Re: [PATCH v3 1/2] hw/nvme: Implement shadow doorbell buffer support

[Guenter Roeck]

On Thu, Dec 08, 2022 at 09:08:12AM +0100, Klaus Jensen wrote:
> On Dec  8 08:16, Klaus Jensen wrote:
> > On Dec  7 09:49, Guenter Roeck wrote:
> > > Hi,
> > > 
> > > On Thu, Jun 16, 2022 at 08:34:07PM +0800, Jinhao Fan wrote:
> > > > Implement Doorbel Buffer Config command (Section 5.7 in NVMe Spec 1.3)
> > > > and Shadow Doorbel buffer & EventIdx buffer handling logic (Section 7.13
> > > > in NVMe Spec 1.3). For queues created before the Doorbell Buffer Config
> > > > command, the nvme_dbbuf_config function tries to associate each existing
> > > > SQ and CQ with its Shadow Doorbel buffer and EventIdx buffer address.
> > > > Queues created after the Doorbell Buffer Config command will have the
> > > > doorbell buffers associated with them when they are initialized.
> > > > 
> > > > In nvme_process_sq and nvme_post_cqe, proactively check for Shadow
> > > > Doorbell buffer changes instead of wait for doorbell register changes.
> > > > This reduces the number of MMIOs.
> > > > 
> > > > In nvme_process_db(), update the shadow doorbell buffer value with
> > > > the doorbell register value if it is the admin queue. This is a hack
> > > > since hosts like Linux NVMe driver and SPDK do not use shadow
> > > > doorbell buffer for the admin queue. Copying the doorbell register
> > > > value to the shadow doorbell buffer allows us to support these hosts
> > > > as well as spec-compliant hosts that use shadow doorbell buffer for
> > > > the admin queue.
> > > > 
> > > > Signed-off-by: Jinhao Fan <fanjinhao21s@ict.ac.cn>
> > > 
> > > I noticed that I can no longer boot Linux kernels from nvme on riscv64
> > > systems. The problem is seen with qemu v7.1 and qemu v7.2-rc4.
> > > The log shows:
> > > 
> > > [   35.904128] nvme nvme0: I/O 642 (I/O Cmd) QID 1 timeout, aborting
> > > [   35.905000] EXT4-fs (nvme0n1): mounting ext2 file system using the 
> > > ext4 subsystem
> > > [   66.623863] nvme nvme0: I/O 643 (I/O Cmd) QID 1 timeout, aborting
> > > [   97.343989] nvme nvme0: Abort status: 0x0
> > > [   97.344355] nvme nvme0: Abort status: 0x0
> > > [   97.344647] nvme nvme0: I/O 7 QID 0 timeout, reset controller
> > > [   97.350568] nvme nvme0: I/O 644 (I/O Cmd) QID 1 timeout, aborting
> > > 
> > > This is with the mainline Linux kernel (v6.1-rc8).
> > > 
> > > Bisect points to this patch. Reverting this patch and a number of 
> > > associated
> > > patches (to fix conflicts) fixes the problem.
> > > 
> > > 06143d8771 Revert "hw/nvme: Implement shadow doorbell buffer support"
> > > acb4443e3a Revert "hw/nvme: Use ioeventfd to handle doorbell updates"
> > > d5fd309feb Revert "hw/nvme: do not enable ioeventfd by default"
> > > 1ca1e6c47c Revert "hw/nvme: unregister the event notifier handler on the 
> > > main loop"
> > > 2d26abd51e Revert "hw/nvme: skip queue processing if notifier is cleared"
> > > 99d411b5a5 Revert "hw/nvme: reenable cqe batching"
> > > 2293d3ca6c Revert "hw/nvme: Add trace events for shadow doorbell buffer"
> > > 
> > > Qemu command line:
> > > 
> > > qemu-system-riscv64 -M virt -m 512M \
> > >      -kernel arch/riscv/boot/Image -snapshot \
> > >      -device nvme,serial=foo,drive=d0 \
> > >      -drive file=rootfs.ext2,if=none,format=raw,id=d0 \
> > >      -bios default \
> > >      -append "root=/dev/nvme0n1 console=ttyS0,115200 
> > > earlycon=uart8250,mmio,0x10000000,115200" \
> > >      -nographic -monitor none
> > > 
> > > Guenter
> > 
> > Hi Guenter,
> > 
> > Thanks for the bisect.
> > 
> > The shadow doorbell is also an obvious candidate for this regression. I
> > wonder if this could be a kernel bug, since we are not observing this on
> > other architectures. The memory barriers required are super finicky, but
> > in QEMU all the operations are associated with full memory barriers. The
> > barriers are more fine grained in the kernel though.
> > 
> > I will dig into this together with Keith.
> 
> A cq head doorbell mmio is skipped... And it is not the fault of the
> kernel. The kernel is in it's good right to skip the mmio since the cq
> eventidx is not properly updated.
> 
> Adding that and it boots properly on riscv. But I'm perplexed as to why
> this didnt show up on our regularly tested platforms.
> 
> Gonna try to get this in for 7.2!

I see another problem with sparc64.

[    5.261508] could not locate request for tag 0x0
[    5.261711] nvme nvme0: invalid id 0 completed on queue 1

That is seen repeatedly until the request times out. I'll test with
your patch to see if it resolves this problem as well, and will bisect
otherwise.

Guenter