[PATCH] nvme: fixup boot failure on nvme-pci

[PATCH] nvme: fixup boot failure on nvme-pci

[hare]

From: Hannes Reinecke <hare@kernel.org>

Commit 62baf70c3274 caused the ANA log page to be re-read, even on systems
where the ANA is not supported.

Fixes: 62baf70c3274 ("nvme: re-read ANA log page after ns scan completes")

Signed-off-by: Hannes Reinecke <hare@kernel.org>
Tested-by: Srikanth Aithal <sraithal@amd.com>
---
 drivers/nvme/host/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index b502ac07483b..eb6ea8acb3cc 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -4300,7 +4300,7 @@ static void nvme_scan_work(struct work_struct *work)
 	if (test_bit(NVME_AER_NOTICE_NS_CHANGED, &ctrl->events))
 		nvme_queue_scan(ctrl);
 #ifdef CONFIG_NVME_MULTIPATH
-	else
+	else if (ctrl->ana_log_buf)
 		/* Re-read the ANA log page to not miss updates */
 		queue_work(nvme_wq, &ctrl->ana_work);
 #endif
-- 
2.35.3

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Christoph Hellwig]

On Mon, Apr 14, 2025 at 02:05:09PM +0200, hare@kernel.org wrote:
> From: Hannes Reinecke <hare@kernel.org>
> 
> Commit 62baf70c3274 caused the ANA log page to be re-read, even on systems
> where the ANA is not supported.

And unsupported log page should normally not cause a boot failure, but
it seems the controller in question does not handle it well.  I've
applied the patch with a better subjet and commit message explaining this.

But if the controller handles unsupported log pages so badly it will
probably cause trouble in the future as well, or even now when
applications ask for unsupported log pages using the passthrough
interfaces.

Srikanth: what controller is this?  I'd like to add that to the commit
message as well.

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Aithal, Srikanth]

On 4/14/2025 5:56 PM, Christoph Hellwig wrote:
> On Mon, Apr 14, 2025 at 02:05:09PM +0200, hare@kernel.org wrote:
>> From: Hannes Reinecke <hare@kernel.org>
>>
>> Commit 62baf70c3274 caused the ANA log page to be re-read, even on systems
>> where the ANA is not supported.
> And unsupported log page should normally not cause a boot failure, but
> it seems the controller in question does not handle it well.  I've
> applied the patch with a better subjet and commit message explaining this.
>
> But if the controller handles unsupported log pages so badly it will
> probably cause trouble in the future as well, or even now when
> applications ask for unsupported log pages using the passthrough
> interfaces.
>
> Srikanth: what controller is this?  I'd like to add that to the commit
> message as well.

It is a Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe 
SSD Controller SM981/PM981/PM983


>

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Domenico Andreoli]

On Mon, Apr 14, 2025 at 02:26:16PM +0200, Christoph Hellwig wrote:
> On Mon, Apr 14, 2025 at 02:05:09PM +0200, hare@kernel.org wrote:
> > From: Hannes Reinecke <hare@kernel.org>
> > 
> > Commit 62baf70c3274 caused the ANA log page to be re-read, even on systems
> > where the ANA is not supported.
> 
> And unsupported log page should normally not cause a boot failure, but
> it seems the controller in question does not handle it well.  I've
> applied the patch with a better subjet and commit message explaining this.

These are the messages that hang my FriendlyELEC NanoPI M4 SBC at boot:

[    4.342362] nvme nvme0: 6/0/0 default/read/poll queues
[    4.359986]  nvme0n1: p2 p3 p8 p9
[   35.830402] nvme nvme0: controller is down; will reset: CSTS=0x3, PCI_STATUS=0x1010
[   35.879571] nvme0n1: I/O Cmd(0x2) @ LBA 1000215040, 8 blocks, I/O Error (sct 0x3 / sc 0x71)
[   35.880346] I/O error, dev nvme0n1, sector 1000215040 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
[   35.881320] nvme nvme0: Failed to get ANA log: -4
[   35.926157] nvme nvme0: D3 entry latency set to 8 seconds
[   35.939041] nvme nvme0: 6/0/0 default/read/poll queues
[   66.550428] nvme nvme0: controller is down; will reset: CSTS=0x3, PCI_STATUS=0x1010
[   66.600020] nvme0n1: I/O Cmd(0x2) @ LBA 1000215152, 8 blocks, I/O Error (sct 0x3 / sc 0x71)
[   66.600806] I/O error, dev nvme0n1, sector 1000215152 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
[   66.610493] nvme nvme0: Failed to get ANA log: -4
[   66.654065] nvme nvme0: D3 entry latency set to 8 seconds
[   66.667024] nvme nvme0: 6/0/0 default/read/poll queues
[   97.270420] nvme nvme0: controller is down; will reset: CSTS=0x3, PCI_STATUS=0x1010
[   97.320023] nvme0n1: I/O Cmd(0x2) @ LBA 1000214240, 8 blocks, I/O Error (sct 0x3 / sc 0x71)
[   97.320796] I/O error, dev nvme0n1, sector 1000214240 op 0x0:(READ) flags 0x80700 phys_seg 1 prio class 0
[   97.330486] nvme nvme0: Failed to get ANA log: -4
[   97.378015] nvme nvme0: D3 entry latency set to 8 seconds
[   97.390837] nvme nvme0: 6/0/0 default/read/poll queues

> But if the controller handles unsupported log pages so badly it will
> probably cause trouble in the future as well, or even now when
> applications ask for unsupported log pages using the passthrough
> interfaces.
> 
> Srikanth: what controller is this?  I'd like to add that to the commit
> message as well.

00:00.0 PCI bridge: Rockchip Electronics Co., Ltd RK3399 PCI Express Root Port
01:00.0 Non-Volatile memory controller: Samsung Electronics Co Ltd NVMe SSD Controller SM981/PM981/PM983

With the patch above everything works again.

Thanks,
Dom

-- 
rsa4096: 3B10 0CA1 8674 ACBA B4FE  FCD2 CE5B CF17 9960 DE13
ed25519: FFB4 0CC3 7F2E 091D F7DA  356E CC79 2832 ED38 CB05

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Aithal, Srikanth]

Hello,

When will this fix be included in the Linux-next build? Please let me know.


On 4/14/2025 5:56 PM, Christoph Hellwig wrote:
> On Mon, Apr 14, 2025 at 02:05:09PM +0200, hare@kernel.org wrote:
>> From: Hannes Reinecke <hare@kernel.org>
>>
>> Commit 62baf70c3274 caused the ANA log page to be re-read, even on systems
>> where the ANA is not supported.
> And unsupported log page should normally not cause a boot failure, but
> it seems the controller in question does not handle it well.  I've
> applied the patch with a better subjet and commit message explaining this.
>
> But if the controller handles unsupported log pages so badly it will
> probably cause trouble in the future as well, or even now when
> applications ask for unsupported log pages using the passthrough
> interfaces.
>
> Srikanth: what controller is this?  I'd like to add that to the commit
> message as well.
>

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Keith Busch]

On Wed, Apr 16, 2025 at 12:18:54PM +0530, Aithal, Srikanth wrote:
> Hello,
> 
> When will this fix be included in the Linux-next build? Please let me know.

Pull requests from the linux-nvme tree usually go out on Thursday, so
tomorrow.

Re: [PATCH] nvme: fixup boot failure on nvme-pci

[Sagi Grimberg]

Reviewed-by: Sagi Grimberg <sagi@grimberg.me>