Re: [BUG] RCU hang with io_uring nvme polling

[Ben Carey]

On Fri, Jun 26, 2026 at 12:48 PM Keith Busch <kbusch@xxxxxxxxxx> wrote:
>
> On Fri, Jun 26, 2026 at 10:35:56AM -0600, Jens Axboe wrote:
> > Yes, it's a bad configuration. I bet it's as simple as:
> >
> > https://lore.kernel.org/linux-block/20260617155051.1266079-1-anuj20.g@xxxxxxxxxxx/
>
> Yep, that's definitely the same problem. Thanks, I hadn't seen that
> thread yet.

(Resending this, didn't enable plain-text mode)

Jens, Keith,

I appreciate you both for the quick responses.

After putting that patch into the kernel, the fio job ran, but most of the
I/O's completed after about 2 milliseconds (makes sense, given the offset.)

Part of me wonders if there's a race between the NVMe driver and the hardware
controller, but I can't back this up right now.

Ben Carey