Andrew Morton wrote:
> You'll need to revert signal-race-fix.patch until the various arch guys
> catch up.
Did that, no change, looks like we have something missing in signal.c
for x86_64.
HOSTCC usr/gen_init_cpio
CPIO usr/initramfs_data.cpio
GZIP usr/initramfs_data.cpio.gz
AS usr/initramfs_data.o
LD usr/built-in.o
CC arch/x86_64/kernel/process.o
CC arch/x86_64/kernel/semaphore.o
CC arch/x86_64/kernel/signal.o
arch/x86_64/kernel/signal.c: In function `do_signal':
arch/x86_64/kernel/signal.c:426: warning: passing arg 2 of
`get_signal_to_deliver' from incompatible pointer type
arch/x86_64/kernel/signal.c:426: error: too few arguments to function
`get_signal_to_deliver'
make[1]: *** [arch/x86_64/kernel/signal.o] Error 1
make: *** [arch/x86_64/kernel] Error 2
Regards
Sid.
Andrew Morton <[email protected]> remarked:
>
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.5-rc2/2.6.5-rc2-mm4/
Hi Andrew,
On a Thinkpad T40p, 2.6.5-rc2-mm4 does not boot successfully,
whereas 2.6.5-rc2 does. All the typical things (ACPI, CPUFREQ, APIC
& IOAPIC, etc.) are built in. The -mm4 kernel config was derived
from the working config for 2.6.5-rc2.
I have to pass the nolapic kernel option to disable the local APIC
in order to successfully boot 2.6.x kernels. It appears that with
2.6.5-rc2-mm4, this option is ignored, or perhaps there's a ordering
issue with when it is checked. With -mm4, a message saying that it
is enabling the local APIC appears _before_ the "Kernel command
line: ... nolapic" line appears.
Without the -mm4 patch, the nolapic command line option is parsed
before it would have tried to reenable the local APIC. With -mm4,
it is parsed after it tries to renable it. Boom, and then it locks
on "Calibrating delay loop...".
Any relevant config options I should try with/without to help narrow
it down?
Thanks,
Dan
Dan Hopper <[email protected]> wrote:
>
> On a Thinkpad T40p, 2.6.5-rc2-mm4 does not boot successfully,
> whereas 2.6.5-rc2 does. All the typical things (ACPI, CPUFREQ, APIC
> & IOAPIC, etc.) are built in. The -mm4 kernel config was derived
> from the working config for 2.6.5-rc2.
>
> I have to pass the nolapic kernel option to disable the local APIC
> in order to successfully boot 2.6.x kernels. It appears that with
> 2.6.5-rc2-mm4, this option is ignored, or perhaps there's a ordering
> issue with when it is checked. With -mm4, a message saying that it
> is enabling the local APIC appears _before_ the "Kernel command
> line: ... nolapic" line appears.
>
> Without the -mm4 patch, the nolapic command line option is parsed
> before it would have tried to reenable the local APIC. With -mm4,
> it is parsed after it tries to renable it. Boom, and then it locks
> on "Calibrating delay loop...".
>
> Any relevant config options I should try with/without to help narrow
> it down?
Does this fix it?
---
25-akpm/arch/i386/kernel/apic.c | 2 +-
1 files changed, 1 insertion(+), 1 deletion(-)
diff -puN arch/i386/kernel/apic.c~early-param-i386-nolapic-fix arch/i386/kernel/apic.c
--- 25/arch/i386/kernel/apic.c~early-param-i386-nolapic-fix 2004-03-28 16:41:06.685335888 -0800
+++ 25-akpm/arch/i386/kernel/apic.c 2004-03-28 16:42:17.299600888 -0800
@@ -616,7 +616,7 @@ static int __init lapic_disable(char *st
clear_bit(X86_FEATURE_APIC, boot_cpu_data.x86_capability);
return 0;
}
-__setup("nolapic", lapic_disable);
+__early_param("nolapic", lapic_disable);
static int __init lapic_enable(char *str)
{
_
Andrew Morton <[email protected]> remarked:
> Dan Hopper <[email protected]> wrote:
> >
> > On a Thinkpad T40p, 2.6.5-rc2-mm4 does not boot successfully,
> > whereas 2.6.5-rc2 does. All the typical things (ACPI, CPUFREQ, APIC
> > & IOAPIC, etc.) are built in. The -mm4 kernel config was derived
> > from the working config for 2.6.5-rc2.
> >
> > I have to pass the nolapic kernel option to disable the local APIC
> > in order to successfully boot 2.6.x kernels. It appears that with
> > 2.6.5-rc2-mm4, this option is ignored, or perhaps there's a ordering
> > issue with when it is checked. With -mm4, a message saying that it
> > is enabling the local APIC appears _before_ the "Kernel command
> > line: ... nolapic" line appears.
> >
> > Without the -mm4 patch, the nolapic command line option is parsed
> > before it would have tried to reenable the local APIC. With -mm4,
> > it is parsed after it tries to renable it. Boom, and then it locks
> > on "Calibrating delay loop...".
> >
> > Any relevant config options I should try with/without to help narrow
> > it down?
>
> Does this fix it?
Yep, it sure does. Thanks!
Dan
>
> ---
>
> 25-akpm/arch/i386/kernel/apic.c | 2 +-
> 1 files changed, 1 insertion(+), 1 deletion(-)
>
> diff -puN arch/i386/kernel/apic.c~early-param-i386-nolapic-fix arch/i386/kernel/apic.c
> --- 25/arch/i386/kernel/apic.c~early-param-i386-nolapic-fix 2004-03-28 16:41:06.685335888 -0800
> +++ 25-akpm/arch/i386/kernel/apic.c 2004-03-28 16:42:17.299600888 -0800
> @@ -616,7 +616,7 @@ static int __init lapic_disable(char *st
> clear_bit(X86_FEATURE_APIC, boot_cpu_data.x86_capability);
> return 0;
> }
> -__setup("nolapic", lapic_disable);
> +__early_param("nolapic", lapic_disable);
>
> static int __init lapic_enable(char *str)
> {
>
> _
>