Hi All
Stomped over this one this morning; tried to boot
2.5.72 (config attached) into my 2xP3 Coppermine
933 MHz and it would get stuck just after printing
the mttr banner.
The NMI watchdog would kick in and print a dump
pointing to arch/i386/kernel/cpu/mtrr/main.c:ipi_handler:
NMI Watchdog detected LOCKUP on CPU1, eip c0113b70, registers:
CPU: 1
EIP: 0060:[<c0113b70>] Not tainted
EFLAGS: 00000082
EIP is at ipi_handler+0x60/0x80
eax: c014b2fe ebx: c2bf5f60 ecx: 000002ff edx: 00000000
esi: 00000086 edi: c2bf2000 ebp: c2bf3f5c esp: c2bf3f44
ds: 007b es: 007b ss: 0068
Process swapper (pid: 0, threadinfo=c2bf2000 task=c2bf72d0)
Stack: c0464228 00000046 c0495088 00000001 c2bf2000 00000000 c2bf3f70
c0118412
c2bf5f60 c01072a0 c2bf2000 c2bf3fac c010a50e c01072a0 00000000
c2bf2000
c2bf2000 c2bf2000 c2bf3fac 00000000 0000007b c2bf007b fffffffb
c01072d0
Call Trace:
[<c0118412>] smp_call_function_interrupt+0x42/0xa0
[<c01072a0>] default_idle+0x0/0x40
[<c010a50e>] call_function_interrupt+0x1a/0x20
[<c01072a0>] default_idle+0x0/0x40
[<c01072d0>] default_idle+0x30/0x40
[<c010736a>] cpu_idle+0x4a/0x60
[<c012532d>] printk+0x1dd/0x290
Code: f3 90 8b 43 04 85 c0 75 f7 56 9d 83 c4 10 5b 5e 5d c3 a1 44
console shuts up ...
Sooo .. decided to do the binary search thingie and tracked
it down to patch-2.5.70-bk15-bk15.gz, change set 1.28
(http://linux.bkbits.net:8080/linux-2.5/diffs/arch/i386/kernel/cpu/mtrr/main
[email protected]?nav=index.html|tags|[email protected]..|[email protected]),
where mtrr_init() is changed from being a core_initcall to be a
subsys_initcall(). Reverting that changes Fixes It For Me (tm)
both in 2.5.70-bk16 and 2.5.72. No adversities seen so far.
Please apply:
--- arch/i386/kernel/cpu/mtrr/main.c~ 2003-06-16 21:19:39.000000000 -0700
+++ arch/i386/kernel/cpu/mtrr/main.c 2003-06-19 18:52:38.000000000 -0700
@@ -701,5 +701,5 @@
"write-back", /* 6 */
};
-subsys_initcall(mtrr_init);
+core_initcall(mtrr_init);
I?aky P?rez-Gonz?lez -- Not speaking for Intel -- all opinions are my own
(and my fault)
> From: Perez-Gonzalez, Inaky [mailto:[email protected]]
>
> Sooo .. decided to do the binary search thingie and tracked
> it down to patch-2.5.70-bk15-bk15.gz, change set 1.28
>
(http://linux.bkbits.net:8080/linux-2.5/diffs/arch/i386/kernel/cpu/mtrr/main
> [email protected]?nav=index.html|tags|[email protected]..|[email protected]),
God I hate Outlook ...
http://linux.bkbits.net:8080/linux-2.5/[email protected]
And seems the cset is not 1.28 ... kind of complex this BK thing
I?aky P?rez-Gonz?lez -- Not speaking for Intel -- all opinions are my own
(and my fault)
This patch fixes my hang.
Perez-Gonzalez, Inaky wrote:
>Hi All
>
>Stomped over this one this morning; tried to boot
>2.5.72 (config attached) into my 2xP3 Coppermine
>933 MHz and it would get stuck just after printing
>the mttr banner.
>
>The NMI watchdog would kick in and print a dump
>pointing to arch/i386/kernel/cpu/mtrr/main.c:ipi_handler:
>
>NMI Watchdog detected LOCKUP on CPU1, eip c0113b70, registers:
>CPU: 1
>EIP: 0060:[<c0113b70>] Not tainted
>EFLAGS: 00000082
>EIP is at ipi_handler+0x60/0x80
>eax: c014b2fe ebx: c2bf5f60 ecx: 000002ff edx: 00000000
>esi: 00000086 edi: c2bf2000 ebp: c2bf3f5c esp: c2bf3f44
>ds: 007b es: 007b ss: 0068
>Process swapper (pid: 0, threadinfo=c2bf2000 task=c2bf72d0)
>Stack: c0464228 00000046 c0495088 00000001 c2bf2000 00000000 c2bf3f70
>c0118412
> c2bf5f60 c01072a0 c2bf2000 c2bf3fac c010a50e c01072a0 00000000
>c2bf2000
> c2bf2000 c2bf2000 c2bf3fac 00000000 0000007b c2bf007b fffffffb
>c01072d0
>Call Trace:
> [<c0118412>] smp_call_function_interrupt+0x42/0xa0
> [<c01072a0>] default_idle+0x0/0x40
> [<c010a50e>] call_function_interrupt+0x1a/0x20
> [<c01072a0>] default_idle+0x0/0x40
> [<c01072d0>] default_idle+0x30/0x40
> [<c010736a>] cpu_idle+0x4a/0x60
> [<c012532d>] printk+0x1dd/0x290
>
>Code: f3 90 8b 43 04 85 c0 75 f7 56 9d 83 c4 10 5b 5e 5d c3 a1 44
>console shuts up ...
>
>Sooo .. decided to do the binary search thingie and tracked
>it down to patch-2.5.70-bk15-bk15.gz, change set 1.28
>(http://linux.bkbits.net:8080/linux-2.5/diffs/arch/i386/kernel/cpu/mtrr/main
>[email protected]?nav=index.html|tags|[email protected]..|[email protected]),
>
>where mtrr_init() is changed from being a core_initcall to be a
>subsys_initcall(). Reverting that changes Fixes It For Me (tm)
>both in 2.5.70-bk16 and 2.5.72. No adversities seen so far.
>
>Please apply:
>
>--- arch/i386/kernel/cpu/mtrr/main.c~ 2003-06-16 21:19:39.000000000 -0700
>+++ arch/i386/kernel/cpu/mtrr/main.c 2003-06-19 18:52:38.000000000 -0700
>@@ -701,5 +701,5 @@
> "write-back", /* 6 */
> };
>
>-subsys_initcall(mtrr_init);
>+core_initcall(mtrr_init);
>
>I?aky P?rez-Gonz?lez -- Not speaking for Intel -- all opinions are my own
>(and my fault)
>
>
>
>
>
On Thu, 2003-06-19 at 19:20, Perez-Gonzalez, Inaky wrote:
> Hi All
>
> Stomped over this one this morning; tried to boot
> 2.5.72 (config attached) into my 2xP3 Coppermine
> 933 MHz and it would get stuck just after printing
> the mttr banner.
>
> The NMI watchdog would kick in and print a dump
> pointing to arch/i386/kernel/cpu/mtrr/main.c:ipi_handler:
>
> NMI Watchdog detected LOCKUP on CPU1, eip c0113b70, registers:
> CPU: 1
> EIP: 0060:[<c0113b70>] Not tainted
> EFLAGS: 00000082
> EIP is at ipi_handler+0x60/0x80
> eax: c014b2fe ebx: c2bf5f60 ecx: 000002ff edx: 00000000
> esi: 00000086 edi: c2bf2000 ebp: c2bf3f5c esp: c2bf3f44
> ds: 007b es: 007b ss: 0068
> Process swapper (pid: 0, threadinfo=c2bf2000 task=c2bf72d0)
> Stack: c0464228 00000046 c0495088 00000001 c2bf2000 00000000 c2bf3f70
> c0118412
> c2bf5f60 c01072a0 c2bf2000 c2bf3fac c010a50e c01072a0 00000000
> c2bf2000
> c2bf2000 c2bf2000 c2bf3fac 00000000 0000007b c2bf007b fffffffb
> c01072d0
> Call Trace:
> [<c0118412>] smp_call_function_interrupt+0x42/0xa0
> [<c01072a0>] default_idle+0x0/0x40
> [<c010a50e>] call_function_interrupt+0x1a/0x20
> [<c01072a0>] default_idle+0x0/0x40
> [<c01072d0>] default_idle+0x30/0x40
> [<c010736a>] cpu_idle+0x4a/0x60
> [<c012532d>] printk+0x1dd/0x290
>
> Code: f3 90 8b 43 04 85 c0 75 f7 56 9d 83 c4 10 5b 5e 5d c3 a1 44
> console shuts up ...
>
> Sooo .. decided to do the binary search thingie and tracked
> it down to patch-2.5.70-bk15-bk15.gz, change set 1.28
> (http://linux.bkbits.net:8080/linux-2.5/diffs/arch/i386/kernel/cpu/mtrr/main
> [email protected]?nav=index.html|tags|[email protected]..|[email protected]),
>
> where mtrr_init() is changed from being a core_initcall to be a
> subsys_initcall(). Reverting that changes Fixes It For Me (tm)
> both in 2.5.70-bk16 and 2.5.72. No adversities seen so far.
>
> Please apply:
>
> --- arch/i386/kernel/cpu/mtrr/main.c~ 2003-06-16 21:19:39.000000000 -0700
> +++ arch/i386/kernel/cpu/mtrr/main.c 2003-06-19 18:52:38.000000000 -0700
> @@ -701,5 +701,5 @@
> "write-back", /* 6 */
> };
>
> -subsys_initcall(mtrr_init);
> +core_initcall(mtrr_init);
This fixes a boot hang problem that I have been seeing on my three
machines.
--
Mark Haverkamp <[email protected]>