Skip to contentSkip to conversation

3.2-rcX regression: boot failure on AMD K6-III/450

BUG Bob Tracy 2 participants 4 messages Nov 26, 2011 – Nov 28, 2011 x86

lore
Share & tools
lore

Bob Tracy

author
#0 lore raw

Verified for 3.2-rc[1-3]. Subject system using LILO as bootloader. The following messages appear on the console:

Loading linux-dev.
BIOS data check successful
Probing EDD (edd=off to disable)... ok
<blank line>
Decompressing Linux... Parsing ELF... done.
Booting the kernel.
<hang here -- no further output>

The console output is normal and expected up to the point of the hang. Versions 3.1 and prior work fine. I had earlier thought the floppy regression Pavel reported applied to my case (and it may), but I don't seem to be getting that far in the boot process.

--Bob

#1 lore raw

Hi,

Quoted 11 lines · Bob Tracy

Verified for 3.2-rc[1-3]. Subject system using LILO as bootloader. The following messages appear on the console:

Loading linux-dev.
BIOS data check successful
Probing EDD (edd=off to disable)... ok
<blank line>
Decompressing Linux... Parsing ELF... done.
Booting the kernel.
<hang here -- no further output>

Quoted 4 lines

The console output is normal and expected up to the point of the hang. Versions 3.1 and prior work fine. I had earlier thought the floppy regression Pavel reported applied to my case (and it may), but I don't seem to be getting that far in the boot process.

can you try the following patch:

--
From: Borislav Petkov <[email protected]>
Date: Wed, 9 Nov 2011 18:08:40 +0100
Subject: [PATCH] x86, microcode, AMD: Restrict microcode reporting

bcb80e53877c ("x86, microcode, AMD: Add microcode revision to /proc/cpuinfo") added support for dumping microcode patch level in /proc/cpuinfo on AMD. Although we were very cautious and did rdmsr_safe the microcode patch level MSR, this broke booting on Geode LX because this CPU doesn't have that MSR and at the time we do the read, exception tables which are used by rdmsr_safe are not yet initialized, leading to the boot hang.

This is the simple fix which enables the microcode line in /proc/cpuinfo for K8 and newer (it is not that interesting for older boxes anyway :-)).

---
 arch/x86/kernel/cpu/amd.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
arch/x86/kernel/cpu/amd.c +2 -1
index c7e46cb..a72d183 100644
@@ -474,7 +474,8 @@ static void __cpuinit early_init_amd(struct cpuinfo_x86 *c)
 	}
 #endif
 
-	rdmsr_safe(MSR_AMD64_PATCH_LEVEL, &c->microcode, &dummy);
+	if (c->x86 >= 0xf)
+		rdmsr_safe(MSR_AMD64_PATCH_LEVEL, &c->microcode, &dummy);
 }
 
 static void __cpuinit init_amd(struct cpuinfo_x86 *c)
Signature · 13 lines
1.7.8.rc0

Thanks.

-- 
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551
author
#2 lore raw
Quoted 10 lines · Borislav Petkov

can you try the following patch:
(...)
bcb80e53877c ("x86, microcode, AMD: Add microcode revision to /proc/cpuinfo") added support for dumping microcode patch level in /proc/cpuinfo on AMD. Although we were very cautious and did rdmsr_safe the microcode patch level MSR, this broke booting on Geode LX because this CPU doesn't have that MSR and at the time we do the read, exception tables which are used by rdmsr_safe are not yet initialized, leading to the boot hang.

Quoted 6 lines

This is the simple fix which enables the microcode line in /proc/cpuinfo for K8 and newer (it is not that interesting for older boxes anyway :-)).

That fixed it. Thank you!

--Bob

#3 lore raw
Quoted 1 line · Bob Tracy

That fixed it. Thank you!

Ok, thanks for testing.

Signature · 8 lines
Regards/Gruss,
Boris.

Advanced Micro Devices GmbH
Einsteinring 24, 85609 Dornach
GM: Alberto Bozzo
Reg: Dornach, Landkreis Muenchen
HRB Nr. 43632 WEEE Registernr: 129 19551