Skip to contentSkip to conversation

scsi: storvsc: Increase the timeouts to storvsc_timeout

PATCH Applied Dexuan Cui 3 participants 3 messages Jun 6, 2025 – Jun 10, 2025

Subsystems (3)
SCSI Hyper-V/Azure core and drivers Microsoft MANA RDMA driver
lore
Share & tools
lore

Dexuan Cui

author maintainer
#0 lore raw

PATCH 1/1 scsi: storvsc: Increase the timeouts to storvsc_timeout

Currently storvsc_timeout is only used in storvsc_sdev_configure(), and 5s and 10s are used elsewhere. It turns out that rarely the 5s is not enough on Azure, so let's use storvsc_timeout everywhere.

In case a timeout happens and storvsc_channel_init() returns an error, close the VMBus channel so that any host-to-guest messages in the channel's ringbuffer, which might come late, can be safely ignored.

Add a "const" to storvsc_timeout.

Cc [email protected]
Signed-off-by Dexuan Cui <[email protected]>
---
 drivers/scsi/storvsc_drv.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/scsi/storvsc_drv.c +6 -4
index 2e6b2412d2c9..d9e59204a9c3 100644
@@ -362,7 +362,7 @@ MODULE_PARM_DESC(ring_avail_percent_lowater,
 /*
  * Timeout in seconds for all devices managed by this driver.
  */
-static int storvsc_timeout = 180;
+static const int storvsc_timeout = 180;
 
 #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS)
 static struct scsi_transport_template *fc_transport_template;

@@ -768,7 +768,7 @@ static void  handle_multichannel_storage(struct hv_device *device, int max_chns)
 		return;
 	}
 
-	t = wait_for_completion_timeout(&request->wait_event, 10*HZ);
+	t = wait_for_completion_timeout(&request->wait_event, storvsc_timeout * HZ);
 	if (t == 0) {
 		dev_err(dev, "Failed to create sub-channel: timed out\n");
 		return;

@@ -833,7 +833,7 @@ static int storvsc_execute_vstor_op(struct hv_device *device,
 	if (ret != 0)
 		return ret;
 
-	t = wait_for_completion_timeout(&request->wait_event, 5*HZ);
+	t = wait_for_completion_timeout(&request->wait_event, storvsc_timeout * HZ);
 	if (t == 0)
 		return -ETIMEDOUT;
 

@@ -1350,6 +1350,8 @@ static int storvsc_connect_to_vsp(struct hv_device *device, u32 ring_size,
 		return ret;
 
 	ret = storvsc_channel_init(device, is_fc);
+	if (ret)
+		vmbus_close(device->channel);
 
 	return ret;
 }

@@ -1668,7 +1670,7 @@ static int storvsc_host_reset_handler(struct scsi_cmnd *scmnd)
 	if (ret != 0)
 		return FAILED;
 
-	t = wait_for_completion_timeout(&request->wait_event, 5*HZ);
+	t = wait_for_completion_timeout(&request->wait_event, storvsc_timeout * HZ);
 	if (t == 0)
 		return TIMEOUT_ERROR;
Signature · 1 line
2.25.1
Reviewed-by on v1 1/1
#1 lore raw
Quoted 6 lines

Subject: [PATCH] scsi: storvsc: Increase the timeouts to storvsc_timeout

Currently storvsc_timeout is only used in storvsc_sdev_configure(), and 5s and 10s are used elsewhere. It turns out that rarely the 5s is not enough on Azure, so let's use storvsc_timeout everywhere.

Quoted 8 lines

In case a timeout happens and storvsc_channel_init() returns an error, close the VMBus channel so that any host-to-guest messages in the channel's ringbuffer, which might come late, can be safely ignored.

Add a "const" to storvsc_timeout.

Cc [email protected]
Signed-off-by Dexuan Cui <[email protected]>
Reviewed-by Long Li <[email protected]>
Quoted 65 lines
---
 drivers/scsi/storvsc_drv.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)
drivers/scsi/storvsc_drv.c index renamed from drivers/scsi/storvsc_drv.c +0 -0

2e6b2412d2c9..d9e59204a9c3 100644

drivers/scsi/storvsc_drv.c +2 -1
@@ -362,7 +362,7 @@ MODULE_PARM_DESC(ring_avail_percent_lowater,
 /*
  * Timeout in seconds for all devices managed by this driver.
  */
-static int storvsc_timeout = 180;
+static const int storvsc_timeout = 180;
 
 #if IS_ENABLED(CONFIG_SCSI_FC_ATTRS)
 static struct scsi_transport_template *fc_transport_template; @@ -768,7
+768,7 @@ static void  handle_multichannel_storage(struct hv_device *device,

int max_chns)

drivers/scsi/storvsc_drv.c +1 -1
 		return;
 	}
 
-	t = wait_for_completion_timeout(&request->wait_event, 10*HZ);
+	t = wait_for_completion_timeout(&request->wait_event,

storvsc_timeout

drivers/scsi/storvsc_drv.c +2 -1
+* HZ);
 	if (t == 0) {
 		dev_err(dev, "Failed to create sub-channel: timed out\n");
 		return;

@@ -833,7 +833,7 @@ static int storvsc_execute_vstor_op(struct hv_device
 *device,
 	if (ret != 0)
 		return ret;
 
-	t = wait_for_completion_timeout(&request->wait_event, 5*HZ);
+	t = wait_for_completion_timeout(&request->wait_event,

storvsc_timeout

drivers/scsi/storvsc_drv.c +3 -0
+* HZ);
 	if (t == 0)
 		return -ETIMEDOUT;

@@ -1350,6 +1350,8 @@ static int storvsc_connect_to_vsp(struct hv_device
 *device, u32 ring_size,
 		return ret;
 
 	ret = storvsc_channel_init(device, is_fc);
+	if (ret)
+		vmbus_close(device->channel);
 
 	return ret;

}

drivers/scsi/storvsc_drv.c +1 -1
@@ -1668,7 +1670,7 @@ static int storvsc_host_reset_handler(struct scsi_cmnd
 *scmnd)
 	if (ret != 0)
 		return FAILED;
 
-	t = wait_for_completion_timeout(&request->wait_event, 5*HZ);
+	t = wait_for_completion_timeout(&request->wait_event,

storvsc_timeout

drivers/scsi/storvsc_drv.c +1 -1
+* HZ);
 	if (t == 0)
 		return TIMEOUT_ERROR;
 
--

2.25.1

Applied
#2 lore raw
In reply to Dexuan Cui
Quoted 9 lines · Dexuan Cui

Currently storvsc_timeout is only used in storvsc_sdev_configure(), and 5s and 10s are used elsewhere. It turns out that rarely the 5s is not enough on Azure, so let's use storvsc_timeout everywhere.

In case a timeout happens and storvsc_channel_init() returns an error, close the VMBus channel so that any host-to-guest messages in the channel's ringbuffer, which might come late, can be safely ignored.

[...]

Applied to 6.16/scsi-fixes, thanks!

[1/1] scsi: storvsc: Increase the timeouts to storvsc_timeout

      https://git.kernel.org/mkp/scsi/c/b2f966568faa
Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering