2.6.21-git5 -- powerpc link failure

2.6.21-git5 -- powerpc link failure

[Andy Whitcroft]

Since 2.6.21-git5 we've been getting the following link failure on powerpc:

  CC      drivers/pci/hotplug/rpadlpar_core.o
  CC      drivers/pci/hotplug/rpadlpar_sysfs.o
drivers/pci/hotplug/rpadlpar_sysfs.c:132: error: unknown field
‘subsys’ specified in initializer
drivers/pci/hotplug/rpadlpar_sysfs.c:132: warning: initialization from
incompatible pointer type
make[3]: *** [drivers/pci/hotplug/rpadlpar_sysfs.o] Error 1
make[2]: *** [drivers/pci/hotplug] Error 2
make[1]: *** [drivers/pci] Error 2
make: *** [drivers] Error 2


I would think this is caused by the changes in the commit below:

commit 823bccfc4002296ba88c3ad0f049e1abd8108d30
Author: Greg Kroah-Hartman <gregkh@suse.de>
Date:   Fri Apr 13 13:15:19 2007 -0700

    remove "struct subsystem" as it is no longer needed

-apw

[PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Andy Whitcroft]

Move the rpadlpar device from "struct subsystem" to "struct kset"
following the changes in sysfs.

Signed-off-by: Andy Whitcroft <apw@shadowen.org>
---

	Ok, this patch seems to sort out the compile problem
	here and indeed boots and runs kernbench.  Perhaps
	you could confirm this is sufficient.


diff --git a/drivers/pci/hotplug/rpadlpar_sysfs.c b/drivers/pci/hotplug/rpadlpar_sysfs.c
index 6c5be3f..ed4de59 100644
--- a/drivers/pci/hotplug/rpadlpar_sysfs.c
+++ b/drivers/pci/hotplug/rpadlpar_sysfs.c
@@ -129,13 +129,13 @@ struct kobj_type ktype_dlpar_io = {
 };
 
 struct kset dlpar_io_kset = {
-	.subsys = &pci_hotplug_slots_subsys,
 	.kobj = {.name = DLPAR_KOBJ_NAME, .ktype=&ktype_dlpar_io,},
 	.ktype = &ktype_dlpar_io,
 };
 
 int dlpar_sysfs_init(void)
 {
+	kobj_set_kset_s(&dlpar_io_kset, pci_hotplug_slots_subsys);
 	if (kset_register(&dlpar_io_kset)) {
 		printk(KERN_ERR "rpadlpar_io: cannot register kset for %s\n",
 				dlpar_io_kset.kobj.name);

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Greg KH]

On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
> 
> Move the rpadlpar device from "struct subsystem" to "struct kset"
> following the changes in sysfs.
> 
> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
> ---
> 
> 	Ok, this patch seems to sort out the compile problem
> 	here and indeed boots and runs kernbench.  Perhaps
> 	you could confirm this is sufficient.

As per the discussion on the pci hotplug list, no, this doesn't seem to
fix the problem.  The developers there are looking into it.  If you can
test out patches for this, I'm sure the people there would appreciate
the help.

thanks,

greg k-h

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Andy Whitcroft]

Greg KH wrote:
> On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
>> Move the rpadlpar device from "struct subsystem" to "struct kset"
>> following the changes in sysfs.
>>
>> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
>> ---
>>
>> 	Ok, this patch seems to sort out the compile problem
>> 	here and indeed boots and runs kernbench.  Perhaps
>> 	you could confirm this is sufficient.
> 
> As per the discussion on the pci hotplug list, no, this doesn't seem to
> fix the problem.  The developers there are looking into it.  If you can
> test out patches for this, I'm sure the people there would appreciate
> the help.

Sure anything they have for testing, send them to me ...

-apw

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Greg KH]

On Thu, May 10, 2007 at 04:54:41PM +0100, Andy Whitcroft wrote:
> Greg KH wrote:
> > On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
> >> Move the rpadlpar device from "struct subsystem" to "struct kset"
> >> following the changes in sysfs.
> >>
> >> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
> >> ---
> >>
> >> 	Ok, this patch seems to sort out the compile problem
> >> 	here and indeed boots and runs kernbench.  Perhaps
> >> 	you could confirm this is sufficient.
> > 
> > As per the discussion on the pci hotplug list, no, this doesn't seem to
> > fix the problem.  The developers there are looking into it.  If you can
> > test out patches for this, I'm sure the people there would appreciate
> > the help.
> 
> Sure anything they have for testing, send them to me ...

They have the same patch that you made (I made it), yet they reported
that it didn't work properly for them.

Can you test your patch out on "real" hardware?

thanks,

greg k-h

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Michael Ellerman]

[-- Attachment #1: Type: text/plain, Size: 1424 bytes --]

On Fri, 2007-05-11 at 00:16 -0700, Greg KH wrote:
> On Thu, May 10, 2007 at 04:54:41PM +0100, Andy Whitcroft wrote:
> > Greg KH wrote:
> > > On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
> > >> Move the rpadlpar device from "struct subsystem" to "struct kset"
> > >> following the changes in sysfs.
> > >>
> > >> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
> > >> ---
> > >>
> > >> 	Ok, this patch seems to sort out the compile problem
> > >> 	here and indeed boots and runs kernbench.  Perhaps
> > >> 	you could confirm this is sufficient.
> > > 
> > > As per the discussion on the pci hotplug list, no, this doesn't seem to
> > > fix the problem.  The developers there are looking into it.  If you can
> > > test out patches for this, I'm sure the people there would appreciate
> > > the help.
> > 
> > Sure anything they have for testing, send them to me ...
> 
> They have the same patch that you made (I made it), yet they reported
> that it didn't work properly for them.
> 
> Can you test your patch out on "real" hardware?

I tested it on real hardware, but it can't hurt for Andy to try it too I
guess.

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Andy Whitcroft]

Michael Ellerman wrote:
> On Fri, 2007-05-11 at 00:16 -0700, Greg KH wrote:
>> On Thu, May 10, 2007 at 04:54:41PM +0100, Andy Whitcroft wrote:
>>> Greg KH wrote:
>>>> On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
>>>>> Move the rpadlpar device from "struct subsystem" to "struct kset"
>>>>> following the changes in sysfs.
>>>>>
>>>>> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
>>>>> ---
>>>>>
>>>>> 	Ok, this patch seems to sort out the compile problem
>>>>> 	here and indeed boots and runs kernbench.  Perhaps
>>>>> 	you could confirm this is sufficient.
>>>> As per the discussion on the pci hotplug list, no, this doesn't seem to
>>>> fix the problem.  The developers there are looking into it.  If you can
>>>> test out patches for this, I'm sure the people there would appreciate
>>>> the help.
>>> Sure anything they have for testing, send them to me ...
>> They have the same patch that you made (I made it), yet they reported
>> that it didn't work properly for them.
>>
>> Can you test your patch out on "real" hardware?
> 
> I tested it on real hardware, but it can't hurt for Andy to try it too I
> guess.

To be fair I am not sure I have a clue how to test it.  Got a recipe?
My patch was based on how other drivers seemed to be converted which is
a concern for those drivers.

What sort of failure do you see?

-apw

Re: [PATCH] update sysfs kset initialisation in PPC64 DLPAR IO driver

[Michael Ellerman]

[-- Attachment #1: Type: text/plain, Size: 2032 bytes --]

On Fri, 2007-05-11 at 11:40 +0100, Andy Whitcroft wrote:
> Michael Ellerman wrote:
> > On Fri, 2007-05-11 at 00:16 -0700, Greg KH wrote:
> >> On Thu, May 10, 2007 at 04:54:41PM +0100, Andy Whitcroft wrote:
> >>> Greg KH wrote:
> >>>> On Thu, May 10, 2007 at 03:00:50PM +0100, Andy Whitcroft wrote:
> >>>>> Move the rpadlpar device from "struct subsystem" to "struct kset"
> >>>>> following the changes in sysfs.
> >>>>>
> >>>>> Signed-off-by: Andy Whitcroft <apw@shadowen.org>
> >>>>> ---
> >>>>>
> >>>>> 	Ok, this patch seems to sort out the compile problem
> >>>>> 	here and indeed boots and runs kernbench.  Perhaps
> >>>>> 	you could confirm this is sufficient.
> >>>> As per the discussion on the pci hotplug list, no, this doesn't seem to
> >>>> fix the problem.  The developers there are looking into it.  If you can
> >>>> test out patches for this, I'm sure the people there would appreciate
> >>>> the help.
> >>> Sure anything they have for testing, send them to me ...
> >> They have the same patch that you made (I made it), yet they reported
> >> that it didn't work properly for them.
> >>
> >> Can you test your patch out on "real" hardware?
> > 
> > I tested it on real hardware, but it can't hurt for Andy to try it too I
> > guess.
> 
> To be fair I am not sure I have a clue how to test it.  Got a recipe?
> My patch was based on how other drivers seemed to be converted which is
> a concern for those drivers.
> 
> What sort of failure do you see?

Prior to the removal of struct subysystem I get two files called
'add_slot' and 'remove_slot' under /sys/bus/pci/slots/control.

With Greg's patch I get the directory /sys/bus/pci/slots/control, but
nothing under it. 

Apparently John Rose is looking into it.

cheers

-- 
Michael Ellerman
OzLabs, IBM Australia Development Lab

wwweb: http://michael.ellerman.id.au
phone: +61 2 6212 1183 (tie line 70 21183)

We do not inherit the earth from our ancestors,
we borrow it from our children. - S.M.A.R.T Person

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]