[KJ] compile warnings due to change of pci resource element type

[KJ] compile warnings due to change of pci resource element type

[Stephen Biggs]

[-- Attachment #1: Mail message body --]
[-- Type: text/plain, Size: 587 bytes --]

Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount 
of warnings in a lot of files due to printf formats (and a couple of 
other minor issues) that have %lx instead of %llx for the "start" element 
in the pci_resource struct whose type was changed from u32 to u64.

These are relatively easy to fix and need to be fixed because of 
potential stack issues with var arg functions like printk and friends.

My question is: should I do this or is someone else working on this, 
perhaps a better fix than simply changing the majority of occurrences of 
"%lx" to "%llx"?


[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Randy.Dunlap]

Stephen Biggs wrote:
> Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount 
> of warnings in a lot of files due to printf formats (and a couple of 
> other minor issues) that have %lx instead of %llx for the "start" element 
> in the pci_resource struct whose type was changed from u32 to u64.
> 
> These are relatively easy to fix and need to be fixed because of 
> potential stack issues with var arg functions like printk and friends.
> 
> My question is: should I do this or is someone else working on this, 
> perhaps a better fix than simply changing the majority of occurrences of 
> "%lx" to "%llx"?

I read messages on lkml that the pci resource patch was being dropped
temporarily to fix some issues.... Perhaps its patch submitter will
be fixing those, I dunno.

-- 
~Randy
_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Greg KH]

[-- Attachment #1: Type: text/plain, Size: 1115 bytes --]

On Fri, Mar 04, 2005 at 07:49:45AM -0800, Randy.Dunlap wrote:
> Stephen Biggs wrote:
> >Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount 
> >of warnings in a lot of files due to printf formats (and a couple of 
> >other minor issues) that have %lx instead of %llx for the "start" element 
> >in the pci_resource struct whose type was changed from u32 to u64.
> >
> >These are relatively easy to fix and need to be fixed because of 
> >potential stack issues with var arg functions like printk and friends.
> >
> >My question is: should I do this or is someone else working on this, 
> >perhaps a better fix than simply changing the majority of occurrences of 
> >"%lx" to "%llx"?
> 
> I read messages on lkml that the pci resource patch was being dropped
> temporarily to fix some issues.... Perhaps its patch submitter will
> be fixing those, I dunno.

Odds are, it will be me :(

So yes, I will gladly take those kinds of patches.  But you need to
convert:
	"%lx", resource_variable
 to:
	"%llx", (unsigned long long)resource_variable

to make it work on all platforms.

thanks,

greg k-h

[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Stephen Biggs]

[-- Attachment #1: Mail message body --]
[-- Type: text/plain, Size: 1746 bytes --]

On 4 Mar 2005 at 8:57, Greg KH wrote:

> On Fri, Mar 04, 2005 at 07:49:45AM -0800, Randy.Dunlap wrote:
> > Stephen Biggs wrote:
> > >Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount 
> > >of warnings in a lot of files due to printf formats (and a couple of 
> > >other minor issues) that have %lx instead of %llx for the "start" element 
> > >in the pci_resource struct whose type was changed from u32 to u64.
> > >
> > >These are relatively easy to fix and need to be fixed because of 
> > >potential stack issues with var arg functions like printk and friends.
> > >
> > >My question is: should I do this or is someone else working on this, 
> > >perhaps a better fix than simply changing the majority of occurrences of 
> > >"%lx" to "%llx"?
> > 
> > I read messages on lkml that the pci resource patch was being dropped
> > temporarily to fix some issues.... Perhaps its patch submitter will
> > be fixing those, I dunno.
> 
> Odds are, it will be me :(
> 
> So yes, I will gladly take those kinds of patches.

Yes, OK, I'll start working on it soon.

Quick question: I see these problems on 2.6.11-rc5-mm1 but NOT on the 
mainline kernel of 2.6.11.  Also, see my previous email about 2.6.11-kj 
patch not applying cleanly at all to the rc5-mm1 BUT applies perfectly to 
the mainline 2.6.11.

Which kernel version should I be working with?  Since there is now a 
release of 2.6.11, isn't the proper version for us now 2.6.12-rc1, as 
soon as it's created??

I am confused.

>  But you need to convert: 
> 	"%lx", resource_variable
>  to:
> 	"%llx", (unsigned long long)resource_variable
> 
> to make it work on all platforms.

Shouldn't it be (u64) instead of (unsigned long long)?  

> 
> thanks,
> 
> greg k-h
> 




[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Alexey Dobriyan]

[-- Attachment #1: Type: text/plain, Size: 531 bytes --]

On Saturday 05 March 2005 11:33, Stephen Biggs wrote:

> >  But you need to convert: 
> > 	"%lx", resource_variable
> >  to:
> > 	"%llx", (unsigned long long)resource_variable
> > 
> > to make it work on all platforms.
> 
> Shouldn't it be (u64) instead of (unsigned long long)?  

No. Otherwise it will generate warnings on 64-bit machines. Ie:

drivers/block/aoe/aoeblk.c:245: warning: long long unsigned int format, u64 arg (arg 2)
drivers/block/aoe/aoeblk.c:31: warning: long long unsigned int format, u64 arg (arg 4)

	Alexey

[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Nish Aravamudan]

[-- Attachment #1: Type: text/plain, Size: 2227 bytes --]

On Sat, 05 Mar 2005 11:33:37 +0200, Stephen Biggs
<yrgrknmxpzlk@gawab.com> wrote:
> On 4 Mar 2005 at 8:57, Greg KH wrote:
> 
> > On Fri, Mar 04, 2005 at 07:49:45AM -0800, Randy.Dunlap wrote:
> > > Stephen Biggs wrote:
> > > >Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount
> > > >of warnings in a lot of files due to printf formats (and a couple of
> > > >other minor issues) that have %lx instead of %llx for the "start" element
> > > >in the pci_resource struct whose type was changed from u32 to u64.
> > > >
> > > >These are relatively easy to fix and need to be fixed because of
> > > >potential stack issues with var arg functions like printk and friends.
> > > >
> > > >My question is: should I do this or is someone else working on this,
> > > >perhaps a better fix than simply changing the majority of occurrences of
> > > >"%lx" to "%llx"?
> > >
> > > I read messages on lkml that the pci resource patch was being dropped
> > > temporarily to fix some issues.... Perhaps its patch submitter will
> > > be fixing those, I dunno.
> >
> > Odds are, it will be me :(
> >
> > So yes, I will gladly take those kinds of patches.
> 
> Yes, OK, I'll start working on it soon.
> 
> Quick question: I see these problems on 2.6.11-rc5-mm1 but NOT on the
> mainline kernel of 2.6.11.  Also, see my previous email about 2.6.11-kj
> patch not applying cleanly at all to the rc5-mm1 BUT applies perfectly to
> the mainline 2.6.11.
> 
> Which kernel version should I be working with?  Since there is now a
> release of 2.6.11, isn't the proper version for us now 2.6.12-rc1, as
> soon as it's created??
> 
> I am confused.

I am not sure why you would think 2.6.11-kj would apply to anything
except 2.6.11. That's why it's *2.6.11*-kj not 2.6.11-rc5-mm1-kj.
That's the whole point of labelling the releases that way.

So, if you want to apply -kj, you apply to it's prefix. Sometimes,
admittedly, it will apply with fuzz to other trees, but that's
incidental.

Also, please read the threads regarding the new kernel naming
procedure. You should apply your patches against mainline (currently
2.6.11), unless the issue only exists in -mm and the patch which
causes the issue is also only in -mm.

Thanks,
Nish

[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Stephen Biggs]

[-- Attachment #1: Mail message body --]
[-- Type: text/plain, Size: 3434 bytes --]

On 5 Mar 2005 at 8:04, Nish Aravamudan wrote:

> On Sat, 05 Mar 2005 11:33:37 +0200, Stephen Biggs
> <yrgrknmxpzlk@gawab.com> wrote:
> > On 4 Mar 2005 at 8:57, Greg KH wrote:
> > 
> > > On Fri, Mar 04, 2005 at 07:49:45AM -0800, Randy.Dunlap wrote:
> > > > Stephen Biggs wrote:
> > > > >Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount
> > > > >of warnings in a lot of files due to printf formats (and a couple of
> > > > >other minor issues) that have %lx instead of %llx for the "start" element
> > > > >in the pci_resource struct whose type was changed from u32 to u64.
> > > > >
> > > > >These are relatively easy to fix and need to be fixed because of
> > > > >potential stack issues with var arg functions like printk and friends.
> > > > >
> > > > >My question is: should I do this or is someone else working on this,
> > > > >perhaps a better fix than simply changing the majority of occurrences of
> > > > >"%lx" to "%llx"?
> > > >
> > > > I read messages on lkml that the pci resource patch was being dropped
> > > > temporarily to fix some issues.... Perhaps its patch submitter will
> > > > be fixing those, I dunno.
> > >
> > > Odds are, it will be me :(
> > >
> > > So yes, I will gladly take those kinds of patches.
> > 
> > Yes, OK, I'll start working on it soon.
> > 
> > Quick question: I see these problems on 2.6.11-rc5-mm1 but NOT on the
> > mainline kernel of 2.6.11.  Also, see my previous email about 2.6.11-kj
> > patch not applying cleanly at all to the rc5-mm1 BUT applies perfectly to
> > the mainline 2.6.11.
> > 
> > Which kernel version should I be working with?  Since there is now a
> > release of 2.6.11, isn't the proper version for us now 2.6.12-rc1, as
> > soon as it's created??
> > 
> > I am confused.
> 
> I am not sure why you would think 2.6.11-kj would apply to anything
> except 2.6.11. That's why it's *2.6.11*-kj not 2.6.11-rc5-mm1-kj.
> That's the whole point of labelling the releases that way.

Yes, thanks for the clear explanation.

> 
> So, if you want to apply -kj, you apply to it's prefix. Sometimes,
> admittedly, it will apply with fuzz to other trees, but that's
> incidental.

Ok, this I can understand and accept.

> 
> Also, please read the threads regarding the new kernel naming
> procedure. You should apply your patches against mainline (currently
> 2.6.11), unless the issue only exists in -mm and the patch which
> causes the issue is also only in -mm.

Here is the source of my confusion:
http://lists.osdl.org/pipermail/kernel-janitors/2005-February/003315.html

... wherein Jim Nelson says, in part: "Work off of -mm, since most of the 
janitorial work will go through Andrew Morton's series for awhile before 
being pushed into mainline."

Can you please send me links to the threads you are referring to so I can 
read them for myself and get a grip on this?

However, even with the naming issues cleared up in my head, I am still 
confused by the other issue. I am assuming that the pci resource retype 
is only in -mm?  If so, I am still confused about which version to work 
off of.  That is, if I apply the -kj patch to the mainline, I should then 
apply the -mm patch that just now appeared in Andrew Morton's directore 
for the mainline?

Maybe this is moot right now, since I see that this particular patch WAS 
in 2.6.11-rc5-mm1, but is not in 2.6.11-mm1.  Guess I'll wait until this 
is placed back in.

> 
> Thanks,
> Nish
> 




[-- Attachment #2: Type: text/plain, Size: 167 bytes --]

_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors

Re: [KJ] compile warnings due to change of pci resource element type

[Randy.Dunlap]

Stephen Biggs wrote:
> On 5 Mar 2005 at 8:04, Nish Aravamudan wrote:
> 
> 
>>On Sat, 05 Mar 2005 11:33:37 +0200, Stephen Biggs
>><yrgrknmxpzlk@gawab.com> wrote:
>>
>>>On 4 Mar 2005 at 8:57, Greg KH wrote:
>>>
>>>
>>>>On Fri, Mar 04, 2005 at 07:49:45AM -0800, Randy.Dunlap wrote:
>>>>
>>>>>Stephen Biggs wrote:
>>>>>
>>>>>>Trying to comple the latest version (2.6.11-rc5-mm1) gives a huge amount
>>>>>>of warnings in a lot of files due to printf formats (and a couple of
>>>>>>other minor issues) that have %lx instead of %llx for the "start" element
>>>>>>in the pci_resource struct whose type was changed from u32 to u64.
>>>>>>
>>>>>>These are relatively easy to fix and need to be fixed because of
>>>>>>potential stack issues with var arg functions like printk and friends.
>>>>>>
>>>>>>My question is: should I do this or is someone else working on this,
>>>>>>perhaps a better fix than simply changing the majority of occurrences of
>>>>>>"%lx" to "%llx"?
>>>>>
>>>>>I read messages on lkml that the pci resource patch was being dropped
>>>>>temporarily to fix some issues.... Perhaps its patch submitter will
>>>>>be fixing those, I dunno.
>>>>
>>>>Odds are, it will be me :(
>>>>
>>>>So yes, I will gladly take those kinds of patches.
>>>
>>>Yes, OK, I'll start working on it soon.
>>>
>>>Quick question: I see these problems on 2.6.11-rc5-mm1 but NOT on the
>>>mainline kernel of 2.6.11.  Also, see my previous email about 2.6.11-kj
>>>patch not applying cleanly at all to the rc5-mm1 BUT applies perfectly to
>>>the mainline 2.6.11.
>>>
>>>Which kernel version should I be working with?  Since there is now a
>>>release of 2.6.11, isn't the proper version for us now 2.6.12-rc1, as
>>>soon as it's created??
>>>
>>>I am confused.
>>
>>I am not sure why you would think 2.6.11-kj would apply to anything
>>except 2.6.11. That's why it's *2.6.11*-kj not 2.6.11-rc5-mm1-kj.
>>That's the whole point of labelling the releases that way.
> 
> 
> Yes, thanks for the clear explanation.
> 
> 
>>So, if you want to apply -kj, you apply to it's prefix. Sometimes,
>>admittedly, it will apply with fuzz to other trees, but that's
>>incidental.
> 
> 
> Ok, this I can understand and accept.
> 
> 
>>Also, please read the threads regarding the new kernel naming
>>procedure. You should apply your patches against mainline (currently
>>2.6.11), unless the issue only exists in -mm and the patch which
>>causes the issue is also only in -mm.
> 
> 
> Here is the source of my confusion:
> http://lists.osdl.org/pipermail/kernel-janitors/2005-February/003315.html
> 
> ... wherein Jim Nelson says, in part: "Work off of -mm, since most of the 
> janitorial work will go through Andrew Morton's series for awhile before 
> being pushed into mainline."

This is the problem, then.  Most developers work off of Linus's
tree (bk or bk snapshots) as a base.  Even Andrew suggests
doing that unless the code that you are patching is only in
-mm.  Andrew must work 20 hours/day, because he is willing
to convert patches to merge into his -mm patchset.

> Can you please send me links to the threads you are referring to so I can 
> read them for myself and get a grip on this?
> 
> However, even with the naming issues cleared up in my head, I am still 
> confused by the other issue. I am assuming that the pci resource retype 
> is only in -mm?  If so, I am still confused about which version to work 
> off of.  That is, if I apply the -kj patch to the mainline, I should then 
> apply the -mm patch that just now appeared in Andrew Morton's directore 
> for the mainline?
> 
> Maybe this is moot right now, since I see that this particular patch WAS 
> in 2.6.11-rc5-mm1, but is not in 2.6.11-mm1.  Guess I'll wait until this 
> is placed back in.

or find the most recent version of that patch (from 2.6.11-rc5-mm1),
apply it to 2.6.11-linus, and start making fixes to it.
Post the updates to lkml and linux-pci mailing list and cc:
Greg on them.

-- 
~Randy
_______________________________________________
Kernel-janitors mailing list
Kernel-janitors@lists.osdl.org
http://lists.osdl.org/mailman/listinfo/kernel-janitors