[v2] fs: ocfs2: Adding new return type vm_fault_t

Commit Message

Andrew Morton June 1, 2018, 11:47 p.m. UTC

On Wed, 23 May 2018 21:02:58 +0530 Souptick Joarder <[e-mail]> wrote:

> 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 4.18.
> 
> Fix one checkpatch.pl warning by replacing BUG_ON()
> with WARN_ON()

err, no, I'll revert that part.

It could be that if this assertion triggers then filesystem corruption
would ensue, in which case a BUG_ON() is the appropriate handling. 
Such a change should be submitted separately, please.

Comments

Andrew Morton June 4, 2018, 11:05 p.m. UTC | #1

On Sat, 2 Jun 2018 11:46:17 +0530 Souptick Joarder <[e-mail]> wrote:

> On Sat, Jun 2, 2018 at 5:17 AM, Andrew Morton <[e-mail]> wrote:
> > On Wed, 23 May 2018 21:02:58 +0530 Souptick Joarder <[e-mail]> wrote:
> >
> >> 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 4.18.
> >>
> >> Fix one checkpatch.pl warning by replacing BUG_ON()
> >> with WARN_ON()
> >
> > err, no, I'll revert that part.
> >
> > It could be that if this assertion triggers then filesystem corruption
> > would ensue, in which case a BUG_ON() is the appropriate handling.
> > Such a change should be submitted separately, please.
> >
> 
> In few places checkpatch.pl throwing warning to replace BUG_ON()
> with WARN_ON(). Shall we ignore these warning or it will vary across
> patches ?

Such changes are unrelated to the vm_fault_t migration, so please don't
make them in this context.

It is true that we tend to be too eager to use BUG_ON(), but addressing
such things should be done one-at-a-time, as a separate project.

Patch

--- a/fs/ocfs2/mmap.c~fs-ocfs2-adding-new-return-type-vm_fault_t-fix
+++ a/fs/ocfs2/mmap.c
@@ -120,7 +120,7 @@  static vm_fault_t __ocfs2_page_mkwrite(s
 		goto out;
 	}
 	err = ocfs2_write_end_nolock(mapping, pos, len, len, fsdata);
-	WARN_ON(err != len);
+	BUG_ON(err != len);
 	ret = VM_FAULT_LOCKED;
 out:
 	return ret;