[PATCH] afs: Fix mmap

[PATCH] afs: Fix mmap

[David Howells]

Hi Linus,

Could you apply this patch please?  Note the odd email address in the
Reported-by and Tested-by fields.  That's an attempt to note Auristor's
internal testcase ID for their tracking purposes.  Is this a reasonable way
to do it?  It's kind of modelled on what syzbot does.  Would it be
preferable to use some other field or a comment after the Reported-by
instead?

David
---
afs: Fix mmap

Fix afs_add_open_map() to check that the vnode isn't already on the list
when it adds it.  It's possible that afs_drop_open_mmap() decremented the
cb_nr_mmap counter, but hadn't yet got into the locked section to remove
it.

Also vnode->cb_mmap_link should be initialised, so fix that too.

Fixes: 6e0e99d58a65 ("afs: Fix mmap coherency vs 3rd-party changes")
Reported-by: kafs-testing+fedora34_64checkkafs-build-300@auristor.com
Suggested-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: kafs-testing+fedora34_64checkkafs-build-300@auristor.com
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/686465.1639435380@warthog.procyon.org.uk/ # v1
---
 fs/afs/file.c  |    5 +++--
 fs/afs/super.c |    1 +
 2 files changed, 4 insertions(+), 2 deletions(-)


diff --git a/fs/afs/file.c b/fs/afs/file.c
index cb6ad61eec3b..afe4b803f84b 100644
--- a/fs/afs/file.c
+++ b/fs/afs/file.c
@@ -514,8 +514,9 @@ static void afs_add_open_mmap(struct afs_vnode *vnode)
 	if (atomic_inc_return(&vnode->cb_nr_mmap) == 1) {
 		down_write(&vnode->volume->cell->fs_open_mmaps_lock);
 
-		list_add_tail(&vnode->cb_mmap_link,
-			      &vnode->volume->cell->fs_open_mmaps);
+		if (list_empty(&vnode->cb_mmap_link))
+			list_add_tail(&vnode->cb_mmap_link,
+				      &vnode->volume->cell->fs_open_mmaps);
 
 		up_write(&vnode->volume->cell->fs_open_mmaps_lock);
 	}
diff --git a/fs/afs/super.c b/fs/afs/super.c
index d110def8aa8e..34c68724c98b 100644
--- a/fs/afs/super.c
+++ b/fs/afs/super.c
@@ -667,6 +667,7 @@ static void afs_i_init_once(void *_vnode)
 	INIT_LIST_HEAD(&vnode->pending_locks);
 	INIT_LIST_HEAD(&vnode->granted_locks);
 	INIT_DELAYED_WORK(&vnode->lock_work, afs_lock_work);
+	INIT_LIST_HEAD(&vnode->cb_mmap_link);
 	seqlock_init(&vnode->cb_lock);
 }
 

Re: [PATCH] afs: Fix mmap

[Linus Torvalds]

On Tue, Dec 14, 2021 at 1:22 AM David Howells <dhowells@redhat.com> wrote:
>
> Could you apply this patch please?  Note the odd email address in the
> Reported-by and Tested-by fields.  That's an attempt to note Auristor's
> internal testcase ID for their tracking purposes.  Is this a reasonable way
> to do it?  It's kind of modelled on what syzbot does.

Yeah, this looks sane to me.

I don't like bare bug tracking IDs that don't tell anybody else
anything, but the "embedding it in the email" trick means that it now
(a) gives a contact for it so that it contains meaningful information
and (b) also acts a "namespace" for the testcase ID.

Applied.

            Linus

Re: [PATCH] afs: Fix mmap

[David Howells]

I forgot to commit the patch and so lost part of it.  Full patch below.

David
---
commit 23e7d72edeeff56d7bfa47c32c9638385c90eddd
Author: David Howells <dhowells@redhat.com>
Date:   Mon Dec 13 16:26:44 2021 +0000

    afs: Fix mmap
    
    Fix afs_add_open_map() to check that the vnode isn't already on the list
    when it adds it.  It's possible that afs_drop_open_mmap() decremented the
    cb_nr_mmap counter, but hadn't yet got into the locked section to remove
    it.
    
    Also vnode->cb_mmap_link should be initialised, so fix that too.
    
    Fixes: 6e0e99d58a65 ("afs: Fix mmap coherency vs 3rd-party changes")
    Reported-by: kafs-testing+fedora34_64checkkafs-build-300@auristor.com
    Suggested-by: Marc Dionne <marc.dionne@auristor.com>
    Signed-off-by: David Howells <dhowells@redhat.com>
    cc: linux-afs@lists.infradead.org

diff --git a/fs/afs/file.c b/fs/afs/file.c
index cb6ad61eec3b..afe4b803f84b 100644
--- a/fs/afs/file.c
+++ b/fs/afs/file.c
@@ -514,8 +514,9 @@ static void afs_add_open_mmap(struct afs_vnode *vnode)
 	if (atomic_inc_return(&vnode->cb_nr_mmap) == 1) {
 		down_write(&vnode->volume->cell->fs_open_mmaps_lock);
 
-		list_add_tail(&vnode->cb_mmap_link,
-			      &vnode->volume->cell->fs_open_mmaps);
+		if (list_empty(&vnode->cb_mmap_link))
+			list_add_tail(&vnode->cb_mmap_link,
+				      &vnode->volume->cell->fs_open_mmaps);
 
 		up_write(&vnode->volume->cell->fs_open_mmaps_lock);
 	}
diff --git a/fs/afs/super.c b/fs/afs/super.c
index d110def8aa8e..34c68724c98b 100644
--- a/fs/afs/super.c
+++ b/fs/afs/super.c
@@ -667,6 +667,7 @@ static void afs_i_init_once(void *_vnode)
 	INIT_LIST_HEAD(&vnode->pending_locks);
 	INIT_LIST_HEAD(&vnode->granted_locks);
 	INIT_DELAYED_WORK(&vnode->lock_work, afs_lock_work);
+	INIT_LIST_HEAD(&vnode->cb_mmap_link);
 	seqlock_init(&vnode->cb_lock);
 }
 

[PATCH] afs: Fix mmap

[David Howells]

Fix afs_add_open_map() to check that the vnode isn't already on the list
when it adds it.  It's possible that afs_drop_open_mmap() decremented the
cb_nr_mmap counter, but hadn't yet got into the locked section to remove
it.

Also vnode->cb_mmap_link should be initialised, so fix that too.

Fixes: 6e0e99d58a65 ("afs: Fix mmap coherency vs 3rd-party changes")
Reported-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: linux-afs@lists.infradead.org
---

 fs/afs/file.c |    5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/fs/afs/file.c b/fs/afs/file.c
index 572063dad0b3..bcda99dcfdec 100644
--- a/fs/afs/file.c
+++ b/fs/afs/file.c
@@ -539,8 +539,9 @@ static void afs_add_open_mmap(struct afs_vnode *vnode)
 	if (atomic_inc_return(&vnode->cb_nr_mmap) == 1) {
 		down_write(&vnode->volume->cell->fs_open_mmaps_lock);
 
-		list_add_tail(&vnode->cb_mmap_link,
-			      &vnode->volume->cell->fs_open_mmaps);
+		if (list_empty(&vnode->cb_mmap_link))
+			list_add_tail(&vnode->cb_mmap_link,
+				      &vnode->volume->cell->fs_open_mmaps);
 
 		up_write(&vnode->volume->cell->fs_open_mmaps_lock);
 	}