Re: Build regressions/improvements in v4.0-rc1

Re: Build regressions/improvements in v4.0-rc1

[Geert Uytterhoeven]

On Mon, Feb 23, 2015 at 9:33 AM, Geert Uytterhoeven
<geert@linux-m68k.org> wrote:
> Below is the list of build error/warning regressions/improvements in
> v4.0-rc1[1] compared to v3.19[2].
>
> Summarized:
>   - build errors: +11/-10

> [1] http://kisskb.ellerman.id.au/kisskb/head/8494/ (256 out of 257 configs)
> [2] http://kisskb.ellerman.id.au/kisskb/head/8427/ (255 out of 257 configs)
>
>
> *** ERRORS ***
>
> 11 regressions:
>   + /home/kisskb/slave/src/arch/arm/include/asm/atomic.h: error: #error SMP not supported on pre-ARMv6 CPUs:  => 137:2
>   + /home/kisskb/slave/src/arch/arm/include/asm/cmpxchg.h: error: #error "SMP is not supported on this platform":  => 114:2

arm-randconfig

SMP does depend on CPU_V6K || CPU_V7 and on HAVE_SMP... strange...

>   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2

sh-randconfig

>   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'devm_gpiod_get' [-Werror=implicit-function-declaration]:  => 55:2
>   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_direction_output' [-Werror=implicit-function-declaration]:  => 61:2
>   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_set_value' [-Werror=implicit-function-declaration]:  => 30:3

s390-allyesconfig
s390-allmodconfig
parisc-allmodconfig
sh-allmodconfig
sh-allyesconfig
cris-allyesconfig
cris-allmodconfig
mips-allmodconfig

>   + error: asm9260_timer.c: undefined reference to `of_io_request_and_map':  => .init.text+0x4014), .init.text+0x5c84)

sparc64-allmodconfig
sparc-allmodconfig

>   + error: book3s_64_vio_hv.c: undefined reference to `power7_wakeup_loss':  => .text+0x408)

pseries_defconfig

>   + error: smp.c: undefined reference to `.__flush_disable_L1':  => .text+0xb24)

powerpc-randconfig

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Build regressions/improvements in v4.0-rc1

[Arnd Bergmann]

On Monday 23 February 2015 09:45:43 Geert Uytterhoeven wrote:
> On Mon, Feb 23, 2015 at 9:33 AM, Geert Uytterhoeven
> <geert@linux-m68k.org> wrote:
> > Below is the list of build error/warning regressions/improvements in
> > v4.0-rc1[1] compared to v3.19[2].
> >
> > Summarized:
> >   - build errors: +11/-10
> 
> > [1] http://kisskb.ellerman.id.au/kisskb/head/8494/ (256 out of 257 configs)
> > [2] http://kisskb.ellerman.id.au/kisskb/head/8427/ (255 out of 257 configs)
> >
> >
> > *** ERRORS ***
> >
> > 11 regressions:
> >   + /home/kisskb/slave/src/arch/arm/include/asm/atomic.h: error: #error SMP not supported on pre-ARMv6 CPUs:  => 137:2
> >   + /home/kisskb/slave/src/arch/arm/include/asm/cmpxchg.h: error: #error "SMP is not supported on this platform":  => 114:2
> 
> arm-randconfig
> 
> SMP does depend on CPU_V6K || CPU_V7 and on HAVE_SMP... strange...

V6/V7 and  the earlier V4/V5 are normally mutually exclusive, but we sometimes
screw up the dependencies for some platform, in which case you can end up
with a config that includes both and gives the above error. I'll have a look.

I thought I'd fixed all the bugs we introduced in the merge window in yesterday's
fixes branch.

There is one known problem for mach-realview in this area, for which
I have a fix.

> >   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2

DAX support was merged with 'depends on !(ARM || MIPS || SPARC)', so I don't
see how this could happen.

> sh-randconfig
> 
> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'devm_gpiod_get' [-Werror=implicit-function-declaration]:  => 55:2
> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_direction_output' [-Werror=implicit-function-declaration]:  => 61:2
> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_set_value' [-Werror=implicit-function-declaration]:  => 30:3

Something wrong in arch/sh/include/asm/gpio.h? the codec is only
selectable when gpiolib is enabled, but the consumer.h header
apparently does not get included here.

> s390-allyesconfig
> s390-allmodconfig
> parisc-allmodconfig
> sh-allmodconfig
> sh-allyesconfig
> cris-allyesconfig
> cris-allmodconfig
> mips-allmodconfig
> 
> >   + error: asm9260_timer.c: undefined reference to `of_io_request_and_map':  => .init.text+0x4014), .init.text+0x5c84)

asm9260 is arm-specific, we probably need to refine the dependencies here
to ensure that OF_ADDRESS is set. Need to think about what the best set of
dependencies is.

	Arnd

Re: Build regressions/improvements in v4.0-rc1

[Geert Uytterhoeven]

On Mon, Feb 23, 2015 at 10:03 AM, Arnd Bergmann <arnd@arndb.de> wrote:
>> >   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2
>
> DAX support was merged with 'depends on !(ARM || MIPS || SPARC)', so I don't
> see how this could happen.

sh is not ARM, not MIPS, not SPARC, so...?

>> sh-randconfig

>> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'devm_gpiod_get' [-Werror=implicit-function-declaration]:  => 55:2
>> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_direction_output' [-Werror=implicit-function-declaration]:  => 61:2
>> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_set_value' [-Werror=implicit-function-declaration]:  => 30:3
>
> Something wrong in arch/sh/include/asm/gpio.h? the codec is only
> selectable when gpiolib is enabled, but the consumer.h header
> apparently does not get included here.

I think it's an bad select, for which a patch has been available since
about 2 weeks.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Build regressions/improvements in v4.0-rc1

[Arnd Bergmann]

On Monday 23 February 2015 10:14:55 Geert Uytterhoeven wrote:
> On Mon, Feb 23, 2015 at 10:03 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> >> >   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2
> >
> > DAX support was merged with 'depends on !(ARM || MIPS || SPARC)', so I don't
> > see how this could happen.
> 
> sh is not ARM, not MIPS, not SPARC, so...?

I misread your email and thought this was still under arm-randconfig.

I checked the other architectures, and found that the same is missing on
ARC, so we should add both:

diff --git a/fs/Kconfig b/fs/Kconfig
index ec35851e5b71..a24d496787d6 100644
--- a/fs/Kconfig
+++ b/fs/Kconfig
@@ -36,7 +36,7 @@ source "fs/nilfs2/Kconfig"
 config FS_DAX
 	bool "Direct Access (DAX) support"
 	depends on MMU
-	depends on !(ARM || MIPS || SPARC)
+	depends on !(ARC || ARM || MIPS || SH || SPARC)
 	help
 	  Direct Access (DAX) can be used on memory-backed block devices.
 	  If the block device supports DAX and the filesystem supports DAX,


It's also not implemented on c6x and blackfin, but both of them have
MMU=n, so they are already excluded.

> >> sh-randconfig
> 
> >> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'devm_gpiod_get' [-Werror=implicit-function-declaration]:  => 55:2
> >> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_direction_output' [-Werror=implicit-function-declaration]:  => 61:2
> >> >   + /home/kisskb/slave/src/sound/soc/codecs/max98357a.c: error: implicit declaration of function 'gpiod_set_value' [-Werror=implicit-function-declaration]:  => 30:3
> >
> > Something wrong in arch/sh/include/asm/gpio.h? the codec is only
> > selectable when gpiolib is enabled, but the consumer.h header
> > apparently does not get included here.
> 
> I think it's an bad select, for which a patch has been available since
> about 2 weeks.

Ah, right. I was looking at linux-next, which has the fix, so I did not
see the problem at first.

	Arnd

Re: Build regressions/improvements in v4.0-rc1

[Geert Uytterhoeven]

On Mon, Feb 23, 2015 at 11:24 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Monday 23 February 2015 10:14:55 Geert Uytterhoeven wrote:
>> On Mon, Feb 23, 2015 at 10:03 AM, Arnd Bergmann <arnd@arndb.de> wrote:
>> >> >   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2

> I checked the other architectures, and found that the same is missing on
> ARC, so we should add both:
>
> diff --git a/fs/Kconfig b/fs/Kconfig
> index ec35851e5b71..a24d496787d6 100644
> --- a/fs/Kconfig
> +++ b/fs/Kconfig
> @@ -36,7 +36,7 @@ source "fs/nilfs2/Kconfig"
>  config FS_DAX
>         bool "Direct Access (DAX) support"
>         depends on MMU
> -       depends on !(ARM || MIPS || SPARC)
> +       depends on !(ARC || ARM || MIPS || SH || SPARC)
>         help
>           Direct Access (DAX) can be used on memory-backed block devices.
>           If the block device supports DAX and the filesystem supports DAX,
>
>
> It's also not implemented on c6x and blackfin, but both of them have
> MMU=n, so they are already excluded.

Exactly.

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Build regressions/improvements in v4.0-rc1

[Arnd Bergmann]

On Monday 23 February 2015 11:28:12 Geert Uytterhoeven wrote:
> On Mon, Feb 23, 2015 at 11:24 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> > On Monday 23 February 2015 10:14:55 Geert Uytterhoeven wrote:
> >> On Mon, Feb 23, 2015 at 10:03 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> >> >> >   + /home/kisskb/slave/src/fs/dax.c: error: implicit declaration of function 'copy_user_page' [-Werror=implicit-function-declaration]:  => 266:2
> 
> > I checked the other architectures, and found that the same is missing on
> > ARC, so we should add both:
> >
> > diff --git a/fs/Kconfig b/fs/Kconfig
> > index ec35851e5b71..a24d496787d6 100644
> > --- a/fs/Kconfig
> > +++ b/fs/Kconfig
> > @@ -36,7 +36,7 @@ source "fs/nilfs2/Kconfig"
> >  config FS_DAX
> >         bool "Direct Access (DAX) support"
> >         depends on MMU
> > -       depends on !(ARM || MIPS || SPARC)
> > +       depends on !(ARC || ARM || MIPS || SH || SPARC)
> >         help
> >           Direct Access (DAX) can be used on memory-backed block devices.
> >           If the block device supports DAX and the filesystem supports DAX,
> >
> >
> > It's also not implemented on c6x and blackfin, but both of them have
> > MMU=n, so they are already excluded.
> 
> Exactly.
> 
> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Thanks, let me submit that as a proper patch with a changelog then.

	Arnd

Re: Build regressions/improvements in v4.0-rc1

[Michael Ellerman]

On Mon, 2015-02-23 at 09:45 +0100, Geert Uytterhoeven wrote:
> On Mon, Feb 23, 2015 at 9:33 AM, Geert Uytterhoeven
> <geert@linux-m68k.org> wrote:
> > Below is the list of build error/warning regressions/improvements in
> > v4.0-rc1[1] compared to v3.19[2].
> >
> > Summarized:
> >   - build errors: +11/-10
> 
> > [1] http://kisskb.ellerman.id.au/kisskb/head/8494/ (256 out of 257 configs)
> > [2] http://kisskb.ellerman.id.au/kisskb/head/8427/ (255 out of 257 configs)
 
...

> >   + error: book3s_64_vio_hv.c: undefined reference to `power7_wakeup_loss':  => .text+0x408)
> 
> pseries_defconfig

This one is actually from pseries_defconfig+POWERNV=n, so I think I
broke your script with the + notation in the config name :)

http://kisskb.ellerman.id.au/kisskb/buildresult/12372031/

cheers

Re: Build regressions/improvements in v4.0-rc1

[Geert Uytterhoeven]

Hi Michael,

On Tue, Feb 24, 2015 at 5:52 AM, Michael Ellerman <mpe@ellerman.id.au> wrote:
>> >   + error: book3s_64_vio_hv.c: undefined reference to `power7_wakeup_loss':  => .text+0x408)
>>
>> pseries_defconfig
>
> This one is actually from pseries_defconfig+POWERNV=n, so I think I

Thanks!

> broke your script with the + notation in the config name :)

Nope, my brain used the wrong separator.

However, my scripts do have a problem with the subdirectories
in arch/powerpc/configs/ (4xx/currituck_defconfig)...

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

Re: Build regressions/improvements in v4.0-rc1

[Michael Ellerman]

On Tue, 2015-02-24 at 10:38 +0100, Geert Uytterhoeven wrote:
> Hi Michael,
> 
> On Tue, Feb 24, 2015 at 5:52 AM, Michael Ellerman <mpe@ellerman.id.au> wrote:
> >> >   + error: book3s_64_vio_hv.c: undefined reference to `power7_wakeup_loss':  => .text+0x408)
> >>
> >> pseries_defconfig
> >
> > This one is actually from pseries_defconfig+POWERNV=n, so I think I
> 
> Thanks!
> 
> > broke your script with the + notation in the config name :)
> 
> Nope, my brain used the wrong separator.

I can't help with that :)

> However, my scripts do have a problem with the subdirectories
> in arch/powerpc/configs/ (4xx/currituck_defconfig)...

Yeah sorry, they are a bit of a pain. I'm sure some horrible regexp can deal
with it ;)

cheers