Received-SPF: pass (google.com: domain of linux-nfs+bounces-3464-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
Date: Tue, 28 May 2024 23:30:28 -0700
From: Christoph Hellwig <hch@infradead.org>
To: "Matthew Wilcox (Oracle)" <willy@infradead.org>
Cc: linux-fsdevel@vger.kernel.org, linux-nfs@vger.kernel.org,
	Trond Myklebust <trond.myklebust@hammerspace.com>,
	Anna Schumaker <Anna.Schumaker@netapp.com>,
	Christoph Hellwig <hch@infradead.org>
Subject: Re: [PATCH] nfs: Fix misuses of folio_shift() and folio_order()
Message-ID: <ZlbLhHB4vucmIRgR@infradead.org>
References: <20240527163616.1135968-1-hch@lst.de>
 <20240528210407.2158964-1-willy@infradead.org>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240528210407.2158964-1-willy@infradead.org>

On Tue, May 28, 2024 at 10:03:15PM +0100, Matthew Wilcox (Oracle) wrote:
> Page cache indices are in units of PAGE_SIZE, not in units of
> the folio size.  Revert the change in nfs_grow_file(), and
> pass the inode to nfs_folio_length() so it can be reimplemented
> in terms of folio_mkwrite_check_truncate() which handles this
> correctly.

I had to apply the incremental patch below to make the change compile.
With that it causes a new xfstests failure in generic/127 that I haven't
looked into yet.  The mm-level bugs I've seen even with baseline
Linus' tree also happened more often than in my previous tests, but
that might just be coincidence.


diff --git a/fs/nfs/nfstrace.h b/fs/nfs/nfstrace.h
index 1e710654af1173..0a5d5fa9513735 100644
--- a/fs/nfs/nfstrace.h
+++ b/fs/nfs/nfstrace.h
@@ -938,7 +938,7 @@ TRACE_EVENT(nfs_sillyrename_unlink,
 
 DECLARE_EVENT_CLASS(nfs_folio_event,
 		TP_PROTO(
-			const struct inode *inode,
+			struct inode *inode,
 			struct folio *folio
 		),
 
@@ -954,14 +954,14 @@ DECLARE_EVENT_CLASS(nfs_folio_event,
 		),
 
 		TP_fast_assign(
-			const struct nfs_inode *nfsi = NFS_I(inode);
+			struct nfs_inode *nfsi = NFS_I(inode);
 
 			__entry->dev = inode->i_sb->s_dev;
 			__entry->fileid = nfsi->fileid;
 			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
 			__entry->version = inode_peek_iversion_raw(inode);
 			__entry->offset = folio_file_pos(folio);
-			__entry->count = nfs_folio_length(folio);
+			__entry->count = nfs_folio_length(folio, inode);
 		),
 
 		TP_printk(
@@ -977,14 +977,14 @@ DECLARE_EVENT_CLASS(nfs_folio_event,
 #define DEFINE_NFS_FOLIO_EVENT(name) \
 	DEFINE_EVENT(nfs_folio_event, name, \
 			TP_PROTO( \
-				const struct inode *inode, \
+				struct inode *inode, \
 				struct folio *folio \
 			), \
 			TP_ARGS(inode, folio))
 
 DECLARE_EVENT_CLASS(nfs_folio_event_done,
 		TP_PROTO(
-			const struct inode *inode,
+			struct inode *inode,
 			struct folio *folio,
 			int ret
 		),
@@ -1002,14 +1002,14 @@ DECLARE_EVENT_CLASS(nfs_folio_event_done,
 		),
 
 		TP_fast_assign(
-			const struct nfs_inode *nfsi = NFS_I(inode);
+			struct nfs_inode *nfsi = NFS_I(inode);
 
 			__entry->dev = inode->i_sb->s_dev;
 			__entry->fileid = nfsi->fileid;
 			__entry->fhandle = nfs_fhandle_hash(&nfsi->fh);
 			__entry->version = inode_peek_iversion_raw(inode);
 			__entry->offset = folio_file_pos(folio);
-			__entry->count = nfs_folio_length(folio);
+			__entry->count = nfs_folio_length(folio, inode);
 			__entry->ret = ret;
 		),
 
@@ -1026,7 +1026,7 @@ DECLARE_EVENT_CLASS(nfs_folio_event_done,
 #define DEFINE_NFS_FOLIO_EVENT_DONE(name) \
 	DEFINE_EVENT(nfs_folio_event_done, name, \
 			TP_PROTO( \
-				const struct inode *inode, \
+				struct inode *inode, \
 				struct folio *folio, \
 				int ret \
 			), \