Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <CAE=gft4irmMAapAj3O0hWr53PnyRUmcX2AJB+p_PqCJHT0rvNg@mail.gmail.com>
 <ca507480-0b06-5e4d-ebe6-464302d3af92@acm.org>
In-Reply-To: <ca507480-0b06-5e4d-ebe6-464302d3af92@acm.org>
From:   Evan Green <evgreen@chromium.org>
Date:   Mon, 6 May 2019 08:31:37 -0700
Message-ID: <CAE=gft58X0o+_=J81F1F8F5_N58mK+Nqs1zrsmtPC8JDj_Z1LA@mail.gmail.com>
Subject: Re: blkdev_get deadlock
To:     Bart Van Assche <bvanassche@acm.org>
Cc:     linux-block <linux-block@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>, Jens Axboe <axboe@kernel.dk>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, May 3, 2019 at 6:15 PM Bart Van Assche <bvanassche@acm.org> wrote:
>
> On 5/3/19 10:47 AM, Evan Green wrote:
> > Hey blockies,
>   ^^^^^^^^^^^^
>
> That's the weirdest greeting I have encountered so far on the
> linux-block mailing list.

Heh, achievement unlocked.

>
> > I'm seeing a hung task in the kernel, and I wanted to share it in case
> > it's a known issue. I'm still trying to wrap my head around the stacks
> > myself. This is our Chrome OS 4.19 kernel, which is admittedly not
> > 100% vanilla mainline master, but we try to keep it pretty close.
> >
> > I can reproduce this reliably within our chrome OS installer, where
> > it's trying to dd from my system disk (NVMe) to a loop device backed
> > by a removable UFS card (4kb sectors) in a USB dongle.
>
> Although this is not the only possible cause such hangs are often caused
> by a block driver or SCSI LLD not completing a request. A list of
> pending requests can be obtained e.g. by running the attached script.

Thanks for the script. I'll try a few different combinations of dd
involving the UFS card to see if I can at least remove the system disk
from the equation. Hopefully the system will still be responsive
enough to run the script if I keep it in the right place and maybe
pre-warm it up. I also might try an older kernel, since if it's a
misbehaving block device as you suggest then all kernel versions
should lock up.

>
> Bart.