Skip to contentSkip to conversation

s390/cio: correct some uevent issues

PATCH In review Cornelia Huck 5 participants 9 messages Mar 27, 2020 – Apr 6, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover s390/cio: correct some uevent issues

For subchannels, we currently delay the initial ADD uevent to a point in time controlled by the driver bound to it (this is to avoid a storm of useless uevents for I/O subchannels that do not have an operational device behind it and will get deregistered again, which are potentially a lot on LPARs.)

If we unbind from the io_subchannel driver and rebind again later, we'll get a duplicate ADD uevent -- not a common workflow, but might happen e.g. when you use a device in the host, then pass it to a guest via vfio-ccw, and then want to use it in the host again. Fixed by the first patch.

The vfio_ccw subchannel driver did not generate any ADD uevent at all -- currently not a problem, as every I/O subchannel will have been bound to the io_subchannel driver before, but let's fix this anyway (second patch).

[As an aside, setting driver_override via a udev rule does not work as expected, due to the uevent delaying -- a specified driver_override works as designed, but userspace won't get the ADD uevent until after the io_subchannel driver has been bound to the device already... we may want to rethink this whole uevent mechanism for subchannels later, but I don't think it's too pressing an issue.]

Probably easiest for both patches to go via the s390 arch maintainers.

Cornelia Huck (2):
s390/cio: avoid duplicated 'ADD' uevents s390/cio: generate delayed uevent for vfio-ccw subchannels

 drivers/s390/cio/device.c       | 13 +++++++++----
 drivers/s390/cio/vfio_ccw_drv.c |  5 +++++
 2 files changed, 14 insertions(+), 4 deletions(-)
Signature · 1 line
2.21.1
#1 lore raw

PATCH 1/2 s390/cio: avoid duplicated 'ADD' uevents

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The io_subchannel driver will do so when the associated ccw_device has been registered -- but unconditionally, so more ADD uevents will be generated if a subchannel has been unbound from the io_subchannel driver and later rebound.

To fix this, only generate the ADD event if uevents were still suppressed for the device.

Fixes fa1a8c23eb7d ("s390: cio: Delay uevents for subchannels")
---
 drivers/s390/cio/device.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
drivers/s390/cio/device.c +9 -4
index 0c6245fc7706..983f9c9e08de 100644
@@ -849,8 +849,10 @@ static void io_subchannel_register(struct ccw_device *cdev)
 	 * Now we know this subchannel will stay, we can throw
 	 * our delayed uevent.
 	 */
-	dev_set_uevent_suppress(&sch->dev, 0);
-	kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
 	/* make it known to the system */
 	ret = ccw_device_add(cdev);
 	if (ret) {
@@ -1058,8 +1060,11 @@ static int io_subchannel_probe(struct subchannel *sch)
 		 * Throw the delayed uevent for the subchannel, register
 		 * the ccw_device and exit.
 		 */
-		dev_set_uevent_suppress(&sch->dev, 0);
-		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		if (dev_get_uevent_suppress(&sch->dev)) {
+			/* should always be the case for the console */
+			dev_set_uevent_suppress(&sch->dev, 0);
+			kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		}
 		cdev = sch_get_cdev(sch);
 		rc = ccw_device_add(cdev);
 		if (rc) {
Signature · 1 line
2.21.1
maintainer Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] s390/cio: avoid duplicated 'ADD' uevents

Quoted 7 lines · Cornelia Huck

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The io_subchannel driver will do so when the associated ccw_device has been registered -- but unconditionally, so more ADD uevents will be generated if a subchannel has been unbound from the io_subchannel driver and later rebound.

Quoted 6 lines

To fix this, only generate the ADD event if uevents were still suppressed for the device.

Fixes fa1a8c23eb7d ("s390: cio: Delay uevents for subchannels")

Please also add Vineeth on cc of future patches against core CIO code.

This patch looks sane and makes handling of suppressed UDEV events consistent to what is already implemented for CHSC and EADM subchannel types.

Quoted 36 lines
---
 drivers/s390/cio/device.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)
drivers/s390/cio/device.c +9 -4
index 0c6245fc7706..983f9c9e08de 100644
@@ -849,8 +849,10 @@ static void io_subchannel_register(struct ccw_device *cdev)
 	 * Now we know this subchannel will stay, we can throw
 	 * our delayed uevent.
 	 */
-	dev_set_uevent_suppress(&sch->dev, 0);
-	kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
 	/* make it known to the system */
 	ret = ccw_device_add(cdev);
 	if (ret) {
@@ -1058,8 +1060,11 @@ static int io_subchannel_probe(struct subchannel *sch)
 		 * Throw the delayed uevent for the subchannel, register
 		 * the ccw_device and exit.
 		 */
-		dev_set_uevent_suppress(&sch->dev, 0);
-		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		if (dev_get_uevent_suppress(&sch->dev)) {
+			/* should always be the case for the console */
+			dev_set_uevent_suppress(&sch->dev, 0);
+			kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		}
 		cdev = sch_get_cdev(sch);
 		rc = ccw_device_add(cdev);
 		if (rc) {
Signature · 2 lines
Peter Oberparleiter
Linux on Z Development - IBM Germany
Reviewed-by on v1 1/2
#3 lore raw

Re: [PATCH 1/2] s390/cio: avoid duplicated 'ADD' uevents

In reply to Cornelia Huck

The code changes look good and I was able to test this fix successfully.

Quoted 49 lines · Cornelia Huck

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The io_subchannel driver will do so when the associated ccw_device has been registered -- but unconditionally, so more ADD uevents will be generated if a subchannel has been unbound from the io_subchannel driver and later rebound.

To fix this, only generate the ADD event if uevents were still suppressed for the device.

Fixes fa1a8c23eb7d ("s390: cio: Delay uevents for subchannels")
---
  drivers/s390/cio/device.c | 13 +++++++++----
  1 file changed, 9 insertions(+), 4 deletions(-)
drivers/s390/cio/device.c +9 -4
index 0c6245fc7706..983f9c9e08de 100644
@@ -849,8 +849,10 @@ static void io_subchannel_register(struct ccw_device *cdev)
  	 * Now we know this subchannel will stay, we can throw
  	 * our delayed uevent.
  	 */
-	dev_set_uevent_suppress(&sch->dev, 0);
-	kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
  	/* make it known to the system */
  	ret = ccw_device_add(cdev);
  	if (ret) {
@@ -1058,8 +1060,11 @@ static int io_subchannel_probe(struct subchannel *sch)
  		 * Throw the delayed uevent for the subchannel, register
  		 * the ccw_device and exit.
  		 */
-		dev_set_uevent_suppress(&sch->dev, 0);
-		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		if (dev_get_uevent_suppress(&sch->dev)) {
+			/* should always be the case for the console */
+			dev_set_uevent_suppress(&sch->dev, 0);
+			kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+		}
  		cdev = sch_get_cdev(sch);
  		rc = ccw_device_add(cdev);
  		if (rc) {
Signature · 8 lines
Mit freundlichen Grüßen/Kind regards
   Boris Fiuczynski

IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Gregor Pillen
Geschäftsführung: Dirk Wittkopp
Sitz der Gesellschaft: Böblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294
#4 lore raw

PATCH 2/2 s390/cio: generate delayed uevent for vfio-ccw subchannels

In reply to Cornelia Huck

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The vfio-ccw I/O subchannel driver, however, did not do that, and will not generate an ADD uevent for subchannels that had not been bound to a different driver (or none at all, which also triggers the uevent).

Generate the ADD uevent at the end of the probe function if uevents were still suppressed for the device.

Fixes 63f1934d562d ("vfio: ccw: basic implementation for vfio_ccw driver")
---
 drivers/s390/cio/vfio_ccw_drv.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/s390/cio/vfio_ccw_drv.c +5 -0
index e401a3d0aa57..339a6bc0339b 100644
@@ -167,6 +167,11 @@ static int vfio_ccw_sch_probe(struct subchannel *sch)
 	if (ret)
 		goto out_disable;
 
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
+
 	VFIO_CCW_MSG_EVENT(4, "bound to subchannel %x.%x.%04x\n",
 			   sch->schid.cssid, sch->schid.ssid,
 			   sch->schid.sch_no);
Signature · 1 line
2.21.1
maintainer Reviewed-by on v1 2/2
#5 lore raw

Re: [PATCH 2/2] s390/cio: generate delayed uevent for vfio-ccw subchannels

Quoted 7 lines · Cornelia Huck

The common I/O layer delays the ADD uevent for subchannels and delegates generating this uevent to the individual subchannel drivers. The vfio-ccw I/O subchannel driver, however, did not do that, and will not generate an ADD uevent for subchannels that had not been bound to a different driver (or none at all, which also triggers the uevent).

Quoted 5 lines

Generate the ADD uevent at the end of the probe function if uevents were still suppressed for the device.

Fixes 63f1934d562d ("vfio: ccw: basic implementation for vfio_ccw driver")

This seems to line up with the other drivers.

Quoted 21 lines
---
 drivers/s390/cio/vfio_ccw_drv.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/s390/cio/vfio_ccw_drv.c +5 -0
index e401a3d0aa57..339a6bc0339b 100644
@@ -167,6 +167,11 @@ static int vfio_ccw_sch_probe(struct subchannel *sch)
 	if (ret)
 		goto out_disable;
 
+	if (dev_get_uevent_suppress(&sch->dev)) {
+		dev_set_uevent_suppress(&sch->dev, 0);
+		kobject_uevent(&sch->dev.kobj, KOBJ_ADD);
+	}
+
 	VFIO_CCW_MSG_EVENT(4, "bound to subchannel %x.%x.%04x\n",
 			   sch->schid.cssid, sch->schid.ssid,
 			   sch->schid.sch_no);
#6 lore raw
In reply to Cornelia Huck
Quoted 17 lines · On Fri, 27 Mar 2020 13:45:01 +0100 Cornelia Huck <[email protected]> wrote:

For subchannels, we currently delay the initial ADD uevent to a point in time controlled by the driver bound to it (this is to avoid a storm of useless uevents for I/O subchannels that do not have an operational device behind it and will get deregistered again, which are potentially a lot on LPARs.)

If we unbind from the io_subchannel driver and rebind again later, we'll get a duplicate ADD uevent -- not a common workflow, but might happen e.g. when you use a device in the host, then pass it to a guest via vfio-ccw, and then want to use it in the host again. Fixed by the first patch.

The vfio_ccw subchannel driver did not generate any ADD uevent at all -- currently not a problem, as every I/O subchannel will have been bound to the io_subchannel driver before, but let's fix this anyway (second patch).

Quoted 8 lines

[As an aside, setting driver_override via a udev rule does not work as expected, due to the uevent delaying -- a specified driver_override works as designed, but userspace won't get the ADD uevent until after the io_subchannel driver has been bound to the device already... we may want to rethink this whole uevent mechanism for subchannels later, but I don't think it's too pressing an issue.]

Probably easiest for both patches to go via the s390 arch maintainers.

Friendly ping. Anyone taking these?

Quoted 9 lines

Cornelia Huck (2):
s390/cio: avoid duplicated 'ADD' uevents s390/cio: generate delayed uevent for vfio-ccw subchannels

 drivers/s390/cio/device.c       | 13 +++++++++----
 drivers/s390/cio/vfio_ccw_drv.c |  5 +++++
 2 files changed, 14 insertions(+), 4 deletions(-)
maintainer
#7 lore raw
Quoted 27 lines · Cornelia Huck
Quoted 24 lines · On Fri, 27 Mar 2020 13:45:01 +0100 Cornelia Huck <[email protected]> wrote:

For subchannels, we currently delay the initial ADD uevent to a point in time controlled by the driver bound to it (this is to avoid a storm of useless uevents for I/O subchannels that do not have an operational device behind it and will get deregistered again, which are potentially a lot on LPARs.)

If we unbind from the io_subchannel driver and rebind again later, we'll get a duplicate ADD uevent -- not a common workflow, but might happen e.g. when you use a device in the host, then pass it to a guest via vfio-ccw, and then want to use it in the host again. Fixed by the first patch.

The vfio_ccw subchannel driver did not generate any ADD uevent at all -- currently not a problem, as every I/O subchannel will have been bound to the io_subchannel driver before, but let's fix this anyway (second patch).

[As an aside, setting driver_override via a udev rule does not work as expected, due to the uevent delaying -- a specified driver_override works as designed, but userspace won't get the ADD uevent until after the io_subchannel driver has been bound to the device already... we may want to rethink this whole uevent mechanism for subchannels later, but I don't think it's too pressing an issue.]

Quoted 3 lines
Quoted 1 line

Probably easiest for both patches to go via the s390 arch maintainers.

Friendly ping. Anyone taking these?

I've just applied them, thank you!

#8 lore raw
Quoted 28 lines · On Mon, 6 Apr 2020 13:58:07 +0200 Vasily Gorbik <[email protected]> wrote:
Quoted 27 lines · Cornelia Huck
Quoted 24 lines · On Fri, 27 Mar 2020 13:45:01 +0100 Cornelia Huck <[email protected]> wrote:

For subchannels, we currently delay the initial ADD uevent to a point in time controlled by the driver bound to it (this is to avoid a storm of useless uevents for I/O subchannels that do not have an operational device behind it and will get deregistered again, which are potentially a lot on LPARs.)

If we unbind from the io_subchannel driver and rebind again later, we'll get a duplicate ADD uevent -- not a common workflow, but might happen e.g. when you use a device in the host, then pass it to a guest via vfio-ccw, and then want to use it in the host again. Fixed by the first patch.

The vfio_ccw subchannel driver did not generate any ADD uevent at all -- currently not a problem, as every I/O subchannel will have been bound to the io_subchannel driver before, but let's fix this anyway (second patch).

[As an aside, setting driver_override via a udev rule does not work as expected, due to the uevent delaying -- a specified driver_override works as designed, but userspace won't get the ADD uevent until after the io_subchannel driver has been bound to the device already... we may want to rethink this whole uevent mechanism for subchannels later, but I don't think it's too pressing an issue.]

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

Probably easiest for both patches to go via the s390 arch maintainers.

Friendly ping. Anyone taking these?

I've just applied them, thank you!

Wonderful, thanks!