Re: [PATCH] Fix error handling in add_disk

[Nikanth Karthikesan]

On Thursday 25 June 2009 08:41:54 Andrew Morton wrote:
> On Wed, 24 Jun 2009 12:18:13 +0530 Nikanth Karthikesan <knikanth@xxxxxxx> wrote:
> > On Wednesday 24 June 2009 02:53:57 Andrew Morton wrote:
> > > On Wed, 17 Jun 2009 12:31:10 +0530
> > >
> > > Nikanth Karthikesan <knikanth@xxxxxxx> wrote:
> > > > Fix error handling in add_disk. Also add WARN_ON()'s in case
> > > > of error, which can be removed once the callers handle the error.
> > >
> > > I have a vague ancestral memory that some of the unchecked errors which
> > > you're now checking for actually do happen in practice, and that this
> > > "fix" will end up breaking currently-working setups.
> > >
> > > Or maybe I'm thinking of a similar but different piece of code (maybe
> > > it was the partition code?).
> > >
> > > Still, I think it would be prudent to initially make this patch
> > > continue to ignore the errors.  So add the warnings, but don't change
> > > the response to errors.  Then we can get the change distributed for a
> > > bit of testing and if that all looks good then we can add the control
> > > flow changes later.
> >
> > add_disk and blk_register_region are functions returning void masking the
> > error
>
> ugh, we're bad.
>
> > which this patch changes. So no caller check for it's return value! And
> > hence errors are ignored, and nothing breaks.
>
> It _does_ change behaviour.  add_disk() can now bale out if, for
> example, sysfs_create_link() failed.  As it commonly does, due to
> various screwups.
>

ah, yes. This patch does mofiy behaviour. :(

> > > >  	retval = blk_alloc_devt(&disk->part0, &devt);
> > > >  	if (retval) {
> > > >  		WARN_ON(1);
> > > > -		return;
> > > > +		goto err_out;
> > > >  	}
> > > > ...
> > > > +	if (retval) {
> > > > +		WARN_ON(1);
> > > > +		goto err_free_devt;
> > > > +	}
> > > > ...
> > > > +	if (retval) {
> > > > +		WARN_ON(1);
> > > > +		goto err_free_region;
> > > > +	}
> > > > ...
> > > > -	WARN_ON(retval);
> > > > +	if (retval) {
> > > > +		WARN_ON(1);
> > > > +		goto err_free_queue;
> > > > +	}
> > >
> > > These all can be coded as
> > >
> > > 	if (WARN_ON(retval))
> > > 		goto foo;
> >
> > done.
> >
> > Thanks
> > Nikanth
> >
> >
> > Fix error handling in add_disk. Also add WARN_ON()'s in case
> > of error, which can be removed once the callers handle the error.
>
> The changelog hasn't been updated to reflect this discussion.  There's
> information missing here.
>
>
> Also, why was this patch written?  Have you observed some behaviour
> which this patch improved or corrected?
>

No. I am just trying to cleanup.

> I applaud the effort, but it's obviously incomplete.  Do you intend to
> add further error checking and handling in this area?

I was intending to fix the callers one by one. But it seems to be better to do
it in one go. I would work on it and send complete patches. Till then could we
add few more WARN_ON's and return error code for the case where add_disk fails
completely?

Thanks
Nikanth

Add WARN_ON's in case of errors during add_disk. Also return error code when
add_disk fails completely. As akpm suggested, in case of partial failure, dont
change the current behaviour and return success.

Signed-off-by: Nikanth Karthikesan <knikanth@xxxxxxx>

---

diff --git a/block/genhd.c b/block/genhd.c
index fe7ccc0..077bf63 100644
--- a/block/genhd.c
+++ b/block/genhd.c
@@ -472,11 +472,11 @@ static char *bdevt_str(dev_t devt, char *buf)
  * range must be nonzero
  * The hash chain is sorted on range, so that subranges can override.
  */
-void blk_register_region(dev_t devt, unsigned long range, struct module *module,
+int blk_register_region(dev_t devt, unsigned long range, struct module *module,
 			 struct kobject *(*probe)(dev_t, int *, void *),
 			 int (*lock)(dev_t, void *), void *data)
 {
-	kobj_map(bdev_map, devt, range, module, probe, lock, data);
+	return kobj_map(bdev_map, devt, range, module, probe, lock, data);
 }
 
 EXPORT_SYMBOL(blk_register_region);
@@ -513,7 +513,7 @@ static int exact_lock(dev_t devt, void *data)
  *
  * FIXME: error handling
  */
-void add_disk(struct gendisk *disk)
+int add_disk(struct gendisk *disk)
 {
 	struct backing_dev_info *bdi;
 	dev_t devt;
@@ -529,10 +529,9 @@ void add_disk(struct gendisk *disk)
 	disk->flags |= GENHD_FL_UP;
 
 	retval = blk_alloc_devt(&disk->part0, &devt);
-	if (retval) {
-		WARN_ON(1);
-		return;
-	}
+	if (WARN_ON(retval))
+		return retval;
+
 	disk_to_dev(disk)->devt = devt;
 
 	/* ->major and ->first_minor aren't supposed to be
@@ -541,16 +540,22 @@ void add_disk(struct gendisk *disk)
 	disk->major = MAJOR(devt);
 	disk->first_minor = MINOR(devt);
 
-	blk_register_region(disk_devt(disk), disk->minors, NULL,
+	retval = blk_register_region(disk_devt(disk), disk->minors, NULL,
 			    exact_match, exact_lock, disk);
+	WARN_ON(retval);
+
 	register_disk(disk);
-	blk_register_queue(disk);
+
+	retval = blk_register_queue(disk);
+	WARN_ON(retval);
 
 	bdi = &disk->queue->backing_dev_info;
 	bdi_register_dev(bdi, disk_devt(disk));
 	retval = sysfs_create_link(&disk_to_dev(disk)->kobj, &bdi->dev->kobj,
 				   "bdi");
 	WARN_ON(retval);
+
+	return 0;
 }
 
 EXPORT_SYMBOL(add_disk);
diff --git a/include/linux/genhd.h b/include/linux/genhd.h
index 149fda2..df3344a 100644
--- a/include/linux/genhd.h
+++ b/include/linux/genhd.h
@@ -339,7 +339,7 @@ static inline void part_dec_in_flight(struct hd_struct *part)
 extern void part_round_stats(int cpu, struct hd_struct *part);
 
 /* block/genhd.c */
-extern void add_disk(struct gendisk *disk);
+extern int add_disk(struct gendisk *disk);
 extern void del_gendisk(struct gendisk *gp);
 extern void unlink_gendisk(struct gendisk *gp);
 extern struct gendisk *get_gendisk(dev_t dev, int *partno);
@@ -534,7 +534,7 @@ extern struct gendisk *alloc_disk_node(int minors, int node_id);
 extern struct gendisk *alloc_disk(int minors);
 extern struct kobject *get_disk(struct gendisk *disk);
 extern void put_disk(struct gendisk *disk);
-extern void blk_register_region(dev_t devt, unsigned long range,
+extern int blk_register_region(dev_t devt, unsigned long range,
 			struct module *module,
 			struct kobject *(*probe)(dev_t, int *, void *),
 			int (*lock)(dev_t, void *),

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/