Re: [PATCH] Fix race between removing raid1 device and I/O error handling on underlying device

NeilBrown [off-list ref] writes:

quoted

diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 611b5f7..ffdad74 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c

@@ -2005,7 +2005,7 @@ static void fix_read_error(struct r1conf *conf, int read_disk,
 		if (!success) {
 			/* Cannot read from anywhere - mark it bad */
 			struct md_rdev *rdev = conf->mirrors[read_disk].rdev;
-			if (!rdev_set_badblocks(rdev, sect, s, 0))
+			if (rdev && !rdev_set_badblocks(rdev, sect, s, 0))
 				md_error(mddev, rdev);
 			break;
 		}

Hi Jes,
this was already reported on linux-raid and I said I didn't like that fix
because I think it is wrong that the rdev could be removed while we are
trying to fix a read error on it.
So I've applied this following which is in -next.

NeilBrown

Hi Neil,

Whoops, I have to admit I am somewhat backlogged on linux-raid.
Apologies for the noise.

I wasn't wildly happy with the solution myself, but I figured it was the
simplest solution. 

Your patch looks good to me, so I'll go with that.

Cheers,
Jes