[GIT PULL] Blackfin arch fixes (try #2)

[GIT PULL] Blackfin arch fixes (try #2)

[Bryan Wu]

Hi Linus:

Marco's patch will kill the zero file git-pull error.
Jie's patch is required because we will release our new Blackfin toolchain.

Please pull from "master" branch of:

  master.kernel.org:/pub/scm/linux/kernel/git/cooloney/blackfin-2.6.git/ master

Jie Zhang (1):
      Blackfin arch: Add proper -mcpu option according to the cpu and silicon revision configuration

Marco Roeland (1):
      Blackfin arch: remove zero-sized include/asm-blackfin/macros.h

Mike Frysinger (1):
      Blackfin arch: update board defconfig files

Robin Getz (1):
      Blackfin arch: Fix up remaining printks with proper log levels

 arch/blackfin/Kconfig                       |    6 ++++++
 arch/blackfin/Makefile                      |   21 +++++++++++++++++++++
 arch/blackfin/configs/BF533-EZKIT_defconfig |   12 +++++++++++-
 arch/blackfin/configs/BF533-STAMP_defconfig |   26 ++++++++++++++++++++++----
 arch/blackfin/configs/BF537-STAMP_defconfig |   26 ++++++++++++++++++++++----
 arch/blackfin/configs/BF561-EZKIT_defconfig |   12 ++++++++++--
 arch/blackfin/configs/PNAV-10_defconfig     |   13 +++++++++++--
 arch/blackfin/kernel/setup.c                |   18 ++++++++++++++----
 arch/blackfin/kernel/traps.c                |    3 ++-
 include/asm-blackfin/processor.h            |    4 ++++
 10 files changed, 123 insertions(+), 18 deletions(-)
 delete mode 100644 include/asm-blackfin/macros.h

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Jan Dittmer]

Bryan Wu wrote:
> Hi Linus:
> 
> Marco's patch will kill the zero file git-pull error.
> Jie's patch is required because we will release our new Blackfin toolchain.

So, what is the new toolchain version?
gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:

bf533-stamp_defconfig [1]:
   CC      arch/blackfin/kernel/asm-offsets.s
cc1: error: unrecognized command line option "-mcpu=bf533-0.3"
make[2]: *** [arch/blackfin/kernel/asm-offsets.s] Error 1
make[1]: *** [prepare0] Error 2
make: *** [_all] Error 2

bf561-ezkit_defconfig [2]:
   CC      arch/blackfin/kernel/asm-offsets.s
cc1: error: unrecognized command line option "-mcpu=bf561-0.3"
make[2]: *** [arch/blackfin/kernel/asm-offsets.s] Error 1
make[1]: *** [prepare0] Error 2
make: *** [_all] Error 2

previously, there was only this error (defconfig) [3]:
CC      fs/binfmt_flat.o
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c: In function ‘decompress_exec’:
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:293: warning: label ‘out’ defined 
but not used
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c: In function ‘load_flat_file’:
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:462: warning: format ‘%x’ expects 
type ‘unsigned int’, but argument 3 has type ‘long int’
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:462: warning: format ‘%x’ expects 
type ‘unsigned int’, but argument 4 has type ‘long int’
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:549: warning: passing argument 1 
of ‘ksize’ makes pointer from integer without a cast
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:601: warning: passing argument 1 
of ‘ksize’ makes pointer from integer without a cast
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:760:50: error: macro 
"flat_get_addr_from_rp" requires 4 arguments, but only 3 given
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:760: error: 
‘flat_get_addr_from_rp’ undeclared (first use in this function)
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:760: error: (Each undeclared 
identifier is reported only once
/tmp/tmp.RAuANA4724/kernel/fs/binfmt_flat.c:760: error: for each function it 
appears in.)
make[2]: *** [fs/binfmt_flat.o] Error 1
make[1]: *** [fs] Error 2
make: *** [_all] Error 2

Thanks,

Jan

[1] http://l4x.org/k/?d=31517
[2] http://l4x.org/k/?d=31518
[3] http://l4x.org/k/?d=31478

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Mike Frysinger]

On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
> Bryan Wu wrote:
> > Jie's patch is required because we will release our new Blackfin toolchain.
>
> So, what is the new toolchain version?
> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:

we'll post new toolchain binaries in a bit
-mike

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Jan Dittmer]

Mike Frysinger wrote:
> On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
>> Bryan Wu wrote:
>>> Jie's patch is required because we will release our new Blackfin toolchain.
>> So, what is the new toolchain version?
>> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
> 
> we'll post new toolchain binaries in a bit

Hrm, somehow I don't understand it. On [1] you're
talking about gcc 4.1.2 and binutils 2.17 supporting
the -mcpu switch. But if you download the 2007R1 RC9
toolchain from the Files section of the site (tar.gz)
you actually get 4.1.1 without mcpu support. But the
version string from gcc indicates 07r1.
Care to explain?

Thanks,

Jan

ps: All in all it is a bit unfortunate to put features
    in the upstream kernel for a toolchain version just
    available from svn.


[1]
http://docs.blackfin.uclinux.org/doku.php?id=toolchain_release_notes_2007r1

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Mike Frysinger]

On 7/4/07, Jan Dittmer <jdi@l4x.org> wrote:
> Mike Frysinger wrote:
> > On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
> >> Bryan Wu wrote:
> >>> Jie's patch is required because we will release our new Blackfin toolchain.
> >> So, what is the new toolchain version?
> >> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
> >
> > we'll post new toolchain binaries in a bit
>
> Hrm, somehow I don't understand it. On [1] you're
> talking about gcc 4.1.2 and binutils 2.17 supporting
> the -mcpu switch. But if you download the 2007R1 RC9
> toolchain from the Files section of the site (tar.gz)
> you actually get 4.1.1 without mcpu support. But the
> version string from gcc indicates 07r1.
> Care to explain?

the syntax of the -mcpu option was extended

> ps: All in all it is a bit unfortunate to put features
>     in the upstream kernel for a toolchain version just
>     available from svn.

an ideal world we'd have it all be stable and done ... in reality,
we're pushing and trying to stabilize on multiple fronts in parallel
rather than stretching out development time over a much longer period
-mike

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Jan Dittmer]

Mike Frysinger wrote:
> On 7/4/07, Jan Dittmer <jdi@l4x.org> wrote:
>> Mike Frysinger wrote:
>> > On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
>> >> Bryan Wu wrote:
>> >>> Jie's patch is required because we will release our new Blackfin 
>> toolchain.
>> >> So, what is the new toolchain version?
>> >> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
>> >
>> > we'll post new toolchain binaries in a bit
>>
>> Hrm, somehow I don't understand it. On [1] you're
>> talking about gcc 4.1.2 and binutils 2.17 supporting
>> the -mcpu switch. But if you download the 2007R1 RC9
>> toolchain from the Files section of the site (tar.gz)
>> you actually get 4.1.1 without mcpu support. But the
>> version string from gcc indicates 07r1.
>> Care to explain?
> 
> the syntax of the -mcpu option was extended

But the tar-ball does _not_ contain 4.1.2

Jan

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Mike Frysinger]

On 7/6/07, Jan Dittmer <jdi@l4x.org> wrote:
> Mike Frysinger wrote:
> > On 7/4/07, Jan Dittmer <jdi@l4x.org> wrote:
> >> Mike Frysinger wrote:
> >> > On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
> >> >> Bryan Wu wrote:
> >> >>> Jie's patch is required because we will release our new Blackfin
> >> toolchain.
> >> >> So, what is the new toolchain version?
> >> >> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
> >> >
> >> > we'll post new toolchain binaries in a bit
> >>
> >> Hrm, somehow I don't understand it. On [1] you're
> >> talking about gcc 4.1.2 and binutils 2.17 supporting
> >> the -mcpu switch. But if you download the 2007R1 RC9
> >> toolchain from the Files section of the site (tar.gz)
> >> you actually get 4.1.1 without mcpu support. But the
> >> version string from gcc indicates 07r1.
> >> Care to explain?
> >
> > the syntax of the -mcpu option was extended
>
> But the tar-ball does _not_ contain 4.1.2

the version is irrelevant ... the syntax was extended and backported to 4.1.1

i put together nightly snapshots which you can find here:
http://blackfin.uclinux.org/gf/project/toolchain/frs/?action=index
-mike

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Jan Dittmer]

Mike Frysinger wrote:
> On 7/4/07, Jan Dittmer <jdi@l4x.org> wrote:
>> Mike Frysinger wrote:
>>> On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
>>>> Bryan Wu wrote:
>>>>> Jie's patch is required because we will release our new Blackfin toolchain.
>>>> So, what is the new toolchain version?
>>>> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
>>> we'll post new toolchain binaries in a bit
>> Hrm, somehow I don't understand it. On [1] you're
>> talking about gcc 4.1.2 and binutils 2.17 supporting
>> the -mcpu switch. But if you download the 2007R1 RC9
>> toolchain from the Files section of the site (tar.gz)
>> you actually get 4.1.1 without mcpu support. But the
>> version string from gcc indicates 07r1.
>> Care to explain?
> 
> the syntax of the -mcpu option was extended

With the current -elf toolchain from svn (-uclinux toolchain fails to
build) I get the following error:

  CC      fs/binfmt_script.o
  CC      fs/binfmt_elf_fdpic.o
  CC      fs/binfmt_flat.o
fs/binfmt_flat.c: In function ‘decompress_exec’:
fs/binfmt_flat.c:293: warning: label ‘out’ defined but not used
fs/binfmt_flat.c: In function ‘load_flat_file’:
fs/binfmt_flat.c:462: warning: format ‘%x’ expects type ‘unsigned int’,
but argument 3 has type ‘long int’
fs/binfmt_flat.c:462: warning: format ‘%x’ expects type ‘unsigned int’,
but argument 4 has type ‘long int’
fs/binfmt_flat.c:549: warning: passing argument 1 of ‘ksize’ makes
pointer from integer without a cast
fs/binfmt_flat.c:601: warning: passing argument 1 of ‘ksize’ makes
pointer from integer without a cast
fs/binfmt_flat.c:760:50: error: macro "flat_get_addr_from_rp" requires 4
arguments, but only 3 given
fs/binfmt_flat.c:760: error: ‘flat_get_addr_from_rp’ undeclared (first
use in this function)
fs/binfmt_flat.c:760: error: (Each undeclared identifier is reported
only once
fs/binfmt_flat.c:760: error: for each function it appears in.)
make[2]: *** [fs/binfmt_flat.o] Error 1
make[1]: *** [fs] Error 2
make: *** [_all] Error 2

Expected?

Jan

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Robin Getz]

On Tue 10 Jul 2007 14:44, Jan Dittmer pondered:
> With the current -elf toolchain from svn (-uclinux toolchain fails to
> build) 

I built from source yesterday without any issues - what error did you get?

-Robin

Re: [GIT PULL] Blackfin arch fixes (try #2)

[Bryan Wu]

On Tue, 2007-07-10 at 20:44 +0200, Jan Dittmer wrote:
> Mike Frysinger wrote:
> > On 7/4/07, Jan Dittmer <jdi@l4x.org> wrote:
> >> Mike Frysinger wrote:
> >>> On 7/3/07, Jan Dittmer <jdi@l4x.org> wrote:
> >>>> Bryan Wu wrote:
> >>>>> Jie's patch is required because we will release our new Blackfin toolchain.
> >>>> So, what is the new toolchain version?
> >>>> gcc 4.1.1 (adi 07r1) / binutils 2.17 doesn't seem to work anymore:
> >>> we'll post new toolchain binaries in a bit
> >> Hrm, somehow I don't understand it. On [1] you're
> >> talking about gcc 4.1.2 and binutils 2.17 supporting
> >> the -mcpu switch. But if you download the 2007R1 RC9
> >> toolchain from the Files section of the site (tar.gz)
> >> you actually get 4.1.1 without mcpu support. But the
> >> version string from gcc indicates 07r1.
> >> Care to explain?
> > 
> > the syntax of the -mcpu option was extended
> 
> With the current -elf toolchain from svn (-uclinux toolchain fails to
> build) I get the following error:
> 
>   CC      fs/binfmt_script.o
>   CC      fs/binfmt_elf_fdpic.o
>   CC      fs/binfmt_flat.o
> fs/binfmt_flat.c: In function ‘decompress_exec’:
> fs/binfmt_flat.c:293: warning: label ‘out’ defined but not used
> fs/binfmt_flat.c: In function ‘load_flat_file’:
> fs/binfmt_flat.c:462: warning: format ‘%x’ expects type ‘unsigned int’,
> but argument 3 has type ‘long int’
> fs/binfmt_flat.c:462: warning: format ‘%x’ expects type ‘unsigned int’,
> but argument 4 has type ‘long int’
> fs/binfmt_flat.c:549: warning: passing argument 1 of ‘ksize’ makes
> pointer from integer without a cast
> fs/binfmt_flat.c:601: warning: passing argument 1 of ‘ksize’ makes
> pointer from integer without a cast
> fs/binfmt_flat.c:760:50: error: macro "flat_get_addr_from_rp" requires 4
> arguments, but only 3 given
> fs/binfmt_flat.c:760: error: ‘flat_get_addr_from_rp’ undeclared (first
> use in this function)
> fs/binfmt_flat.c:760: error: (Each undeclared identifier is reported
> only once
> fs/binfmt_flat.c:760: error: for each function it appears in.)
> make[2]: *** [fs/binfmt_flat.o] Error 1
> make[1]: *** [fs] Error 2
> make: *** [_all] Error 2
> 
> Expected?
> 
> Jan

Yes, It should be, because a binfmt_flat patch hacking by blackfin is
not merged by mainline. Please apply the 2 patches in the SVN trunk
uclinux-dist/bfin_patch/lkml_patch/ and you can find the information
about these 2 patches in uclinux-dist/bfin_patch/lkml_patch/README.

Thanks
- Bryan Wu