Re: linux-next: Tree for September 7 (scsi/qla2x)

Re: linux-next: Tree for September 7 (scsi/qla2x)

[Randy Dunlap]

On Mon, 7 Sep 2009 21:02:06 +1000 Stephen Rothwell wrote:

> Hi all,
> 
> Changes since 20090904:


when CONFIG_MODULES=n:

drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type

in
	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
	    KOBJ_CHANGE, envp);

---
~Randy
LPC 2009, Sept. 23-25, Portland, Oregon
http://linuxplumbersconf.org/2009/

Re: linux-next: Tree for September 7 (scsi/qla2x)

[Andrew Vasquez]

On Mon, 07 Sep 2009, Randy Dunlap wrote:

> On Mon, 7 Sep 2009 21:02:06 +1000 Stephen Rothwell wrote:
> 
> > Hi all,
> > 
> > Changes since 20090904:
> 
> 
> when CONFIG_MODULES=n:
> 
> drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
> in
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 	    KOBJ_CHANGE, envp);

Argg...  Some history here...  During several unwelcome
hardware/firmware events (ISP system error, mailbox command timeouts,
etc), the qla2xxx driver can store a 'firmware-dump' (essentially a
snapshot of the current state of the ISP firmware).  This snapshot is
then captured via a user-space tool querying a driver sysfs-node
hanging off of a scsi_host's device tree:

	/sys/class/scsi_host/host4/device/fw_dump

The dump is then used by our firmware engineering group to help triage
the issue.

This recent change:

	commit 10a71b40153a19279428053ad9743e15ef414148
	Author: Andrew Vasquez <andrew.vasquez@qlogic.com>
	Date:   Tue Aug 25 11:36:15 2009 -0700

	    [SCSI] qla2xxx: Add firmware-dump kobject uevent notification.

	    Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
	    Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
	    Signed-off-by: James Bottomley <James.Bottomley@suse.de>

attempted to help 'automate' the task of retrieval by signaling udev
to automatically run the 'retrieval' script anytime the driver
captured the firmware-dump.  Here's a snippet of the udev rule:

	# qla2xxx driver
	KERNEL=="qla2xxx", SUBSYSTEM=="module", ACTION=="change", RUN+="qla2xxx_udev.sh"

Any suggestions here on an alternate driver-specific kobject an LLD
can/should use for something like this?  I looked previously at other
callers of kobject_uevent_env(), but didn't really see a simlar
usage-pattern of a driver wanting to signal events to userspace...

Thanks, AV

qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n (was: Re: linux-next: Tree for September 7 (scsi/qla2x))

[Andrew Vasquez]

Randy Dunlap noted:

  when CONFIG_MODULES=n:

	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type

  in
	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
		KOBJ_CHANGE, envp);

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
---

On Tue, 08 Sep 2009, Andrew Vasquez wrote:

> On Mon, 07 Sep 2009, Randy Dunlap wrote:
> 
> > On Mon, 7 Sep 2009 21:02:06 +1000 Stephen Rothwell wrote:
> > 
> > > Hi all,
> > > 
> > > Changes since 20090904:
> > 
> > 
> > when CONFIG_MODULES=n:
> > 
> > drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > 
> > in
> > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > 	    KOBJ_CHANGE, envp);
> 
> Argg...  Some history here...  During several unwelcome
> hardware/firmware events (ISP system error, mailbox command timeouts,
> etc), the qla2xxx driver can store a 'firmware-dump' (essentially a
> snapshot of the current state of the ISP firmware).  This snapshot is
> then captured via a user-space tool querying a driver sysfs-node
> hanging off of a scsi_host's device tree:
> 
> 	/sys/class/scsi_host/host4/device/fw_dump
> 
> The dump is then used by our firmware engineering group to help triage
> the issue.
> 
> This recent change:
> 
> 	commit 10a71b40153a19279428053ad9743e15ef414148
> 	Author: Andrew Vasquez <andrew.vasquez@qlogic.com>
> 	Date:   Tue Aug 25 11:36:15 2009 -0700
> 
> 	    [SCSI] qla2xxx: Add firmware-dump kobject uevent notification.
> 
> 	    Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> 	    Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
> 	    Signed-off-by: James Bottomley <James.Bottomley@suse.de>
> 
> attempted to help 'automate' the task of retrieval by signaling udev
> to automatically run the 'retrieval' script anytime the driver
> captured the firmware-dump.  Here's a snippet of the udev rule:
> 
> 	# qla2xxx driver
> 	KERNEL=="qla2xxx", SUBSYSTEM=="module", ACTION=="change", RUN+="qla2xxx_udev.sh"
> 
> Any suggestions here on an alternate driver-specific kobject an LLD
> can/should use for something like this?  I looked previously at other
> callers of kobject_uevent_env(), but didn't really see a simlar
> usage-pattern of a driver wanting to signal events to userspace...
> 
> Thanks, AV

Ok, So any strong objections to just having the functionality present
when module support is enabled?

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 29396c0..3887adb 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -2671,6 +2671,7 @@ qla2x00_post_uevent_work(struct scsi_qla_host *vha, u32 code)
 static void
 qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
 {
+#ifdef CONFIG_MODULES
 	char event_string[40];
 	char *envp[] = { event_string, NULL };
 
@@ -2685,6 +2686,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
 	}
 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
 	    KOBJ_CHANGE, envp);
+#endif
 }
 
 void

Re: qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n (was: Re: linux-next: Tree for September 7 (scsi/qla2x))

[Randy Dunlap]

On Fri, 11 Sep 2009 10:53:41 -0700 Andrew Vasquez wrote:

> Randy Dunlap noted:
> 
>   when CONFIG_MODULES=n:
> 
> 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
>   in
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 		KOBJ_CHANGE, envp);
> 
> Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>

Acked-by: Randy Dunlap <randy.dunlap@oracle.com>


> Ok, So any strong objections to just having the functionality present
> when module support is enabled?

Fine with me.  Thanks.

> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 29396c0..3887adb 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -2671,6 +2671,7 @@ qla2x00_post_uevent_work(struct scsi_qla_host *vha, u32 code)
>  static void
>  qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  {
> +#ifdef CONFIG_MODULES
>  	char event_string[40];
>  	char *envp[] = { event_string, NULL };
>  
> @@ -2685,6 +2686,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  	}
>  	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
>  	    KOBJ_CHANGE, envp);
> +#endif
>  }
>  
>  void
> 


---
~Randy
LPC 2009, Sept. 23-25, Portland, Oregon
http://linuxplumbersconf.org/2009/

Re: qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n (was: Re: linux-next: Tree for September 7 (scsi/qla2x))

[James Bottomley]

On Fri, 2009-09-11 at 10:53 -0700, Andrew Vasquez wrote:
> Randy Dunlap noted:
> 
>   when CONFIG_MODULES=n:
> 
> 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
>   in
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 		KOBJ_CHANGE, envp);
> 
> Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> ---
> 
> On Tue, 08 Sep 2009, Andrew Vasquez wrote:
> 
> > On Mon, 07 Sep 2009, Randy Dunlap wrote:
> > 
> > > On Mon, 7 Sep 2009 21:02:06 +1000 Stephen Rothwell wrote:
> > > 
> > > > Hi all,
> > > > 
> > > > Changes since 20090904:
> > > 
> > > 
> > > when CONFIG_MODULES=n:
> > > 
> > > drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > 
> > > in
> > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > 	    KOBJ_CHANGE, envp);
> > 
> > Argg...  Some history here...  During several unwelcome
> > hardware/firmware events (ISP system error, mailbox command timeouts,
> > etc), the qla2xxx driver can store a 'firmware-dump' (essentially a
> > snapshot of the current state of the ISP firmware).  This snapshot is
> > then captured via a user-space tool querying a driver sysfs-node
> > hanging off of a scsi_host's device tree:
> > 
> > 	/sys/class/scsi_host/host4/device/fw_dump
> > 
> > The dump is then used by our firmware engineering group to help triage
> > the issue.
> > 
> > This recent change:
> > 
> > 	commit 10a71b40153a19279428053ad9743e15ef414148
> > 	Author: Andrew Vasquez <andrew.vasquez@qlogic.com>
> > 	Date:   Tue Aug 25 11:36:15 2009 -0700
> > 
> > 	    [SCSI] qla2xxx: Add firmware-dump kobject uevent notification.
> > 
> > 	    Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> > 	    Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
> > 	    Signed-off-by: James Bottomley <James.Bottomley@suse.de>
> > 
> > attempted to help 'automate' the task of retrieval by signaling udev
> > to automatically run the 'retrieval' script anytime the driver
> > captured the firmware-dump.  Here's a snippet of the udev rule:
> > 
> > 	# qla2xxx driver
> > 	KERNEL=="qla2xxx", SUBSYSTEM=="module", ACTION=="change", RUN+="qla2xxx_udev.sh"
> > 
> > Any suggestions here on an alternate driver-specific kobject an LLD
> > can/should use for something like this?  I looked previously at other
> > callers of kobject_uevent_env(), but didn't really see a simlar
> > usage-pattern of a driver wanting to signal events to userspace...
> > 
> > Thanks, AV
> 
> Ok, So any strong objections to just having the functionality present
> when module support is enabled?
> 
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 29396c0..3887adb 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -2671,6 +2671,7 @@ qla2x00_post_uevent_work(struct scsi_qla_host *vha, u32 code)
>  static void
>  qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  {
> +#ifdef CONFIG_MODULES
>  	char event_string[40];
>  	char *envp[] = { event_string, NULL };
>  
> @@ -2685,6 +2686,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  	}
>  	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
>  	    KOBJ_CHANGE, envp);
> +#endif

Only emitting events if the thing is compiled as a module doesn't really
look like the right solution.  The first question that springs to mind
is why are you emitting events against the module kobject in the first
place?  Why not emit them against the device kobject (which is always
present)?

James

Re: qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n (was: Re: linux-next: Tree for September 7 (scsi/qla2x))

[Andrew Vasquez]

On Fri, 11 Sep 2009, James Bottomley wrote:

> On Fri, 2009-09-11 at 10:53 -0700, Andrew Vasquez wrote:
> > Randy Dunlap noted:
> > 
> >   when CONFIG_MODULES=n:
> > 
> > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > 
> >   in
> > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > 		KOBJ_CHANGE, envp);
> > 
> > Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> > ---
> > 
> > On Tue, 08 Sep 2009, Andrew Vasquez wrote:
> > 
> > > On Mon, 07 Sep 2009, Randy Dunlap wrote:
> > > 
> > > > On Mon, 7 Sep 2009 21:02:06 +1000 Stephen Rothwell wrote:
> > > > 
> > > > > Hi all,
> > > > > 
> > > > > Changes since 20090904:
> > > > 
> > > > 
> > > > when CONFIG_MODULES=n:
> > > > 
> > > > drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > 
> > > > in
> > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > 	    KOBJ_CHANGE, envp);
> > > 
> > > Argg...  Some history here...  During several unwelcome
> > > hardware/firmware events (ISP system error, mailbox command timeouts,
> > > etc), the qla2xxx driver can store a 'firmware-dump' (essentially a
> > > snapshot of the current state of the ISP firmware).  This snapshot is
> > > then captured via a user-space tool querying a driver sysfs-node
> > > hanging off of a scsi_host's device tree:
> > > 
> > > 	/sys/class/scsi_host/host4/device/fw_dump
> > > 
> > > The dump is then used by our firmware engineering group to help triage
> > > the issue.
> > > 
> > > This recent change:
> > > 
> > > 	commit 10a71b40153a19279428053ad9743e15ef414148
> > > 	Author: Andrew Vasquez <andrew.vasquez@qlogic.com>
> > > 	Date:   Tue Aug 25 11:36:15 2009 -0700
> > > 
> > > 	    [SCSI] qla2xxx: Add firmware-dump kobject uevent notification.
> > > 
> > > 	    Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> > > 	    Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
> > > 	    Signed-off-by: James Bottomley <James.Bottomley@suse.de>
> > > 
> > > attempted to help 'automate' the task of retrieval by signaling udev
> > > to automatically run the 'retrieval' script anytime the driver
> > > captured the firmware-dump.  Here's a snippet of the udev rule:
> > > 
> > > 	# qla2xxx driver
> > > 	KERNEL=="qla2xxx", SUBSYSTEM=="module", ACTION=="change", RUN+="qla2xxx_udev.sh"
> > > 
> > > Any suggestions here on an alternate driver-specific kobject an LLD
> > > can/should use for something like this?  I looked previously at other
> > > callers of kobject_uevent_env(), but didn't really see a simlar
> > > usage-pattern of a driver wanting to signal events to userspace...
> > > 
> > > Thanks, AV
> > 
> > Ok, So any strong objections to just having the functionality present
> > when module support is enabled?
> > 
> > diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> > index 29396c0..3887adb 100644
> > --- a/drivers/scsi/qla2xxx/qla_os.c
> > +++ b/drivers/scsi/qla2xxx/qla_os.c
> > @@ -2671,6 +2671,7 @@ qla2x00_post_uevent_work(struct scsi_qla_host *vha, u32 code)
> >  static void
> >  qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
> >  {
> > +#ifdef CONFIG_MODULES
> >  	char event_string[40];
> >  	char *envp[] = { event_string, NULL };
> >  
> > @@ -2685,6 +2686,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
> >  	}
> >  	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> >  	    KOBJ_CHANGE, envp);
> > +#endif
> 
> Only emitting events if the thing is compiled as a module doesn't really
> look like the right solution.  The first question that springs to mind
> is why are you emitting events against the module kobject in the first
> place?  Why not emit them against the device kobject (which is always
> present)?

The struct device's kobj from pdev->dev?

My udev-foo is pathetic, so how exactly would that get multiplexed at
the udev side?

	KERNEL=="???", SUBSYSTEM=="???", ACTION=="change", RUN+="qla2xxx_udev.sh"

Thanks, AV

Re: qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n (was: Re: linux-next: Tree for September 7 (scsi/qla2x))

[Andrew Vasquez]

On Fri, 11 Sep 2009, Andrew Vasquez wrote:

> The struct device's kobj from pdev->dev?
> 
> My udev-foo is pathetic, so how exactly would that get multiplexed at
> the udev side?
> 
> 	KERNEL=="???", SUBSYSTEM=="???", ACTION=="change", RUN+="qla2xxx_udev.sh"

Ok, udevadm helps a bit...  Let me see if I can convert this uevent
off of:

	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);

to something meaningful:

	UEVENT[1252714722.263731] change
	/devices/pci0000:17/0000:17:08.0/0000:1e:00.0 (pci)
	ACTION=change
	DEVPATH=/devices/pci0000:17/0000:17:08.0/0000:1e:00.0
	SUBSYSTEM=pci
	FW_DUMP=6
	DRIVER=qla2xxx
	PHYSDEVBUS=pci
	PHYSDEVDRIVER=qla2xxx
	PCI_CLASS=C0400
	PCI_ID=1077:2532
	PCI_SUBSYS_ID=1077:015C
	PCI_SLOT_NAME=0000:1e:00.0
	MODALIAS=pci:v00001077d00002532sv00001077sd0000015Cbc0Csc04i00
	SEQNUM=3574

Thanks, AV

[PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Andrew Vasquez]

Randy Dunlap noted:

  when CONFIG_MODULES=n:

	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type

  in

	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
		KOBJ_CHANGE, envp);

Trigger kobject event on the 'struct device' hanging off the pci_dev.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
---

> On Fri, 11 Sep 2009, Andrew Vasquez wrote:
> 
> > The struct device's kobj from pdev->dev?
> > 
> > My udev-foo is pathetic, so how exactly would that get multiplexed at
> > the udev side?
> > 
> > 	KERNEL=="???", SUBSYSTEM=="???", ACTION=="change", RUN+="qla2xxx_udev.sh"
> 
> Ok, udevadm helps a bit...  Let me see if I can convert this uevent
> off of:
> 
> 	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);
> 
> to something meaningful:
> 
> 	UEVENT[1252714722.263731] change
> 	/devices/pci0000:17/0000:17:08.0/0000:1e:00.0 (pci)
> 	ACTION=change
> 	DEVPATH=/devices/pci0000:17/0000:17:08.0/0000:1e:00.0
> 	SUBSYSTEM=pci
> 	FW_DUMP=6
> 	DRIVER=qla2xxx
> 	PHYSDEVBUS=pci
> 	PHYSDEVDRIVER=qla2xxx
> 	PCI_CLASS=C0400
> 	PCI_ID=1077:2532
> 	PCI_SUBSYS_ID=1077:015C
> 	PCI_SLOT_NAME=0000:1e:00.0
> 	MODALIAS=pci:v00001077d00002532sv00001077sd0000015Cbc0Csc04i00
> 	SEQNUM=3574

Ok, with the kobject_uevent_env() change and the udev-rule modified to:

	SUBSYSTEM=="pci", ENV{DRIVER}="qla2xxx", ACTION=="change", RUN+="qla2xxx_udev.sh"

we're working again.

This seem reasonable?

Thanks, AV

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 29396c0..369a270 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
 		/* do nothing */
 		break;
 	}
-	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
-	    KOBJ_CHANGE, envp);
+	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);
 }
 
 void

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Greg KH]

On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> Randy Dunlap noted:
> 
>   when CONFIG_MODULES=n:
> 
> 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
>   in
> 
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 		KOBJ_CHANGE, envp);
> 
> Trigger kobject event on the 'struct device' hanging off the pci_dev.

Um, why?  What are you trying to do here?  kobject change should not be
for a device, or a "normal" kobject.

What do you expect userspace to do with this?  Where have you documented
it?

confused,

greg k-h

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Andrew Vasquez]

On Fri, 11 Sep 2009, Greg KH wrote:

> On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > Randy Dunlap noted:
> > 
> >   when CONFIG_MODULES=n:
> > 
> > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > 
> >   in
> > 
> > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > 		KOBJ_CHANGE, envp);
> > 
> > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> 
> Um, why?  What are you trying to do here?  kobject change should not be
> for a device, or a "normal" kobject.
> 
> What do you expect userspace to do with this?  Where have you documented
> it?

The purpose was described here:

	http://article.gmane.org/gmane.linux.scsi/54155

Basically we'd like to instruct user-space to retrieve a blob of data
automatically. Original implementation used the kboject hanging off
the module which does not exist when CONFIG_MODULES=n.  It was
suggested that perhaps an alternative would be to use 'struct device'
kobj.  Any tips on how to trigger such a driver-specific event,
perhaps a dedicated kobject exported by the driver itself???

> confused,

anch'io...

Thanks, AV

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[James Bottomley]

On Fri, 2009-09-11 at 17:38 -0700, Andrew Vasquez wrote:
> Randy Dunlap noted:
> 
>   when CONFIG_MODULES=n:
> 
> 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
>   in
> 
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 		KOBJ_CHANGE, envp);
> 
> Trigger kobject event on the 'struct device' hanging off the pci_dev.
> 
> Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
> ---
> 
> > On Fri, 11 Sep 2009, Andrew Vasquez wrote:
> > 
> > > The struct device's kobj from pdev->dev?
> > > 
> > > My udev-foo is pathetic, so how exactly would that get multiplexed at
> > > the udev side?
> > > 
> > > 	KERNEL=="???", SUBSYSTEM=="???", ACTION=="change", RUN+="qla2xxx_udev.sh"
> > 
> > Ok, udevadm helps a bit...  Let me see if I can convert this uevent
> > off of:
> > 
> > 	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);
> > 
> > to something meaningful:
> > 
> > 	UEVENT[1252714722.263731] change
> > 	/devices/pci0000:17/0000:17:08.0/0000:1e:00.0 (pci)
> > 	ACTION=change
> > 	DEVPATH=/devices/pci0000:17/0000:17:08.0/0000:1e:00.0
> > 	SUBSYSTEM=pci
> > 	FW_DUMP=6
> > 	DRIVER=qla2xxx
> > 	PHYSDEVBUS=pci
> > 	PHYSDEVDRIVER=qla2xxx
> > 	PCI_CLASS=C0400
> > 	PCI_ID=1077:2532
> > 	PCI_SUBSYS_ID=1077:015C
> > 	PCI_SLOT_NAME=0000:1e:00.0
> > 	MODALIAS=pci:v00001077d00002532sv00001077sd0000015Cbc0Csc04i00
> > 	SEQNUM=3574
> 
> Ok, with the kobject_uevent_env() change and the udev-rule modified to:
> 
> 	SUBSYSTEM=="pci", ENV{DRIVER}="qla2xxx", ACTION=="change", RUN+="qla2xxx_udev.sh"
> 
> we're working again.
> 
> This seem reasonable?
> 
> Thanks, AV
> 
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 29396c0..369a270 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  		/* do nothing */
>  		break;
>  	}
> -	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> -	    KOBJ_CHANGE, envp);
> +	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);

Much better.  Of course to be perfect, you might like to remember that
(&x)->y is actually x.y

so 

kobject_uevent_env(&vha->hw->pdev->dev.kobj, ...

James

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Greg KH]

On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> On Fri, 11 Sep 2009, Greg KH wrote:
> 
> > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > Randy Dunlap noted:
> > > 
> > >   when CONFIG_MODULES=n:
> > > 
> > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > 
> > >   in
> > > 
> > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > 		KOBJ_CHANGE, envp);
> > > 
> > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > 
> > Um, why?  What are you trying to do here?  kobject change should not be
> > for a device, or a "normal" kobject.
> > 
> > What do you expect userspace to do with this?  Where have you documented
> > it?
> 
> The purpose was described here:
> 
> 	http://article.gmane.org/gmane.linux.scsi/54155
> 
> Basically we'd like to instruct user-space to retrieve a blob of data
> automatically.

Hm, like a firmware object perhaps?

> Original implementation used the kboject hanging off
> the module which does not exist when CONFIG_MODULES=n.  It was
> suggested that perhaps an alternative would be to use 'struct device'
> kobj.  Any tips on how to trigger such a driver-specific event,
> perhaps a dedicated kobject exported by the driver itself???

Why not use the firmware interface for it, that is what it is designed
for, and you will not have to craft any new udev rules.

thanks,

greg k-h

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[James Bottomley]

On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > On Fri, 11 Sep 2009, Greg KH wrote:
> > 
> > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > Randy Dunlap noted:
> > > > 
> > > >   when CONFIG_MODULES=n:
> > > > 
> > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > 
> > > >   in
> > > > 
> > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > 		KOBJ_CHANGE, envp);
> > > > 
> > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > 
> > > Um, why?  What are you trying to do here?  kobject change should not be
> > > for a device, or a "normal" kobject.
> > > 
> > > What do you expect userspace to do with this?  Where have you documented
> > > it?
> > 
> > The purpose was described here:
> > 
> > 	http://article.gmane.org/gmane.linux.scsi/54155
> > 
> > Basically we'd like to instruct user-space to retrieve a blob of data
> > automatically.
> 
> Hm, like a firmware object perhaps?
> 
> > Original implementation used the kboject hanging off
> > the module which does not exist when CONFIG_MODULES=n.  It was
> > suggested that perhaps an alternative would be to use 'struct device'
> > kobj.  Any tips on how to trigger such a driver-specific event,
> > perhaps a dedicated kobject exported by the driver itself???
> 
> Why not use the firmware interface for it, that is what it is designed
> for, and you will not have to craft any new udev rules.

The data is going the wrong way to use the current firmware interface,
which is designed to load data from userspace into the kernel.  For this
interface, we want the data to go the other way (i.e. the kernel has a
blob of dump data it would like userspace to save if it can).

I'd be amenable to updating the firmware interface to do this, but it
looks like adding a completely new codepath, which it's not clear even
belongs there.

James

[PATCHv3] qla2xxx: Correct compilation issues when CONFIG_MODULES=n.

[Andrew Vasquez]

Randy Dunlap noted:

  when CONFIG_MODULES=n:

	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type

  in

	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
		KOBJ_CHANGE, envp);

Trigger kobject event on the 'struct device' hanging off the pci_dev.

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
---

	On Fri, 11 Sep 2009, James Bottomley wrote:

	> > diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
	> > index 29396c0..369a270 100644
	> > --- a/drivers/scsi/qla2xxx/qla_os.c
	> > +++ b/drivers/scsi/qla2xxx/qla_os.c
	> > @@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
	> >  		/* do nothing */
	> >  		break;
	> >  	}
	> > -	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
	> > -	    KOBJ_CHANGE, envp);
	> > +	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);
	> 
	> Much better.  Of course to be perfect, you might like to remember that
	> (&x)->y is actually x.y
	> 
	> so 
	> 
	> kobject_uevent_env(&vha->hw->pdev->dev.kobj, ...

	Ahh, of course...  Perfection...the enemy of progress...

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 29396c0..86f337f 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
 		/* do nothing */
 		break;
 	}
-	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
-	    KOBJ_CHANGE, envp);
+	kobject_uevent_env(&vha->hw->pdev->dev.kobj, KOBJ_CHANGE, envp);
 }
 
 void

Re: [PATCHv3] qla2xxx: Correct compilation issues when CONFIG_MODULES=n.

[Randy Dunlap]

On Sat, 12 Sep 2009 09:43:56 -0700 Andrew Vasquez wrote:

> Randy Dunlap noted:
> 
>   when CONFIG_MODULES=n:
> 
> 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> 
>   in
> 
> 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 		KOBJ_CHANGE, envp);
> 
> Trigger kobject event on the 'struct device' hanging off the pci_dev.
> 
> Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>

That works.  Thanks for your persistence.

Acked-by: Randy Dunlap <randy.dunlap@oracle.com>


> ---
> 
> 	On Fri, 11 Sep 2009, James Bottomley wrote:
> 
> 	> > diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> 	> > index 29396c0..369a270 100644
> 	> > --- a/drivers/scsi/qla2xxx/qla_os.c
> 	> > +++ b/drivers/scsi/qla2xxx/qla_os.c
> 	> > @@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
> 	> >  		/* do nothing */
> 	> >  		break;
> 	> >  	}
> 	> > -	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> 	> > -	    KOBJ_CHANGE, envp);
> 	> > +	kobject_uevent_env(&(&vha->hw->pdev->dev)->kobj, KOBJ_CHANGE, envp);
> 	> 
> 	> Much better.  Of course to be perfect, you might like to remember that
> 	> (&x)->y is actually x.y
> 	> 
> 	> so 
> 	> 
> 	> kobject_uevent_env(&vha->hw->pdev->dev.kobj, ...
> 
> 	Ahh, of course...  Perfection...the enemy of progress...
> 
> diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
> index 29396c0..86f337f 100644
> --- a/drivers/scsi/qla2xxx/qla_os.c
> +++ b/drivers/scsi/qla2xxx/qla_os.c
> @@ -2683,8 +2683,7 @@ qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
>  		/* do nothing */
>  		break;
>  	}
> -	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> -	    KOBJ_CHANGE, envp);
> +	kobject_uevent_env(&vha->hw->pdev->dev.kobj, KOBJ_CHANGE, envp);
>  }
>  
>  void


---
~Randy
LPC 2009, Sept. 23-25, Portland, Oregon
http://linuxplumbersconf.org/2009/

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Greg KH]

On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > > On Fri, 11 Sep 2009, Greg KH wrote:
> > > 
> > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > > Randy Dunlap noted:
> > > > > 
> > > > >   when CONFIG_MODULES=n:
> > > > > 
> > > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > > 
> > > > >   in
> > > > > 
> > > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > > 		KOBJ_CHANGE, envp);
> > > > > 
> > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > > 
> > > > Um, why?  What are you trying to do here?  kobject change should not be
> > > > for a device, or a "normal" kobject.
> > > > 
> > > > What do you expect userspace to do with this?  Where have you documented
> > > > it?
> > > 
> > > The purpose was described here:
> > > 
> > > 	http://article.gmane.org/gmane.linux.scsi/54155
> > > 
> > > Basically we'd like to instruct user-space to retrieve a blob of data
> > > automatically.
> > 
> > Hm, like a firmware object perhaps?
> > 
> > > Original implementation used the kboject hanging off
> > > the module which does not exist when CONFIG_MODULES=n.  It was
> > > suggested that perhaps an alternative would be to use 'struct device'
> > > kobj.  Any tips on how to trigger such a driver-specific event,
> > > perhaps a dedicated kobject exported by the driver itself???
> > 
> > Why not use the firmware interface for it, that is what it is designed
> > for, and you will not have to craft any new udev rules.
> 
> The data is going the wrong way to use the current firmware interface,
> which is designed to load data from userspace into the kernel.  For this
> interface, we want the data to go the other way (i.e. the kernel has a
> blob of dump data it would like userspace to save if it can).

Ick.

> I'd be amenable to updating the firmware interface to do this, but it
> looks like adding a completely new codepath, which it's not clear even
> belongs there.

Well, I don't think it deserves a kobject change event, as that means
something a bit different to userspace today, right?

Either way, I see no documentation being added to the Documentation/ABI/
directory describing what is happening here, that needs to be added at
the least.

thanks,

greg k-h

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[James Bottomley]

On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > > > On Fri, 11 Sep 2009, Greg KH wrote:
> > > > 
> > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > > > Randy Dunlap noted:
> > > > > > 
> > > > > >   when CONFIG_MODULES=n:
> > > > > > 
> > > > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > > > 
> > > > > >   in
> > > > > > 
> > > > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > > > 		KOBJ_CHANGE, envp);
> > > > > > 
> > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > > > 
> > > > > Um, why?  What are you trying to do here?  kobject change should not be
> > > > > for a device, or a "normal" kobject.
> > > > > 
> > > > > What do you expect userspace to do with this?  Where have you documented
> > > > > it?
> > > > 
> > > > The purpose was described here:
> > > > 
> > > > 	http://article.gmane.org/gmane.linux.scsi/54155
> > > > 
> > > > Basically we'd like to instruct user-space to retrieve a blob of data
> > > > automatically.
> > > 
> > > Hm, like a firmware object perhaps?
> > > 
> > > > Original implementation used the kboject hanging off
> > > > the module which does not exist when CONFIG_MODULES=n.  It was
> > > > suggested that perhaps an alternative would be to use 'struct device'
> > > > kobj.  Any tips on how to trigger such a driver-specific event,
> > > > perhaps a dedicated kobject exported by the driver itself???
> > > 
> > > Why not use the firmware interface for it, that is what it is designed
> > > for, and you will not have to craft any new udev rules.
> > 
> > The data is going the wrong way to use the current firmware interface,
> > which is designed to load data from userspace into the kernel.  For this
> > interface, we want the data to go the other way (i.e. the kernel has a
> > blob of dump data it would like userspace to save if it can).
> 
> Ick.

It's a natural debugging event ... the user can simply program the
system to discard the dump.

> > I'd be amenable to updating the firmware interface to do this, but it
> > looks like adding a completely new codepath, which it's not clear even
> > belongs there.
> 
> Well, I don't think it deserves a kobject change event, as that means
> something a bit different to userspace today, right?

Well, the documentation (in include/linux/kobject.h) does say that
KOBJ_CHANGED is the dumping ground for events that don't match anything
else, which this one seems to qualify as.  What else do you suggest?

> Either way, I see no documentation being added to the Documentation/ABI/
> directory describing what is happening here, that needs to be added at
> the least.

OK, we can add that in the next go around.

Andrew, I actually dropped the patch (just this one, not the rest of the
qla2xxx patches) from the tree, so we can work on updating this as a
whole.

James

[PATCHv4] qla2xxx: Add firmware-dump kobject uevent notification.

[Andrew Vasquez]

Signed-off-by: Andrew Vasquez <andrew.vasquez@qlogic.com>
---

On Tue, 15 Sep 2009, James Bottomley wrote:
> On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > > > > On Fri, 11 Sep 2009, Greg KH wrote:
> > > > > 
> > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > > > > Randy Dunlap noted:
> > > > > > > 
> > > > > > >   when CONFIG_MODULES=n:
> > > > > > > 
> > > > > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > > > > 
> > > > > > >   in
> > > > > > > 
> > > > > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > > > > 		KOBJ_CHANGE, envp);
> > > > > > > 
> > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > > > > 
> > > > > > Um, why?  What are you trying to do here?  kobject change should not be
> > > > > > for a device, or a "normal" kobject.
> > > > > > 
> > > > > > What do you expect userspace to do with this?  Where have you documented
> > > > > > it?
> > > > > 
> > > > > The purpose was described here:
> > > > > 
> > > > > 	http://article.gmane.org/gmane.linux.scsi/54155
> > > > > 
> > > > > Basically we'd like to instruct user-space to retrieve a blob of data
> > > > > automatically.
> > > > 
> > > > Hm, like a firmware object perhaps?
> > > > 
> > > > > Original implementation used the kboject hanging off
> > > > > the module which does not exist when CONFIG_MODULES=n.  It was
> > > > > suggested that perhaps an alternative would be to use 'struct device'
> > > > > kobj.  Any tips on how to trigger such a driver-specific event,
> > > > > perhaps a dedicated kobject exported by the driver itself???
> > > > 
> > > > Why not use the firmware interface for it, that is what it is designed
> > > > for, and you will not have to craft any new udev rules.
> > > 
> > > The data is going the wrong way to use the current firmware interface,
> > > which is designed to load data from userspace into the kernel.  For this
> > > interface, we want the data to go the other way (i.e. the kernel has a
> > > blob of dump data it would like userspace to save if it can).
> > 
> > Ick.
> 
> It's a natural debugging event ... the user can simply program the
> system to discard the dump.
> 
> > > I'd be amenable to updating the firmware interface to do this, but it
> > > looks like adding a completely new codepath, which it's not clear even
> > > belongs there.
> > 
> > Well, I don't think it deserves a kobject change event, as that means
> > something a bit different to userspace today, right?
> 
> Well, the documentation (in include/linux/kobject.h) does say that
> KOBJ_CHANGED is the dumping ground for events that don't match anything
> else, which this one seems to qualify as.  What else do you suggest?
> 
> > Either way, I see no documentation being added to the Documentation/ABI/
> > directory describing what is happening here, that needs to be added at
> > the least.
> 
> OK, we can add that in the next go around.
> 
> Andrew, I actually dropped the patch (just this one, not the rest of the
> qla2xxx patches) from the tree, so we can work on updating this as a
> whole.

Ok, how about this squashed commit as a first stab at documenting the
ABI???

Thanks, AV

 Documentation/ABI/stable/sysfs-driver-qla2xxx |    8 +++
 drivers/scsi/qla2xxx/qla_dbg.c                |   78 +++++++-----------------
 drivers/scsi/qla2xxx/qla_def.h                |    5 ++
 drivers/scsi/qla2xxx/qla_gbl.h                |    1 +
 drivers/scsi/qla2xxx/qla_os.c                 |   35 +++++++++++
 5 files changed, 72 insertions(+), 55 deletions(-)
 create mode 100644 Documentation/ABI/stable/sysfs-driver-qla2xxx

diff --git a/Documentation/ABI/stable/sysfs-driver-qla2xxx b/Documentation/ABI/stable/sysfs-driver-qla2xxx
new file mode 100644
index 0000000..9a59d84
--- /dev/null
+++ b/Documentation/ABI/stable/sysfs-driver-qla2xxx
@@ -0,0 +1,8 @@
+What:		/sys/bus/pci/drivers/qla2xxx/.../devices/*
+Date:		September 2009
+Contact:	QLogic Linux Driver <linux-driver@qlogic.com>
+Description:	qla2xxx-udev.sh currently looks for uevent CHANGE events to
+		signal a firmware-dump has been generated by the driver and is
+		ready for retrieval.
+Users:		qla2xxx-udev.sh.  Proposed changes should be mailed to
+		linux-driver@qlogic.com
diff --git a/drivers/scsi/qla2xxx/qla_dbg.c b/drivers/scsi/qla2xxx/qla_dbg.c
index cca8e4a..cb2eca4 100644
--- a/drivers/scsi/qla2xxx/qla_dbg.c
+++ b/drivers/scsi/qla2xxx/qla_dbg.c
@@ -377,6 +377,24 @@ qla25xx_copy_mq(struct qla_hw_data *ha, void *ptr, uint32_t **last_chain)
 	return ptr + sizeof(struct qla2xxx_mq_chain);
 }
 
+static void
+qla2xxx_dump_post_process(scsi_qla_host_t *vha, int rval)
+{
+	struct qla_hw_data *ha = vha->hw;
+
+	if (rval != QLA_SUCCESS) {
+		qla_printk(KERN_WARNING, ha,
+		    "Failed to dump firmware (%x)!!!\n", rval);
+		ha->fw_dumped = 0;
+	} else {
+		qla_printk(KERN_INFO, ha,
+		    "Firmware dump saved to temp buffer (%ld/%p).\n",
+		    vha->host_no, ha->fw_dump);
+		ha->fw_dumped = 1;
+		qla2x00_post_uevent_work(vha, QLA_UEVENT_CODE_FW_DUMP);
+	}
+}
+
 /**
  * qla2300_fw_dump() - Dumps binary data from the 2300 firmware.
  * @ha: HA context
@@ -530,17 +548,7 @@ qla2300_fw_dump(scsi_qla_host_t *vha, int hardware_locked)
 	if (rval == QLA_SUCCESS)
 		qla2xxx_copy_queues(ha, nxt);
 
-	if (rval != QLA_SUCCESS) {
-		qla_printk(KERN_WARNING, ha,
-		    "Failed to dump firmware (%x)!!!\n", rval);
-		ha->fw_dumped = 0;
-
-	} else {
-		qla_printk(KERN_INFO, ha,
-		    "Firmware dump saved to temp buffer (%ld/%p).\n",
-		    base_vha->host_no, ha->fw_dump);
-		ha->fw_dumped = 1;
-	}
+	qla2xxx_dump_post_process(base_vha, rval);
 
 qla2300_fw_dump_failed:
 	if (!hardware_locked)
@@ -737,17 +745,7 @@ qla2100_fw_dump(scsi_qla_host_t *vha, int hardware_locked)
 	if (rval == QLA_SUCCESS)
 		qla2xxx_copy_queues(ha, &fw->risc_ram[cnt]);
 
-	if (rval != QLA_SUCCESS) {
-		qla_printk(KERN_WARNING, ha,
-		    "Failed to dump firmware (%x)!!!\n", rval);
-		ha->fw_dumped = 0;
-
-	} else {
-		qla_printk(KERN_INFO, ha,
-		    "Firmware dump saved to temp buffer (%ld/%p).\n",
-		    base_vha->host_no, ha->fw_dump);
-		ha->fw_dumped = 1;
-	}
+	qla2xxx_dump_post_process(base_vha, rval);
 
 qla2100_fw_dump_failed:
 	if (!hardware_locked)
@@ -984,17 +982,7 @@ qla24xx_fw_dump(scsi_qla_host_t *vha, int hardware_locked)
 	qla24xx_copy_eft(ha, nxt);
 
 qla24xx_fw_dump_failed_0:
-	if (rval != QLA_SUCCESS) {
-		qla_printk(KERN_WARNING, ha,
-		    "Failed to dump firmware (%x)!!!\n", rval);
-		ha->fw_dumped = 0;
-
-	} else {
-		qla_printk(KERN_INFO, ha,
-		    "Firmware dump saved to temp buffer (%ld/%p).\n",
-		    base_vha->host_no, ha->fw_dump);
-		ha->fw_dumped = 1;
-	}
+	qla2xxx_dump_post_process(base_vha, rval);
 
 qla24xx_fw_dump_failed:
 	if (!hardware_locked)
@@ -1305,17 +1293,7 @@ qla25xx_fw_dump(scsi_qla_host_t *vha, int hardware_locked)
 	}
 
 qla25xx_fw_dump_failed_0:
-	if (rval != QLA_SUCCESS) {
-		qla_printk(KERN_WARNING, ha,
-		    "Failed to dump firmware (%x)!!!\n", rval);
-		ha->fw_dumped = 0;
-
-	} else {
-		qla_printk(KERN_INFO, ha,
-		    "Firmware dump saved to temp buffer (%ld/%p).\n",
-		    base_vha->host_no, ha->fw_dump);
-		ha->fw_dumped = 1;
-	}
+	qla2xxx_dump_post_process(base_vha, rval);
 
 qla25xx_fw_dump_failed:
 	if (!hardware_locked)
@@ -1628,17 +1606,7 @@ qla81xx_fw_dump(scsi_qla_host_t *vha, int hardware_locked)
 	}
 
 qla81xx_fw_dump_failed_0:
-	if (rval != QLA_SUCCESS) {
-		qla_printk(KERN_WARNING, ha,
-		    "Failed to dump firmware (%x)!!!\n", rval);
-		ha->fw_dumped = 0;
-
-	} else {
-		qla_printk(KERN_INFO, ha,
-		    "Firmware dump saved to temp buffer (%ld/%p).\n",
-		    base_vha->host_no, ha->fw_dump);
-		ha->fw_dumped = 1;
-	}
+	qla2xxx_dump_post_process(base_vha, rval);
 
 qla81xx_fw_dump_failed:
 	if (!hardware_locked)
diff --git a/drivers/scsi/qla2xxx/qla_def.h b/drivers/scsi/qla2xxx/qla_def.h
index 2150618..d8ce310 100644
--- a/drivers/scsi/qla2xxx/qla_def.h
+++ b/drivers/scsi/qla2xxx/qla_def.h
@@ -2123,6 +2123,7 @@ enum qla_work_type {
 	QLA_EVT_ASYNC_LOGIN_DONE,
 	QLA_EVT_ASYNC_LOGOUT,
 	QLA_EVT_ASYNC_LOGOUT_DONE,
+	QLA_EVT_UEVENT,
 };
 
 
@@ -2146,6 +2147,10 @@ struct qla_work_evt {
 #define QLA_LOGIO_LOGIN_RETRIED	BIT_0
 			u16 data[2];
 		} logio;
+		struct {
+			u32 code;
+#define QLA_UEVENT_CODE_FW_DUMP	0
+		} uevent;
 	} u;
 };
 
diff --git a/drivers/scsi/qla2xxx/qla_gbl.h b/drivers/scsi/qla2xxx/qla_gbl.h
index f3d1d1a..14e0562 100644
--- a/drivers/scsi/qla2xxx/qla_gbl.h
+++ b/drivers/scsi/qla2xxx/qla_gbl.h
@@ -92,6 +92,7 @@ extern int qla2x00_post_async_logout_work(struct scsi_qla_host *, fc_port_t *,
     uint16_t *);
 extern int qla2x00_post_async_logout_done_work(struct scsi_qla_host *,
     fc_port_t *, uint16_t *);
+extern int qla2x00_post_uevent_work(struct scsi_qla_host *, u32);
 
 extern int qla81xx_restart_mpi_firmware(scsi_qla_host_t *);
 
diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index b79fca7..ecf2a40 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -11,6 +11,7 @@
 #include <linux/delay.h>
 #include <linux/kthread.h>
 #include <linux/mutex.h>
+#include <linux/kobject.h>
 
 #include <scsi/scsi_tcq.h>
 #include <scsi/scsicam.h>
@@ -2653,6 +2654,37 @@ qla2x00_post_async_work(login_done, QLA_EVT_ASYNC_LOGIN_DONE);
 qla2x00_post_async_work(logout, QLA_EVT_ASYNC_LOGOUT);
 qla2x00_post_async_work(logout_done, QLA_EVT_ASYNC_LOGOUT_DONE);
 
+int
+qla2x00_post_uevent_work(struct scsi_qla_host *vha, u32 code)
+{
+	struct qla_work_evt *e;
+
+	e = qla2x00_alloc_work(vha, QLA_EVT_UEVENT);
+	if (!e)
+		return QLA_FUNCTION_FAILED;
+
+	e->u.uevent.code = code;
+	return qla2x00_post_work(vha, e);
+}
+
+static void
+qla2x00_uevent_emit(struct scsi_qla_host *vha, u32 code)
+{
+	char event_string[40];
+	char *envp[] = { event_string, NULL };
+
+	switch (code) {
+	case QLA_UEVENT_CODE_FW_DUMP:
+		snprintf(event_string, sizeof(event_string), "FW_DUMP=%ld",
+		    vha->host_no);
+		break;
+	default:
+		/* do nothing */
+		break;
+	}
+	kobject_uevent_env(&vha->hw->pdev->dev.kobj, KOBJ_CHANGE, envp);
+}
+
 void
 qla2x00_do_work(struct scsi_qla_host *vha)
 {
@@ -2690,6 +2722,9 @@ qla2x00_do_work(struct scsi_qla_host *vha)
 			qla2x00_async_logout_done(vha, e->u.logio.fcport,
 			    e->u.logio.data);
 			break;
+		case QLA_EVT_UEVENT:
+			qla2x00_uevent_emit(vha, e->u.uevent.code);
+			break;
 		}
 		if (e->flags & QLA_EVT_FLAG_FREE)
 			kfree(e);
-- 
1.6.5.rc0

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Greg KH]

On Tue, Sep 15, 2009 at 10:47:11AM -0500, James Bottomley wrote:
> On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > > > > On Fri, 11 Sep 2009, Greg KH wrote:
> > > > > 
> > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > > > > Randy Dunlap noted:
> > > > > > > 
> > > > > > >   when CONFIG_MODULES=n:
> > > > > > > 
> > > > > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > > > > 
> > > > > > >   in
> > > > > > > 
> > > > > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > > > > 		KOBJ_CHANGE, envp);
> > > > > > > 
> > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > > > > 
> > > > > > Um, why?  What are you trying to do here?  kobject change should not be
> > > > > > for a device, or a "normal" kobject.
> > > > > > 
> > > > > > What do you expect userspace to do with this?  Where have you documented
> > > > > > it?
> > > > > 
> > > > > The purpose was described here:
> > > > > 
> > > > > 	http://article.gmane.org/gmane.linux.scsi/54155
> > > > > 
> > > > > Basically we'd like to instruct user-space to retrieve a blob of data
> > > > > automatically.
> > > > 
> > > > Hm, like a firmware object perhaps?
> > > > 
> > > > > Original implementation used the kboject hanging off
> > > > > the module which does not exist when CONFIG_MODULES=n.  It was
> > > > > suggested that perhaps an alternative would be to use 'struct device'
> > > > > kobj.  Any tips on how to trigger such a driver-specific event,
> > > > > perhaps a dedicated kobject exported by the driver itself???
> > > > 
> > > > Why not use the firmware interface for it, that is what it is designed
> > > > for, and you will not have to craft any new udev rules.
> > > 
> > > The data is going the wrong way to use the current firmware interface,
> > > which is designed to load data from userspace into the kernel.  For this
> > > interface, we want the data to go the other way (i.e. the kernel has a
> > > blob of dump data it would like userspace to save if it can).
> > 
> > Ick.
> 
> It's a natural debugging event ... the user can simply program the
> system to discard the dump.
> 
> > > I'd be amenable to updating the firmware interface to do this, but it
> > > looks like adding a completely new codepath, which it's not clear even
> > > belongs there.
> > 
> > Well, I don't think it deserves a kobject change event, as that means
> > something a bit different to userspace today, right?
> 
> Well, the documentation (in include/linux/kobject.h) does say that
> KOBJ_CHANGED is the dumping ground for events that don't match anything
> else, which this one seems to qualify as.  What else do you suggest?

Ok, I can't think of anything else at the moment, sorry.

This is using a binary sysfs file, right?

thanks,

greg k-h

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Kay Sievers]

On Tue, Sep 15, 2009 at 18:57, Greg KH <greg@kroah.com> wrote:
> On Tue, Sep 15, 2009 at 10:47:11AM -0500, James Bottomley wrote:
>> On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
>> > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
>> > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
>> > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
>> > > > > On Fri, 11 Sep 2009, Greg KH wrote:
>> > > > >
>> > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
>> > > > > > > Randy Dunlap noted:
>> > > > > > >
>> > > > > > >   when CONFIG_MODULES=n:
>> > > > > > >
>> > > > > > >   drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
>> > > > > > >
>> > > > > > >   in
>> > > > > > >
>> > > > > > >   kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
>> > > > > > >           KOBJ_CHANGE, envp);
>> > > > > > >
>> > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
>> > > > > >
>> > > > > > Um, why?  What are you trying to do here?  kobject change should not be
>> > > > > > for a device, or a "normal" kobject.
>> > > > > >
>> > > > > > What do you expect userspace to do with this?  Where have you documented
>> > > > > > it?
>> > > > >
>> > > > > The purpose was described here:
>> > > > >
>> > > > >       http://article.gmane.org/gmane.linux.scsi/54155
>> > > > >
>> > > > > Basically we'd like to instruct user-space to retrieve a blob of data
>> > > > > automatically.
>> > > >
>> > > > Hm, like a firmware object perhaps?
>> > > >
>> > > > > Original implementation used the kboject hanging off
>> > > > > the module which does not exist when CONFIG_MODULES=n.  It was
>> > > > > suggested that perhaps an alternative would be to use 'struct device'
>> > > > > kobj.  Any tips on how to trigger such a driver-specific event,
>> > > > > perhaps a dedicated kobject exported by the driver itself???
>> > > >
>> > > > Why not use the firmware interface for it, that is what it is designed
>> > > > for, and you will not have to craft any new udev rules.
>> > >
>> > > The data is going the wrong way to use the current firmware interface,
>> > > which is designed to load data from userspace into the kernel.  For this
>> > > interface, we want the data to go the other way (i.e. the kernel has a
>> > > blob of dump data it would like userspace to save if it can).
>> >
>> > Ick.
>>
>> It's a natural debugging event ... the user can simply program the
>> system to discard the dump.
>>
>> > > I'd be amenable to updating the firmware interface to do this, but it
>> > > looks like adding a completely new codepath, which it's not clear even
>> > > belongs there.
>> >
>> > Well, I don't think it deserves a kobject change event, as that means
>> > something a bit different to userspace today, right?
>>
>> Well, the documentation (in include/linux/kobject.h) does say that
>> KOBJ_CHANGED is the dumping ground for events that don't match anything
>> else, which this one seems to qualify as.  What else do you suggest?
>
> Ok, I can't think of anything else at the moment, sorry.

Wouldn't kernel/kmod.c :: call_usermodehelper_pipe() somehow fit for
that interface?

Kay

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Andrew Vasquez]

On Tue, 15 Sep 2009, Greg KH wrote:

> On Tue, Sep 15, 2009 at 10:47:11AM -0500, James Bottomley wrote:
> > On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> > > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> > > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> > > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> > > > > > On Fri, 11 Sep 2009, Greg KH wrote:
> > > > > > 
> > > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> > > > > > > > Randy Dunlap noted:
> > > > > > > > 
> > > > > > > >   when CONFIG_MODULES=n:
> > > > > > > > 
> > > > > > > > 	drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> > > > > > > > 
> > > > > > > >   in
> > > > > > > > 
> > > > > > > > 	kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> > > > > > > > 		KOBJ_CHANGE, envp);
> > > > > > > > 
> > > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> > > > > > > 
> > > > > > > Um, why?  What are you trying to do here?  kobject change should not be
> > > > > > > for a device, or a "normal" kobject.
> > > > > > > 
> > > > > > > What do you expect userspace to do with this?  Where have you documented
> > > > > > > it?
> > > > > > 
> > > > > > The purpose was described here:
> > > > > > 
> > > > > > 	http://article.gmane.org/gmane.linux.scsi/54155
> > > > > > 
> > > > > > Basically we'd like to instruct user-space to retrieve a blob of data
> > > > > > automatically.
> > > > > 
> > > > > Hm, like a firmware object perhaps?
> > > > > 
> > > > > > Original implementation used the kboject hanging off
> > > > > > the module which does not exist when CONFIG_MODULES=n.  It was
> > > > > > suggested that perhaps an alternative would be to use 'struct device'
> > > > > > kobj.  Any tips on how to trigger such a driver-specific event,
> > > > > > perhaps a dedicated kobject exported by the driver itself???
> > > > > 
> > > > > Why not use the firmware interface for it, that is what it is designed
> > > > > for, and you will not have to craft any new udev rules.
> > > > 
> > > > The data is going the wrong way to use the current firmware interface,
> > > > which is designed to load data from userspace into the kernel.  For this
> > > > interface, we want the data to go the other way (i.e. the kernel has a
> > > > blob of dump data it would like userspace to save if it can).
> > > 
> > > Ick.
> > 
> > It's a natural debugging event ... the user can simply program the
> > system to discard the dump.
> > 
> > > > I'd be amenable to updating the firmware interface to do this, but it
> > > > looks like adding a completely new codepath, which it's not clear even
> > > > belongs there.
> > > 
> > > Well, I don't think it deserves a kobject change event, as that means
> > > something a bit different to userspace today, right?
> > 
> > Well, the documentation (in include/linux/kobject.h) does say that
> > KOBJ_CHANGED is the dumping ground for events that don't match anything
> > else, which this one seems to qualify as.  What else do you suggest?
> 
> Ok, I can't think of anything else at the moment, sorry.
> 
> This is using a binary sysfs file, right?

Yes, the userspace tool retrieves the dump from a binary sysfs node:

	/sys/class/scsi_host/hostX/device/fw_dump

-- AV

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Greg KH]

On Tue, Sep 15, 2009 at 07:22:03PM +0200, Kay Sievers wrote:
> On Tue, Sep 15, 2009 at 18:57, Greg KH <greg@kroah.com> wrote:
> > On Tue, Sep 15, 2009 at 10:47:11AM -0500, James Bottomley wrote:
> >> On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> >> > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> >> > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> >> > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> >> > > > > On Fri, 11 Sep 2009, Greg KH wrote:
> >> > > > >
> >> > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> >> > > > > > > Randy Dunlap noted:
> >> > > > > > >
> >> > > > > > >   when CONFIG_MODULES=n:
> >> > > > > > >
> >> > > > > > >   drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> >> > > > > > >
> >> > > > > > >   in
> >> > > > > > >
> >> > > > > > >   kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> >> > > > > > >           KOBJ_CHANGE, envp);
> >> > > > > > >
> >> > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> >> > > > > >
> >> > > > > > Um, why?  What are you trying to do here?  kobject change should not be
> >> > > > > > for a device, or a "normal" kobject.
> >> > > > > >
> >> > > > > > What do you expect userspace to do with this?  Where have you documented
> >> > > > > > it?
> >> > > > >
> >> > > > > The purpose was described here:
> >> > > > >
> >> > > > >       http://article.gmane.org/gmane.linux.scsi/54155
> >> > > > >
> >> > > > > Basically we'd like to instruct user-space to retrieve a blob of data
> >> > > > > automatically.
> >> > > >
> >> > > > Hm, like a firmware object perhaps?
> >> > > >
> >> > > > > Original implementation used the kboject hanging off
> >> > > > > the module which does not exist when CONFIG_MODULES=n.  It was
> >> > > > > suggested that perhaps an alternative would be to use 'struct device'
> >> > > > > kobj.  Any tips on how to trigger such a driver-specific event,
> >> > > > > perhaps a dedicated kobject exported by the driver itself???
> >> > > >
> >> > > > Why not use the firmware interface for it, that is what it is designed
> >> > > > for, and you will not have to craft any new udev rules.
> >> > >
> >> > > The data is going the wrong way to use the current firmware interface,
> >> > > which is designed to load data from userspace into the kernel.  For this
> >> > > interface, we want the data to go the other way (i.e. the kernel has a
> >> > > blob of dump data it would like userspace to save if it can).
> >> >
> >> > Ick.
> >>
> >> It's a natural debugging event ... the user can simply program the
> >> system to discard the dump.
> >>
> >> > > I'd be amenable to updating the firmware interface to do this, but it
> >> > > looks like adding a completely new codepath, which it's not clear even
> >> > > belongs there.
> >> >
> >> > Well, I don't think it deserves a kobject change event, as that means
> >> > something a bit different to userspace today, right?
> >>
> >> Well, the documentation (in include/linux/kobject.h) does say that
> >> KOBJ_CHANGED is the dumping ground for events that don't match anything
> >> else, which this one seems to qualify as.  What else do you suggest?
> >
> > Ok, I can't think of anything else at the moment, sorry.
> 
> Wouldn't kernel/kmod.c :: call_usermodehelper_pipe() somehow fit for
> that interface?

Ah, yeah, good idea.  Anyone tried that?

thanks,

greg k-h

Re: [PATCHv2] qla2xxx: Correct compilation issues when CONFIG_MOUDLES=n.

[Andrew Vasquez]

On Tue, 15 Sep 2009, Kay Sievers wrote:
> On Tue, Sep 15, 2009 at 18:57, Greg KH <greg@kroah.com> wrote:
> > On Tue, Sep 15, 2009 at 10:47:11AM -0500, James Bottomley wrote:
> >> On Tue, 2009-09-15 at 08:33 -0700, Greg KH wrote:
> >> > On Sat, Sep 12, 2009 at 09:30:07AM -0500, James Bottomley wrote:
> >> > > On Fri, 2009-09-11 at 21:33 -0700, Greg KH wrote:
> >> > > > On Fri, Sep 11, 2009 at 07:56:23PM -0700, Andrew Vasquez wrote:
> >> > > > > On Fri, 11 Sep 2009, Greg KH wrote:
> >> > > > >
> >> > > > > > On Fri, Sep 11, 2009 at 05:38:08PM -0700, Andrew Vasquez wrote:
> >> > > > > > > Randy Dunlap noted:
> >> > > > > > >
> >> > > > > > >   when CONFIG_MODULES=n:
> >> > > > > > >
> >> > > > > > >   drivers/scsi/qla2xxx/qla_os.c:2685: error: dereferencing pointer to incomplete type
> >> > > > > > >
> >> > > > > > >   in
> >> > > > > > >
> >> > > > > > >   kobject_uevent_env(&(&vha->hw->pdev->driver->driver)->owner->mkobj.kobj,
> >> > > > > > >           KOBJ_CHANGE, envp);
> >> > > > > > >
> >> > > > > > > Trigger kobject event on the 'struct device' hanging off the pci_dev.
> >> > > > > >
> >> > > > > > Um, why?  What are you trying to do here?  kobject change should not be
> >> > > > > > for a device, or a "normal" kobject.
> >> > > > > >
> >> > > > > > What do you expect userspace to do with this?  Where have you documented
> >> > > > > > it?
> >> > > > >
> >> > > > > The purpose was described here:
> >> > > > >
> >> > > > >       http://article.gmane.org/gmane.linux.scsi/54155
> >> > > > >
> >> > > > > Basically we'd like to instruct user-space to retrieve a blob of data
> >> > > > > automatically.
> >> > > >
> >> > > > Hm, like a firmware object perhaps?
> >> > > >
> >> > > > > Original implementation used the kboject hanging off
> >> > > > > the module which does not exist when CONFIG_MODULES=n.  It was
> >> > > > > suggested that perhaps an alternative would be to use 'struct device'
> >> > > > > kobj.  Any tips on how to trigger such a driver-specific event,
> >> > > > > perhaps a dedicated kobject exported by the driver itself???
> >> > > >
> >> > > > Why not use the firmware interface for it, that is what it is designed
> >> > > > for, and you will not have to craft any new udev rules.
> >> > >
> >> > > The data is going the wrong way to use the current firmware interface,
> >> > > which is designed to load data from userspace into the kernel.  For this
> >> > > interface, we want the data to go the other way (i.e. the kernel has a
> >> > > blob of dump data it would like userspace to save if it can).
> >> >
> >> > Ick.
> >>
> >> It's a natural debugging event ... the user can simply program the
> >> system to discard the dump.
> >>
> >> > > I'd be amenable to updating the firmware interface to do this, but it
> >> > > looks like adding a completely new codepath, which it's not clear even
> >> > > belongs there.
> >> >
> >> > Well, I don't think it deserves a kobject change event, as that means
> >> > something a bit different to userspace today, right?
> >>
> >> Well, the documentation (in include/linux/kobject.h) does say that
> >> KOBJ_CHANGED is the dumping ground for events that don't match anything
> >> else, which this one seems to qualify as.  What else do you suggest?
> >
> > Ok, I can't think of anything else at the moment, sorry.
> 
> Wouldn't kernel/kmod.c :: call_usermodehelper_pipe() somehow fit for
> that interface?

The only current consumer of call_usermodehelper_pipe() is
fs/exec.c::do_coredump() and it appears to manipulate the blob of data
via file-level operations.  Would this be a 'notification'
alternative?  Or, are you suggesting the whole process be redone (both
notification and blob-transfer)?  Uevent notifications with a udev
user-space front-end appears to be a much more simplified and flexible
mechanism.

-- av
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html