
> index 83e644bd518f..554a1a4615ad 100644
> --- a/fs/netfs/misc.c
> +++ b/fs/netfs/misc.c
> @@ -101,6 +101,8 @@ void netfs_invalidate_folio(struct folio *folio, size_t offset, size_t length)
>   
>   	_enter("{%lx},%zx,%zx", folio->index, offset, length);
>   
> +	folio_wait_private_2(folio); /* [DEPRECATED] */
> +
>   	if (!folio_test_private(folio))
>   		return;
>   
> @@ -165,6 +167,11 @@ bool netfs_release_folio(struct folio *folio, gfp_t gfp)
>   
>   	if (folio_test_private(folio))
>   		return false;
> +	if (unlikely(folio_test_private_2(folio))) { /* [DEPRECATED] */
> +		if (current_is_kswapd() || !(gfp & __GFP_FS))
> +			return false;
> +		folio_wait_private_2(folio);
> +	}
>   	fscache_note_page_release(netfs_i_cookie(ctx));
>   	return true;
>   }

Just back from PTOs.

This LGTM and I will run the test today locally.

Thanks David.

- Xiubo
Xiubo Li Aug. 20, 2024, 6:33 a.m. UTC | #4
On 8/15/24 03:50, David Howells wrote:
>      
> This partially reverts commit 2ff1e97587f4d398686f52c07afde3faf3da4e5c.
>
> In addition to reverting the removal of PG_private_2 wrangling from the
> buffered read code[1][2], the removal of the waits for PG_private_2 from
> netfs_release_folio() and netfs_invalidate_folio() need reverting too.
>
> It also adds a wait into ceph_evict_inode() to wait for netfs read and
> copy-to-cache ops to complete.
>
> Fixes: 2ff1e97587f4 ("netfs: Replace PG_fscache by setting folio->private and marking dirty")
> Signed-off-by: David Howells <dhowells@redhat.com>
> cc: Max Kellermann <max.kellermann@ionos.com>
> cc: Ilya Dryomov <idryomov@gmail.com>
> cc: Xiubo Li <xiubli@redhat.com>
> cc: Jeff Layton <jlayton@kernel.org>
> cc: Matthew Wilcox <willy@infradead.org>
> cc: ceph-devel@vger.kernel.org
> cc: netfs@lists.linux.dev
> cc: linux-fsdevel@vger.kernel.org
> cc: linux-mm@kvack.org
> Link: https://lore.kernel.org/r/3575457.1722355300@warthog.procyon.org.uk [1]
> Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=8e5ced7804cb9184c4a23f8054551240562a8eda [2]
> ---
>   fs/ceph/inode.c |    1 +
>   fs/netfs/misc.c |    7 +++++++
>   2 files changed, 8 insertions(+)
>
> diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
> index 71cd70514efa..4a8eec46254b 100644
> --- a/fs/ceph/inode.c
> +++ b/fs/ceph/inode.c
> @@ -695,6 +695,7 @@ void ceph_evict_inode(struct inode *inode)
>   
>   	percpu_counter_dec(&mdsc->metric.total_inodes);
>   
> +	netfs_wait_for_outstanding_io(inode);
>   	truncate_inode_pages_final(&inode->i_data);
>   	if (inode->i_state & I_PINNING_NETFS_WB)
>   		ceph_fscache_unuse_cookie(inode, true);
> diff --git a/fs/netfs/misc.c b/fs/netfs/misc.c
> index 83e644bd518f..554a1a4615ad 100644
> --- a/fs/netfs/misc.c
> +++ b/fs/netfs/misc.c
> @@ -101,6 +101,8 @@ void netfs_invalidate_folio(struct folio *folio, size_t offset, size_t length)
>   
>   	_enter("{%lx},%zx,%zx", folio->index, offset, length);
>   
> +	folio_wait_private_2(folio); /* [DEPRECATED] */
> +
>   	if (!folio_test_private(folio))
>   		return;
>   
> @@ -165,6 +167,11 @@ bool netfs_release_folio(struct folio *folio, gfp_t gfp)
>   
>   	if (folio_test_private(folio))
>   		return false;
> +	if (unlikely(folio_test_private_2(folio))) { /* [DEPRECATED] */
> +		if (current_is_kswapd() || !(gfp & __GFP_FS))
> +			return false;
> +		folio_wait_private_2(folio);
> +	}
>   	fscache_note_page_release(netfs_i_cookie(ctx));
>   	return true;
>   }
>
Tested it and worked fine.

Reviewed-by: Xiubo Li <xiubli@redhat.com>
Tested-by: Xiubo Li <xiubli@redhat.com>

Thanks
diff mbox series

Patch

diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 71cd70514efa..4a8eec46254b 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -695,6 +695,7 @@  void ceph_evict_inode(struct inode *inode)
 
 	percpu_counter_dec(&mdsc->metric.total_inodes);
 
+	netfs_wait_for_outstanding_io(inode);
 	truncate_inode_pages_final(&inode->i_data);
 	if (inode->i_state & I_PINNING_NETFS_WB)
 		ceph_fscache_unuse_cookie(inode, true);
diff --git a/fs/netfs/misc.c b/fs/netfs/misc.c
index 83e644bd518f..554a1a4615ad 100644
--- a/fs/netfs/misc.c
+++ b/fs/netfs/misc.c
@@ -101,6 +101,8 @@  void netfs_invalidate_folio(struct folio *folio, size_t offset, size_t length)
 
 	_enter("{%lx},%zx,%zx", folio->index, offset, length);
 
+	folio_wait_private_2(folio); /* [DEPRECATED] */
+
 	if (!folio_test_private(folio))
 		return;
 
@@ -165,6 +167,11 @@  bool netfs_release_folio(struct folio *folio, gfp_t gfp)
 
 	if (folio_test_private(folio))
 		return false;
+	if (unlikely(folio_test_private_2(folio))) { /* [DEPRECATED] */
+		if (current_is_kswapd() || !(gfp & __GFP_FS))
+			return false;
+		folio_wait_private_2(folio);
+	}
 	fscache_note_page_release(netfs_i_cookie(ctx));
 	return true;
 }