2.6.24 still doesn't boot on i486..

2.6.24 still doesn't boot on i486..

[Michael Tokarev]

Just a follow-up on my earlier bugreport, when
2.6.23 first come out....

2.6.24 seems to not boot on i486-class hardware.
The machine reboots right after "Booting linux"
message.  2.6.22 was the last worked kernel.

The relevant config entries are:

CONFIG_TICK_ONESHOT=y
CONFIG_NO_HZ=y
CONFIG_HIGH_RES_TIMERS=y
CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
CONFIG_X86_PC=y
CONFIG_M486=y
CONFIG_X86_GENERIC=y
CONFIG_X86_CMPXCHG=y          <-- is it right?
CONFIG_X86_L1_CACHE_SHIFT=7
CONFIG_X86_XADD=y
CONFIG_X86_PPRO_FENCE=y
CONFIG_X86_F00F_BUG=y
CONFIG_X86_WP_WORKS_OK=y
CONFIG_X86_INVLPG=y
CONFIG_X86_BSWAP=y
CONFIG_X86_POPAD_OK=y
CONFIG_X86_ALIGNMENT_16=y
CONFIG_X86_INTEL_USERCOPY=y
CONFIG_X86_MINIMUM_CPU_FAMILY=4
CONFIG_X86_UP_APIC=y
CONFIG_X86_UP_IOAPIC=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y
CONFIG_X86_MCE=y
CONFIG_X86_MCE_NONFATAL=y
CONFIG_X86_MCE_P4THERMAL=y
CONFIG_NOHIGHMEM=y
CONFIG_VMSPLIT_3G=y

Compiled with gcc (GCC) 4.2.3 (Debian 4.2.3-2).

This email is also a JFYI-style email.  Yesterday
I tried to debug this issue but discovered that the
power unit of that machine failed.  Well, it still
works and the machine boots, but at the same time
it smells like frosted/fried plastic ;)  So I don't
have i486 hardware anymore to test things on...

And now I wonder if I'm the only person in this world
with i486 machine... ;)

Thanks!

/mjt

Re: 2.6.24 still doesn't boot on i486..

[Mikael Pettersson]

Michael Tokarev writes:
 > Just a follow-up on my earlier bugreport, when
 > 2.6.23 first come out....
 > 
 > 2.6.24 seems to not boot on i486-class hardware.
 > The machine reboots right after "Booting linux"
 > message.  2.6.22 was the last worked kernel.
 > 
 > The relevant config entries are:
 > 
 > CONFIG_TICK_ONESHOT=y
 > CONFIG_NO_HZ=y
 > CONFIG_HIGH_RES_TIMERS=y
 > CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
 > CONFIG_X86_PC=y
 > CONFIG_M486=y
 > CONFIG_X86_GENERIC=y
 > CONFIG_X86_CMPXCHG=y          <-- is it right?

yes

 > CONFIG_X86_L1_CACHE_SHIFT=7
 > CONFIG_X86_XADD=y
 > CONFIG_X86_PPRO_FENCE=y
 > CONFIG_X86_F00F_BUG=y
 > CONFIG_X86_WP_WORKS_OK=y
 > CONFIG_X86_INVLPG=y
 > CONFIG_X86_BSWAP=y
 > CONFIG_X86_POPAD_OK=y
 > CONFIG_X86_ALIGNMENT_16=y
 > CONFIG_X86_INTEL_USERCOPY=y
 > CONFIG_X86_MINIMUM_CPU_FAMILY=4
 > CONFIG_X86_UP_APIC=y
 > CONFIG_X86_UP_IOAPIC=y
 > CONFIG_X86_LOCAL_APIC=y
 > CONFIG_X86_IO_APIC=y
 > CONFIG_X86_MCE=y
 > CONFIG_X86_MCE_NONFATAL=y
 > CONFIG_X86_MCE_P4THERMAL=y
 > CONFIG_NOHIGHMEM=y
 > CONFIG_VMSPLIT_3G=y
 > 
 > Compiled with gcc (GCC) 4.2.3 (Debian 4.2.3-2).
 > 
 > This email is also a JFYI-style email.  Yesterday
 > I tried to debug this issue but discovered that the
 > power unit of that machine failed.  Well, it still
 > works and the machine boots, but at the same time
 > it smells like frosted/fried plastic ;)  So I don't
 > have i486 hardware anymore to test things on...
 > 
 > And now I wonder if I'm the only person in this world
 > with i486 machine... ;)

Just FYI, 2.6.23, 2.6.24 and 2.6.25-rc6 work fine on my i486DX4.

However, it's clear from your .config fragment above that you've
enabled features that cannot possibly exist on a 486 (APICs and
MCE for starters) and there may be bugs related to that. I suggest
trimming your .config and making it 486-specific.

Re: 2.6.24 still doesn't boot on i486..

[Ingo Molnar]

* Mikael Pettersson <mikpe@it.uu.se> wrote:

>  > And now I wonder if I'm the only person in this world with i486 
>  > machine... ;)
> 
> Just FYI, 2.6.23, 2.6.24 and 2.6.25-rc6 work fine on my i486DX4.
> 
> However, it's clear from your .config fragment above that you've 
> enabled features that cannot possibly exist on a 486 (APICs and MCE 
> for starters) and there may be bugs related to that. I suggest 
> trimming your .config and making it 486-specific.

apics should be auto-detect though. So if a .24 config was plugged into 
.25 and the kernel broke that's still a regression - would be nice to 
figure out which .config detail causes this problem.

	Ingo

Re: 2.6.24 still doesn't boot on i486..

[Lennart Sorensen]

On Fri, Apr 04, 2008 at 01:01:30PM +0400, Michael Tokarev wrote:
> Just a follow-up on my earlier bugreport, when
> 2.6.23 first come out....
> 
> 2.6.24 seems to not boot on i486-class hardware.
> The machine reboots right after "Booting linux"
> message.  2.6.22 was the last worked kernel.
> 
> The relevant config entries are:
> 
> CONFIG_TICK_ONESHOT=y
> CONFIG_NO_HZ=y
> CONFIG_HIGH_RES_TIMERS=y
> CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
> CONFIG_X86_PC=y
> CONFIG_M486=y
> CONFIG_X86_GENERIC=y
> CONFIG_X86_CMPXCHG=y          <-- is it right?
> CONFIG_X86_L1_CACHE_SHIFT=7
> CONFIG_X86_XADD=y
> CONFIG_X86_PPRO_FENCE=y
> CONFIG_X86_F00F_BUG=y
> CONFIG_X86_WP_WORKS_OK=y
> CONFIG_X86_INVLPG=y
> CONFIG_X86_BSWAP=y
> CONFIG_X86_POPAD_OK=y
> CONFIG_X86_ALIGNMENT_16=y
> CONFIG_X86_INTEL_USERCOPY=y
> CONFIG_X86_MINIMUM_CPU_FAMILY=4
> CONFIG_X86_UP_APIC=y
> CONFIG_X86_UP_IOAPIC=y
> CONFIG_X86_LOCAL_APIC=y
> CONFIG_X86_IO_APIC=y
> CONFIG_X86_MCE=y
> CONFIG_X86_MCE_NONFATAL=y
> CONFIG_X86_MCE_P4THERMAL=y
> CONFIG_NOHIGHMEM=y
> CONFIG_VMSPLIT_3G=y
> 
> Compiled with gcc (GCC) 4.2.3 (Debian 4.2.3-2).
> 
> This email is also a JFYI-style email.  Yesterday
> I tried to debug this issue but discovered that the
> power unit of that machine failed.  Well, it still
> works and the machine boots, but at the same time
> it smells like frosted/fried plastic ;)  So I don't
> have i486 hardware anymore to test things on...
> 
> And now I wonder if I'm the only person in this world
> with i486 machine... ;)

I have one that works flawlessly. :)  15 years of mostly 24/7 operation
and still going.  The last 9 years has been as a
firewall/router/fileserver.

-- 
Len Sorensen

Re: 2.6.24 still doesn't boot on i486..

[Lennart Sorensen]

On Fri, Apr 04, 2008 at 02:40:20PM +0200, Mikael Pettersson wrote:
> Michael Tokarev writes:
>  > Just a follow-up on my earlier bugreport, when
>  > 2.6.23 first come out....
>  > 
>  > 2.6.24 seems to not boot on i486-class hardware.
>  > The machine reboots right after "Booting linux"
>  > message.  2.6.22 was the last worked kernel.
>  > 
>  > The relevant config entries are:
>  > 
>  > CONFIG_TICK_ONESHOT=y
>  > CONFIG_NO_HZ=y
>  > CONFIG_HIGH_RES_TIMERS=y
>  > CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
>  > CONFIG_X86_PC=y
>  > CONFIG_M486=y
>  > CONFIG_X86_GENERIC=y
>  > CONFIG_X86_CMPXCHG=y          <-- is it right?
> 
> yes
> 
>  > CONFIG_X86_L1_CACHE_SHIFT=7
>  > CONFIG_X86_XADD=y
>  > CONFIG_X86_PPRO_FENCE=y
>  > CONFIG_X86_F00F_BUG=y
>  > CONFIG_X86_WP_WORKS_OK=y
>  > CONFIG_X86_INVLPG=y
>  > CONFIG_X86_BSWAP=y
>  > CONFIG_X86_POPAD_OK=y
>  > CONFIG_X86_ALIGNMENT_16=y
>  > CONFIG_X86_INTEL_USERCOPY=y
>  > CONFIG_X86_MINIMUM_CPU_FAMILY=4
>  > CONFIG_X86_UP_APIC=y
>  > CONFIG_X86_UP_IOAPIC=y
>  > CONFIG_X86_LOCAL_APIC=y
>  > CONFIG_X86_IO_APIC=y
>  > CONFIG_X86_MCE=y
>  > CONFIG_X86_MCE_NONFATAL=y
>  > CONFIG_X86_MCE_P4THERMAL=y
>  > CONFIG_NOHIGHMEM=y
>  > CONFIG_VMSPLIT_3G=y
>  > 
>  > Compiled with gcc (GCC) 4.2.3 (Debian 4.2.3-2).
>  > 
>  > This email is also a JFYI-style email.  Yesterday
>  > I tried to debug this issue but discovered that the
>  > power unit of that machine failed.  Well, it still
>  > works and the machine boots, but at the same time
>  > it smells like frosted/fried plastic ;)  So I don't
>  > have i486 hardware anymore to test things on...
>  > 
>  > And now I wonder if I'm the only person in this world
>  > with i486 machine... ;)
> 
> Just FYI, 2.6.23, 2.6.24 and 2.6.25-rc6 work fine on my i486DX4.

A DX4 has many features a DX2 does not.  I will try upgrading from
2.6.18 to 2.6.24 on my DX2/66.  It doesn't have any of the CPUID and
other clever enhancements that newer chips have.

> However, it's clear from your .config fragment above that you've
> enabled features that cannot possibly exist on a 486 (APICs and
> MCE for starters) and there may be bugs related to that. I suggest
> trimming your .config and making it 486-specific.

Doesn't APIC and MCE just do nothing if the system doesn't have them?  I
thought normally it did.

-- 
Len Sorensen