2018-10-29 01:01:41

by Singh, Balbir

Subject: [PATCH] Add cmd_flags to print_req_error

I ran into a bug where after hibernation due to incompatible
backends, the block driver returned BLK_STS_NOTSUPP, with the
current message it's hard to find out what the command flags
were. Adding req->cmd_flags help make the problem easier to
diagnose.

Signed-off-by: Balbir Singh <[email protected]>
Reviewed-by: Eduardo Valentin <[email protected]>
---

The patch has been just compile tested only.

block/blk-core.c | 9 +++++----
1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/block/blk-core.c b/block/blk-core.c
index cdfabc5646da..28b8f93835a7 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -256,10 +256,11 @@ static void print_req_error(struct request *req, blk_status_t status)
if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
return;

- printk_ratelimited(KERN_ERR "%s: %s error, dev %s, sector %llu\n",
- __func__, blk_errors[idx].name, req->rq_disk ?
- req->rq_disk->disk_name : "?",
- (unsigned long long)blk_rq_pos(req));
+ printk_ratelimited(KERN_ERR "%s: %s error, dev %s, sector %llu flags %llx\n",
+ __func__, blk_errors[idx].name,
+ req->rq_disk ? req->rq_disk->disk_name : "?",
+ (unsigned long long)blk_rq_pos(req),
+ (unsigned long long)req->cmd_flags);
}

static void req_bio_endio(struct request *rq, struct bio *bio,
--
2.16.2

2018-10-29 22:40:03

by Jens Axboe

Subject: Re: [PATCH] Add cmd_flags to print_req_error

On 10/28/18 6:57 PM, Balbir Singh wrote:
> I ran into a bug where after hibernation due to incompatible
> backends, the block driver returned BLK_STS_NOTSUPP, with the
> current message it's hard to find out what the command flags
> were. Adding req->cmd_flags help make the problem easier to
> diagnose.

I recently did the same thing to debug something, so I like
this change. One comment:

> + printk_ratelimited(KERN_ERR "%s: %s error, dev %s, sector %llu flags %llx\n",
> + __func__, blk_errors[idx].name,
> + req->rq_disk ? req->rq_disk->disk_name : "?",
> + (unsigned long long)blk_rq_pos(req),
> + (unsigned long long)req->cmd_flags);

Why are you casting it to unsigned long long for printing? A simple
%x should suffice, no cast necessary.

--
Jens Axboe

2018-10-30 02:54:28

by Balbir Singh

Subject: Re: [PATCH] Add cmd_flags to print_req_error

On Mon, Oct 29, 2018 at 04:39:05PM -0600, Jens Axboe wrote:
> On 10/28/18 6:57 PM, Balbir Singh wrote:
> > I ran into a bug where after hibernation due to incompatible
> > backends, the block driver returned BLK_STS_NOTSUPP, with the
> > current message it's hard to find out what the command flags
> > were. Adding req->cmd_flags help make the problem easier to
> > diagnose.
>
> I recently did the same thing to debug something, so I like
> this change. One comment:
>
> > + printk_ratelimited(KERN_ERR "%s: %s error, dev %s, sector %llu flags %llx\n",
> > + __func__, blk_errors[idx].name,
> > + req->rq_disk ? req->rq_disk->disk_name : "?",
> > + (unsigned long long)blk_rq_pos(req),
> > + (unsigned long long)req->cmd_flags);
>
> Why are you casting it to unsigned long long for printing? A simple
> %x should suffice, no cast necessary.
>

Thanks! resent

I was looking at blk_dump_rq_flags() and tried to do something similar.
I agree a %x is sufficient, I was trying to be consistent. We can fix
that path up if needed.

> --
> Jens Axboe
>
>