[PATCH net] hv_netvsc: Fix race condition between netvsc_probe and netvsc_remove

[PATCH net] hv_netvsc: Fix race condition between netvsc_probe and netvsc_remove

[Souradeep Chakrabarti]

In commit ac5047671758 ("hv_netvsc: Disable NAPI before closing the
VMBus channel"), napi_disable was getting called for all channels,
including all subchannels without confirming if they are enabled or not.

Which caused hv_netvsc getting hung at napi_disable, when netvsc_probe()
and netvsc_remove() are happening simultaneously and netvsc_remove()
calls cancel_work_sync(&nvdev->subchan_work) before netvsc_sc_open()
calls napi_enable for the sub channels. Which causes NAPIF_STATE_SCHED
bit not getting cleared for the subchannels.

Now during netvsc_device_remove(), when napi_disable is called for those
subchannels, napi_disable gets stuck on infinite msleep.

Call trace:
[  654.559417] task:modprobe        state:D stack:    0 pid: 2321 ppid:  1091 flags:0x00004002
[  654.568030] Call Trace:
[  654.571221]  <TASK>
[  654.573790]  __schedule+0x2d6/0x960
[  654.577733]  schedule+0x69/0xf0
[  654.581214]  schedule_timeout+0x87/0x140
[  654.585463]  ? __bpf_trace_tick_stop+0x20/0x20
[  654.590291]  msleep+0x2d/0x40
[  654.593625]  napi_disable+0x2b/0x80
[  654.597437]  netvsc_device_remove+0x8a/0x1f0 [hv_netvsc]
[  654.603935]  rndis_filter_device_remove+0x194/0x1c0 [hv_netvsc]
[  654.611101]  ? do_wait_intr+0xb0/0xb0
[  654.615753]  netvsc_remove+0x7c/0x120 [hv_netvsc]
[  654.621675]  vmbus_remove+0x27/0x40 [hv_vmbus]

Fixes: ac5047671758 ("hv_netvsc: Disable NAPI before closing the VMBus channel")
Signed-off-by: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>
---
 drivers/net/hyperv/netvsc.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
index 1dafa44155d0..a6fcbda64ecc 100644
--- a/drivers/net/hyperv/netvsc.c
+++ b/drivers/net/hyperv/netvsc.c
@@ -708,7 +708,10 @@ void netvsc_device_remove(struct hv_device *device)
 	/* Disable NAPI and disassociate its context from the device. */
 	for (i = 0; i < net_device->num_chn; i++) {
 		/* See also vmbus_reset_channel_cb(). */
-		napi_disable(&net_device->chan_table[i].napi);
+		/* only disable enabled NAPI channel */
+		if (i < ndev->real_num_rx_queues)
+			napi_disable(&net_device->chan_table[i].napi);
+
 		netif_napi_del(&net_device->chan_table[i].napi);
 	}
 
-- 
2.34.1

RE: [PATCH net] hv_netvsc: Fix race condition between netvsc_probe and netvsc_remove

[Haiyang Zhang]

> -----Original Message-----
> From: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>
> Sent: Tuesday, January 30, 2024 5:16 AM
> To: KY Srinivasan <kys@microsoft.com>; Haiyang Zhang
> <haiyangz@microsoft.com>; wei.liu@kernel.org; Dexuan Cui
> <decui@microsoft.com>; davem@davemloft.net; edumazet@google.com;
> kuba@kernel.org; pabeni@redhat.com; Long Li <longli@microsoft.com>;
> yury.norov@gmail.com; leon@kernel.org; cai.huoqing@linux.dev;
> ssengar@linux.microsoft.com; vkuznets@redhat.com; tglx@linutronix.de;
> linux-hyperv@vger.kernel.org; netdev@vger.kernel.org; linux-
> kernel@vger.kernel.org; linux-rdma@vger.kernel.org
> Cc: Souradeep Chakrabarti <schakrabarti@microsoft.com>; Souradeep
> Chakrabarti <schakrabarti@linux.microsoft.com>
> Subject: [PATCH net] hv_netvsc: Fix race condition between netvsc_probe
> and netvsc_remove
> 
> In commit ac5047671758 ("hv_netvsc: Disable NAPI before closing the
> VMBus channel"), napi_disable was getting called for all channels,
> including all subchannels without confirming if they are enabled or not.
> 
> Which caused hv_netvsc getting hung at napi_disable, when netvsc_probe()
> and netvsc_remove() are happening simultaneously and netvsc_remove()
> calls cancel_work_sync(&nvdev->subchan_work) before netvsc_sc_open()
> calls napi_enable for the sub channels. Which causes NAPIF_STATE_SCHED
> bit not getting cleared for the subchannels.
> 
> Now during netvsc_device_remove(), when napi_disable is called for those
> subchannels, napi_disable gets stuck on infinite msleep.
> 
> Call trace:
> [  654.559417] task:modprobe        state:D stack:    0 pid: 2321 ppid:
> 1091 flags:0x00004002
> [  654.568030] Call Trace:
> [  654.571221]  <TASK>
> [  654.573790]  __schedule+0x2d6/0x960
> [  654.577733]  schedule+0x69/0xf0
> [  654.581214]  schedule_timeout+0x87/0x140
> [  654.585463]  ? __bpf_trace_tick_stop+0x20/0x20
> [  654.590291]  msleep+0x2d/0x40
> [  654.593625]  napi_disable+0x2b/0x80
> [  654.597437]  netvsc_device_remove+0x8a/0x1f0 [hv_netvsc]
> [  654.603935]  rndis_filter_device_remove+0x194/0x1c0 [hv_netvsc]
> [  654.611101]  ? do_wait_intr+0xb0/0xb0
> [  654.615753]  netvsc_remove+0x7c/0x120 [hv_netvsc]
> [  654.621675]  vmbus_remove+0x27/0x40 [hv_vmbus]
> 
> Fixes: ac5047671758 ("hv_netvsc: Disable NAPI before closing the VMBus
> channel")
> Signed-off-by: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>

Please add:
Cc: stable@vger.kernel.org

Otherwise, all look good!

Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>

> ---
>  drivers/net/hyperv/netvsc.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c
> index 1dafa44155d0..a6fcbda64ecc 100644
> --- a/drivers/net/hyperv/netvsc.c
> +++ b/drivers/net/hyperv/netvsc.c
> @@ -708,7 +708,10 @@ void netvsc_device_remove(struct hv_device *device)
>  	/* Disable NAPI and disassociate its context from the device. */
>  	for (i = 0; i < net_device->num_chn; i++) {
>  		/* See also vmbus_reset_channel_cb(). */
> -		napi_disable(&net_device->chan_table[i].napi);
> +		/* only disable enabled NAPI channel */
> +		if (i < ndev->real_num_rx_queues)
> +			napi_disable(&net_device->chan_table[i].napi);
> +
>  		netif_napi_del(&net_device->chan_table[i].napi);
>  	}
> 
> --
> 2.34.1

RE: [PATCH net] hv_netvsc: Fix race condition between netvsc_probe and netvsc_remove

[Dexuan Cui]

> From: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>
> Sent: Tuesday, January 30, 2024 2:16 AM
> [...]
> In commit ac5047671758 ("hv_netvsc: Disable NAPI before closing the
> VMBus channel"), napi_disable was getting called for all channels,
> including all subchannels without confirming if they are enabled or not.

s/enabled/created/

> Which caused hv_netvsc getting hung at napi_disable, when
> netvsc_probe()
> and netvsc_remove() are happening simultaneously and netvsc_remove()

Technically, they are not happening simultaneously: netvsc_probe() itself has
finished, but the work item scheduled by it has not started yet.

> calls cancel_work_sync(&nvdev->subchan_work) before netvsc_sc_open()
> calls napi_enable for the sub channels. Which causes NAPIF_STATE_SCHED

Technically, nvdev->subchan_work has not started to run yet, i.e.
netvsc_subchan_work() -> rndis_set_subchannel() has not created the
sub-channels yet, so netvsc_sc_open() can't run.

It would be great if you could briefly explain how the NAPIF_STATE_SCHED bit
is set and cleared, e.g. it's pre-set in rndis_filter_device_add() -> netif_napi_add()
so if the sub-channels are not created, netvsc_sc_open() -> napi_enable() won't
clear the flag and the flag remains set for ever for the sub-channels. 

> bit not getting cleared for the subchannels.
> 
> Now during netvsc_device_remove(), when napi_disable is called for those
> subchannels, napi_disable gets stuck on infinite msleep.

The patch body looks good to me. Please post v2 with an updated changelog.

Reviewed-by: Dexuan Cui <decui@microsoft.com>