Thread (7 messages) 7 messages, 3 authors, 2021-10-15

Re: [PATCH v2 1/2] block: make __register_blkdev() return an error

From: Luis Chamberlain <mcgrof@kernel.org>
Date: 2021-10-15 18:34:16
Also in: linux-fsdevel, linux-raid, linux-scsi, lkml

On Tue, Sep 28, 2021 at 09:19:47AM +0900, Tetsuo Handa wrote:
On 2021/09/28 7:03, Luis Chamberlain wrote:
quoted
diff --git a/drivers/block/ataflop.c b/drivers/block/ataflop.c
index 5dc9b3d32415..be0627345b21 100644
--- a/drivers/block/ataflop.c
+++ b/drivers/block/ataflop.c
@@ -1989,24 +1989,34 @@ static int ataflop_alloc_disk(unsigned int drive, unsigned int type)
 
 static DEFINE_MUTEX(ataflop_probe_lock);
 
-static void ataflop_probe(dev_t dev)
+static int ataflop_probe(dev_t dev)
 {
 	int drive = MINOR(dev) & 3;
 	int type  = MINOR(dev) >> 2;
+	int err = 0;
 
 	if (type)
 		type--;
 
-	if (drive >= FD_MAX_UNITS || type >= NUM_DISK_MINORS)
-		return;
+	if (drive >= FD_MAX_UNITS || type >= NUM_DISK_MINORS) {
+		err = -EINVAL;
+		goto out;
+	}
+
 	mutex_lock(&ataflop_probe_lock);
 	if (!unit[drive].disk[type]) {
-		if (ataflop_alloc_disk(drive, type) == 0) {
-			add_disk(unit[drive].disk[type]);
+		err = ataflop_alloc_disk(drive, type);
+		if (err == 0) {
+			err = add_disk(unit[drive].disk[type]);
+			if (err)
+				blk_cleanup_disk(unit[drive].disk[type]);
 			unit[drive].registered[type] = true;
Why setting registered to true despite add_disk() failed?
del_gendisk() without successful add_disk() sounds wrong.
That was a mistake, fixed.
Don't we need to undo ataflop_alloc_disk() because it sets
unit[drive].disk[type] to non-NULL ?
ataflop_alloc_disk() just calls blk_mq_alloc_disk() for its
allocation, and so blk_cleanup_disk() does that for us. Please
let me know if I missed anything.
quoted
diff --git a/drivers/block/brd.c b/drivers/block/brd.c
index c2bf4946f4e3..82a93044de95 100644
--- a/drivers/block/brd.c
+++ b/drivers/block/brd.c
@@ -426,10 +426,11 @@ static int brd_alloc(int i)
 	return err;
 }
 
-static void brd_probe(dev_t dev)
+static int brd_probe(dev_t dev)
 {
 	int i = MINOR(dev) / max_part;
 	struct brd_device *brd;
+	int err = 0;
 
 	mutex_lock(&brd_devices_mutex);
 	list_for_each_entry(brd, &brd_devices, brd_list) {
@@ -437,9 +438,11 @@ static void brd_probe(dev_t dev)
 			goto out_unlock;
 	}
 
-	brd_alloc(i);
+	err = brd_alloc(i);
 out_unlock:
 	mutex_unlock(&brd_devices_mutex);
+
+	return err;
 }
 
 static void brd_del_one(struct brd_device *brd)
https://lkml.kernel.org/r/e205f13d-18ff-a49c-0988-7de6ea5ff823@i-love.sakura.ne.jp
will require this part to be updated.
Indeed, rebased, thanks for the heads up!
quoted
diff --git a/drivers/block/floppy.c b/drivers/block/floppy.c
index 0434f28742e7..95a1c8ef62f7 100644
--- a/drivers/block/floppy.c
+++ b/drivers/block/floppy.c
@@ -4517,21 +4517,27 @@ static int floppy_alloc_disk(unsigned int drive, unsigned int type)
 
 static DEFINE_MUTEX(floppy_probe_lock);
 
-static void floppy_probe(dev_t dev)
+static int floppy_probe(dev_t dev)
 {
 	unsigned int drive = (MINOR(dev) & 3) | ((MINOR(dev) & 0x80) >> 5);
 	unsigned int type = (MINOR(dev) >> 2) & 0x1f;
+	int err = 0;
 
 	if (drive >= N_DRIVE || !floppy_available(drive) ||
 	    type >= ARRAY_SIZE(floppy_type))
-		return;
+		return -EINVAL;
 
 	mutex_lock(&floppy_probe_lock);
 	if (!disks[drive][type]) {
-		if (floppy_alloc_disk(drive, type) == 0)
-			add_disk(disks[drive][type]);
+		if (floppy_alloc_disk(drive, type) == 0) {
+			err = add_disk(disks[drive][type]);
+			if (err)
+				blk_cleanup_disk(disks[drive][type]);
This makes future floppy_probe() no-op once add_disk() failed (or maybe a bad
thing happens somewhere else), for disks[drive][type] was set to non-NULL by
floppy_alloc_disk() but blk_cleanup_disk() does not reset it to NULL.
Thanks!

I think just setting disks[drive][type] = NULL after the
blk_cleanup_disk() fixes that issue.
According to floppy_module_exit() which tries to cleanup it, implementing
undo might be complicated...
I can't see what would be missing from just setting disks[drive][type] = NULL.
Can you clarify?

  Luis
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help