Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <5a883e670b6f38fc7c3edde7343a205fbc56474b.camel@linux.ibm.com>
Subject: Re: nvme-pci: NULL pointer dereference in nvme_dev_disable() on
 linux-next
From:   Gerd Bayer <gbayer@linux.ibm.com>
To:     Christoph Hellwig <hch@lst.de>
Cc:     Jens Axboe <axboe@fb.com>, Sagi Grimberg <sagi@grimberg.me>,
        Niklas Schnelle <schnelle@linux.ibm.com>,
        linux-kernel@vger.kernel.org, linux-next@vger.kernel.org,
        linux-nvme@lists.infradead.org
Date:   Tue, 08 Nov 2022 18:11:01 +0100
In-Reply-To: <20221108074846.GA22674@lst.de>
References: <fad4d2d5e24eabe1a4fcab75c5d080a6229dc88b.camel@linux.ibm.com>
         <20221108074846.GA22674@lst.de>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0
Precedence: bulk

Hi Christoph,

with your minimal fix

On Tue, 2022-11-08 at 08:48 +0100, Christoph Hellwig wrote:
> Below is the minimal fix.  I'll see if I sort out the mess that is
> probe/reset failure vs ->remove a bit better, though.
> 
> diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
> index f94b05c585cbc..577bacdcfee08 100644
> --- a/drivers/nvme/host/core.c
> +++ b/drivers/nvme/host/core.c
> @@ -5160,6 +5160,8 @@ EXPORT_SYMBOL_GPL(nvme_start_freeze);
>  
>  void nvme_stop_queues(struct nvme_ctrl *ctrl)
>  {
> +	if (!ctrl->tagset)
> +		return;
>  	if (!test_and_set_bit(NVME_CTRL_STOPPED, &ctrl->flags))
>  		blk_mq_quiesce_tagset(ctrl->tagset);
>  	else
> @@ -5169,6 +5171,8 @@ EXPORT_SYMBOL_GPL(nvme_stop_queues);
>  
>  void nvme_start_queues(struct nvme_ctrl *ctrl)
>  {
> +	if (!ctrl->tagset)
> +		return;
>  	if (test_and_clear_bit(NVME_CTRL_STOPPED, &ctrl->flags))
>  		blk_mq_unquiesce_tagset(ctrl->tagset);
>  }

on next-20221108 the kernel does not crash any
more when I run the short test-script.

dmesg shows:
Nov 08 17:38:51 a46lp24.lnxne.boe kernel: nvme nvme0: pci function 0004:00:00.0
Nov 08 17:38:51 a46lp24.lnxne.boe kernel:
nvme nvme0: failed to mark controller CONNECTING
Nov 08 17:38:51 a46lp24.lnxne.boe kernel: nvme nvme0: Removing after
probe failure status: -16
Nov 08 17:38:52 a46lp24.lnxne.boe kernel: pci 0004:00:00.0: Removing from iommu group 0

while kernel remains up.
I can even do 
- rescan on the pci bus (to bring back the nvme drive), and
- run the test script
multiple times.

So from my point of view this band-aid is valuable to be incorporated while the larger 
overhaul in

https://lore.kernel.org/linux-nvme/20221108150252.2123727-1-hch@lst.de/
is out for review and test.

Feel free to add my
Tested-by: Gerd Bayer <gbayer@linux.ibm.com>

Thank you,
Gerd Bayer