Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Domenico Andreoli]

On Mon, Aug 09, 2004 at 02:21:57PM +0900, GOTO Masanori wrote:
> At Sat, 7 Aug 2004 00:50:06 +0200,
> Domenico Andreoli wrote:
> > i have attached a workaround for bug #263348 in case nobody wants to
> > fix it linux-kernel-headers.
> > 
> > it should work although being not that elegant. i know it is fragile
> > since relies on _PPC_ERRNO_H macro being defined by asm/errno.h on
> > powerpc and that could change in any moment.
> > 

the header file used is actually asm/byteroder.h and macro is
_PPC_BYTEORDER_H. 1:3.6.18-2 used asm/error.h but this broke things in
a manner i didn't understand.

> Looking through the kernel source, it seems lkh should be fixed.  I
> put the patch to lkh cvs.
> 

good :)

> However, now we're base-freeze period, and at least I don't decide to
> put new lkh package -18 into unstable and testing-proposed-updates
> yet, so it's good idea to put workaround fix for reiserfsprogs.
> 

already put, uploaded and successfully built :)

> BTW, I think reiserfsprogs is tightly coupled with the kernel version,
> so I wonder it's OK to use lkh package, instead of using kernel
> headers which are included from the latest kernel source to
> reiserfsprogs.
> 

don't know, Hans Reiser should be able to answer this question.

cheers
domenico

-----[ Domenico Andreoli, aka cavok
 --[ http://people.debian.org/~cavok/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Hans Reiser]

Domenico Andreoli wrote:

>On Mon, Aug 09, 2004 at 02:21:57PM +0900, GOTO Masanori wrote:
>  
>
>>At Sat, 7 Aug 2004 00:50:06 +0200,
>>Domenico Andreoli wrote:
>>    
>>
>>>i have attached a workaround for bug #263348 in case nobody wants to
>>>fix it linux-kernel-headers.
>>>
>>>it should work although being not that elegant. i know it is fragile
>>>since relies on _PPC_ERRNO_H macro being defined by asm/errno.h on
>>>powerpc and that could change in any moment.
>>>
>>>      
>>>
>
>the header file used is actually asm/byteroder.h and macro is
>_PPC_BYTEORDER_H. 1:3.6.18-2 used asm/error.h but this broke things in
>a manner i didn't understand.
>
>  
>
>>Looking through the kernel source, it seems lkh should be fixed.  I
>>put the patch to lkh cvs.
>>
>>    
>>
>
>good :)
>
>  
>
>>However, now we're base-freeze period, and at least I don't decide to
>>put new lkh package -18 into unstable and testing-proposed-updates
>>yet, so it's good idea to put workaround fix for reiserfsprogs.
>>
>>    
>>
>
>already put, uploaded and successfully built :)
>
>  
>
>>BTW, I think reiserfsprogs is tightly coupled with the kernel version,
>>so I wonder it's OK to use lkh package, instead of using kernel
>>headers which are included from the latest kernel source to
>>reiserfsprogs.
>>
>>    
>>
>
>don't know, Hans Reiser should be able to answer this question.
>
>cheers
>domenico
>
>-----[ Domenico Andreoli, aka cavok
> --[ http://people.debian.org/~cavok/gpgkey.asc
>   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50
>
>
>  
>
vitaly is the maintainer for progs.  Vitaly?

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Vitaly Fertman]

On Monday 09 August 2004 21:50, Domenico Andreoli wrote:
> On Mon, Aug 09, 2004 at 02:21:57PM +0900, GOTO Masanori wrote:
> > At Sat, 7 Aug 2004 00:50:06 +0200,
> >
> > Domenico Andreoli wrote:
> > > i have attached a workaround for bug #263348 in case nobody wants to
> > > fix it linux-kernel-headers.
> > >
> > > it should work although being not that elegant. i know it is fragile
> > > since relies on _PPC_ERRNO_H macro being defined by asm/errno.h on
> > > powerpc and that could change in any moment.

#if defined(__ppc__) || defined(ppc) || defined(__ppc) || \
  defined(__PPC) || defined(powerpc) || defined(__powerpc__)

?

> the header file used is actually asm/byteroder.h and macro is
> _PPC_BYTEORDER_H. 1:3.6.18-2 used asm/error.h but this broke things in
> a manner i didn't understand.
>
> > Looking through the kernel source, it seems lkh should be fixed.  I
> > put the patch to lkh cvs.
>
> good :)
>
> > However, now we're base-freeze period, and at least I don't decide to
> > put new lkh package -18 into unstable and testing-proposed-updates
> > yet, so it's good idea to put workaround fix for reiserfsprogs.
>
> already put, uploaded and successfully built :)
>
> > BTW, I think reiserfsprogs is tightly coupled with the kernel version,
> > so I wonder it's OK to use lkh package, instead of using kernel
> > headers which are included from the latest kernel source to
> > reiserfsprogs.
>
> don't know, Hans Reiser should be able to answer this question.

kernel has all its includes in the kernel source tree and does not 
use standard ones, reiserfsprogs do not take anything from
any kernel source tree and use their own or standard includes.
asm/unaligned.h is included from glibc-devel on my computer.
Could you check the above macros whether it works for you? 

-- 
Thanks,
Vitaly Fertman

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Domenico Andreoli]

On Thu, Aug 12, 2004 at 05:20:54PM +0400, Vitaly Fertman wrote:
> Hi Domenico, 

hi Vitaly,

> seems that the problem will be solved in the linux-libc-headers finally,
> as reiserfsprogs need a fix anyway, I am going to include the attached

i did not understand you had the bug also out of debian, i would have
replied earlier :)

> one, it looks better than _PPC_ERRNO_H, doesn't it? unfortunately

yes, it does a lot. of course.

> I do not have hardware to test building. if you have, try it please and
> tell me what you think about it.

my powerpc hardware is actually that of debian, which is currently
a little busy (wild and massive rebuild of packages) due to imminent
release of the distribution. as soon as i'm able to check i will report
you (in this moment i not even able to login).

regards
domenico

-----[ Domenico Andreoli, aka cavok
 --[ http://people.debian.org/~cavok/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Domenico Andreoli]

hi Vitaly,

On Thu, Aug 12, 2004 at 03:57:30PM +0200, Domenico Andreoli wrote:
> On Thu, Aug 12, 2004 at 05:20:54PM +0400, Vitaly Fertman wrote:
> >
> > I do not have hardware to test building. if you have, try it please and
> > tell me what you think about it.
> 
> my powerpc hardware is actually that of debian, which is currently
> a little busy (wild and massive rebuild of packages) due to imminent
> release of the distribution. as soon as i'm able to check i will report
> you (in this moment i not even able to login).
> 

i can now confirm that you patch to fix broken powerpc linux kernel
headers does it work, the build succeeded.

cheers
domenico

-----[ Domenico Andreoli, aka cavok
 --[ http://people.debian.org/~cavok/gpgkey.asc
   ---[ 3A0F 2F80 F79C 678A 8936  4FEE 0677 9033 A20E BC50

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Vitaly Fertman]

On Wednesday 18 August 2004 15:03, Domenico Andreoli wrote:
> hi Vitaly,
>
> On Thu, Aug 12, 2004 at 03:57:30PM +0200, Domenico Andreoli wrote:
> > On Thu, Aug 12, 2004 at 05:20:54PM +0400, Vitaly Fertman wrote:
> > > I do not have hardware to test building. if you have, try it please and
> > > tell me what you think about it.
> >
> > my powerpc hardware is actually that of debian, which is currently
> > a little busy (wild and massive rebuild of packages) due to imminent
> > release of the distribution. as soon as i'm able to check i will report
> > you (in this moment i not even able to login).
>
> i can now confirm that you patch to fix broken powerpc linux kernel
> headers does it work, the build succeeded.

Thank You

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[GOTO Masanori]

Hi,

At Tue, 10 Aug 2004 15:39:10 +0400,
Vitaly Fertman wrote:
> > the header file used is actually asm/byteroder.h and macro is
> > _PPC_BYTEORDER_H. 1:3.6.18-2 used asm/error.h but this broke things in
> > a manner i didn't understand.
> >
> > > Looking through the kernel source, it seems lkh should be fixed.  I
> > > put the patch to lkh cvs.
> >
> > good :)
> >
> > > However, now we're base-freeze period, and at least I don't decide to
> > > put new lkh package -18 into unstable and testing-proposed-updates
> > > yet, so it's good idea to put workaround fix for reiserfsprogs.
> >
> > already put, uploaded and successfully built :)
> >
> > > BTW, I think reiserfsprogs is tightly coupled with the kernel version,
> > > so I wonder it's OK to use lkh package, instead of using kernel
> > > headers which are included from the latest kernel source to
> > > reiserfsprogs.
> >
> > don't know, Hans Reiser should be able to answer this question.
> 
> kernel has all its includes in the kernel source tree and does not 
> use standard ones, reiserfsprogs do not take anything from
> any kernel source tree and use their own or standard includes.
> asm/unaligned.h is included from glibc-devel on my computer.

I guess "glibc-devel" in Fedora Core is RedHat .rpm package, and I
confirmed that it does not have any header files.  If reiserfsprogs
uses asm/unaligned.h in FC, then it's included in "glibc-kernheaders"
package and this file is actually derived from linux kernel.  In
Debian, we use "linux-kernel-headers" package that role is equivalent
to FC's glibc-kernheaders package.

I sent the patch to ppc guys to fix this problem.  If they apply my
patch, the problem should be gone away.

However, I think the question is still remained: should reiserfsprogs
include some kernel version headers for each reiserfs modification
versions, instead of using system headers directly?

Regards,
-- gotom

Re: Bug#263348: linux-kernel-headers: asm/unaligned.h on powerpc is useless if __KERNEL__ is undefined

[Vitaly Fertman]

On Wednesday 18 August 2004 11:18, GOTO Masanori wrote:
> Hi,
>
> At Tue, 10 Aug 2004 15:39:10 +0400,
>
> Vitaly Fertman wrote:
> > > the header file used is actually asm/byteroder.h and macro is
> > > _PPC_BYTEORDER_H. 1:3.6.18-2 used asm/error.h but this broke things in
> > > a manner i didn't understand.
> > >
> > > > Looking through the kernel source, it seems lkh should be fixed.  I
> > > > put the patch to lkh cvs.
> > >
> > > good :)
> > >
> > > > However, now we're base-freeze period, and at least I don't decide to
> > > > put new lkh package -18 into unstable and testing-proposed-updates
> > > > yet, so it's good idea to put workaround fix for reiserfsprogs.
> > >
> > > already put, uploaded and successfully built :)
> > >
> > > > BTW, I think reiserfsprogs is tightly coupled with the kernel
> > > > version, so I wonder it's OK to use lkh package, instead of using
> > > > kernel headers which are included from the latest kernel source to
> > > > reiserfsprogs.
> > >
> > > don't know, Hans Reiser should be able to answer this question.
> >
> > kernel has all its includes in the kernel source tree and does not
> > use standard ones, reiserfsprogs do not take anything from
> > any kernel source tree and use their own or standard includes.
> > asm/unaligned.h is included from glibc-devel on my computer.
>
> I guess "glibc-devel" in Fedora Core is RedHat .rpm package, and I
> confirmed that it does not have any header files.  If reiserfsprogs
> uses asm/unaligned.h in FC, then it's included in "glibc-kernheaders"
> package and this file is actually derived from linux kernel.  In
> Debian, we use "linux-kernel-headers" package that role is equivalent
> to FC's glibc-kernheaders package.
>
> I sent the patch to ppc guys to fix this problem.  If they apply my
> patch, the problem should be gone away.
>
> However, I think the question is still remained: should reiserfsprogs
> include some kernel version headers for each reiserfs modification
> versions, instead of using system headers directly?

what headers are you talking about? if about reiserfs headers, progs do 
not include reiserfs kernel headers, they have their own.

-- 
Thanks,
Vitaly Fertman