2003-06-06 05:37:45

by Warren Togami

[permalink] [raw]
Subject: 2.5.70 thru bk10 amd64 compile failure

http://www.togami.com/~warren/archive/2003/opteron/kernel-2.5.70.cfg

kernel-2.5.70, 2.5.70-bk9 and 2.5.70-bk10 all fail compilation here on
my amd64 with gcc-3.2.2-10 on stock RedHat GinGin64. Please pardon me
if this is a duplicate report, I am now subscribing in order to keep a
closer eye on this list.

CHK include/linux/compile.h
UPD include/linux/compile.h
CC init/version.o
LD init/built-in.o
LD vmlinux
arch/x86_64/ia32/built-in.o(.text+0x6885): In function `vt_check':
: undefined reference to `vt_ioctl'
arch/x86_64/ia32/built-in.o(.text+0x6978): In function `do_fontx_ioctl':
: undefined reference to `fg_console'
arch/x86_64/ia32/built-in.o(.text+0x69a9): In function `do_fontx_ioctl':
: undefined reference to `con_font_op'
arch/x86_64/ia32/built-in.o(.text+0x69c2): In function `do_fontx_ioctl':
: undefined reference to `fg_console'
arch/x86_64/ia32/built-in.o(.text+0x69f3): In function `do_fontx_ioctl':
: undefined reference to `con_font_op'
arch/x86_64/ia32/built-in.o(.text+0x6ad5): In function
`do_kdfontop_ioctl':
: undefined reference to `con_font_op'
arch/x86_64/ia32/built-in.o(.text+0x6b89): In function
`do_unimap_ioctl':
: undefined reference to `fg_console'
arch/x86_64/ia32/built-in.o(.text+0x6b8e): In function
`do_unimap_ioctl':
: undefined reference to `con_set_unimap'
arch/x86_64/ia32/built-in.o(.text+0x6ba1): In function
`do_unimap_ioctl':
: undefined reference to `fg_console'
arch/x86_64/ia32/built-in.o(.text+0x6ba6): In function
`do_unimap_ioctl':
: undefined reference to `con_get_unimap'
make: *** [vmlinux] Error 1



2003-06-06 16:22:25

by Bryan O'Sullivan

[permalink] [raw]
Subject: Re: 2.5.70 thru bk10 amd64 compile failure

On Thu, 2003-06-05 at 22:50, Warren Togami wrote:

> kernel-2.5.70, 2.5.70-bk9 and 2.5.70-bk10 all fail compilation here on
> my amd64 with gcc-3.2.2-10 on stock RedHat GinGin64. Please pardon me
> if this is a duplicate report, I am now subscribing in order to keep a
> closer eye on this list.

You're compiling with CONFIG_VT turned off. Turn it on for now, as I
don't have a patch for that problem yet.

<b

2003-06-06 21:25:33

by Andi Kleen

[permalink] [raw]
Subject: Re: 2.5.70 thru bk10 amd64 compile failure

Bryan O'Sullivan <[email protected]> writes:

> On Thu, 2003-06-05 at 22:50, Warren Togami wrote:
>
> > kernel-2.5.70, 2.5.70-bk9 and 2.5.70-bk10 all fail compilation here on
> > my amd64 with gcc-3.2.2-10 on stock RedHat GinGin64. Please pardon me
> > if this is a duplicate report, I am now subscribing in order to keep a
> > closer eye on this list.
>
> You're compiling with CONFIG_VT turned off. Turn it on for now, as I
> don't have a patch for that problem yet.

See it as a feature. Compiling with CONFIG_VT off is in 99.999999999%
of all cases a mistake. On AMD64 it is more likely 100% ;-)

Seriously is quite possible that obscure configurations do not compile on amd64
(obscure is anything that is not like arch/x86_64/defconfig ;)
They are not regularly tested. Patches are welcome.

As for CONFIG_VT I think this option should really be wrapped by
CONFIG_EMBEDDED, Turning it off is near always a mistake.
Same for the keyboard drivers in the input layer at least for CONFIG_X86.
This would work with make oldconfig too from 2.4 config files
unlike the current way.

-Andi

2003-06-06 21:38:21

by Warren Togami

[permalink] [raw]
Subject: Re: 2.5.70 thru bk10 amd64 compile failure

On Fri, 2003-06-06 at 11:39, Andi Kleen wrote:
> Bryan O'Sullivan <[email protected]> writes:
>
> > On Thu, 2003-06-05 at 22:50, Warren Togami wrote:
> >
> > > kernel-2.5.70, 2.5.70-bk9 and 2.5.70-bk10 all fail compilation here on
> > > my amd64 with gcc-3.2.2-10 on stock RedHat GinGin64. Please pardon me
> > > if this is a duplicate report, I am now subscribing in order to keep a
> > > closer eye on this list.
> >
> > You're compiling with CONFIG_VT turned off. Turn it on for now, as I
> > don't have a patch for that problem yet.
>
> See it as a feature. Compiling with CONFIG_VT off is in 99.999999999%
> of all cases a mistake. On AMD64 it is more likely 100% ;-)
>
> Seriously is quite possible that obscure configurations do not compile on amd64
> (obscure is anything that is not like arch/x86_64/defconfig ;)
> They are not regularly tested. Patches are welcome.
>
> As for CONFIG_VT I think this option should really be wrapped by
> CONFIG_EMBEDDED, Turning it off is near always a mistake.
> Same for the keyboard drivers in the input layer at least for CONFIG_X86.
> This would work with make oldconfig too from 2.4 config files
> unlike the current way.
>
> -Andi

Thanks for the cluestick. I indeed made the mistake of using "make
oldconfig" from a 2.4 kernel config file. I hope that the
CONFIG_EMBEDDED thing does happen so others do not experience this
problem.

Warren