Skip to contentSkip to conversation

2.6.14 boots on sgi altix?

DISCUSSION Chen, Kenneth W 3 participants 3 messages Nov 18, 2005

lore
Share & tools
lore
#0 lore raw

Is there something special I need to do to boot 2.6.14 on SGI altix machine? I took 7 day old 2.6.14.2 kernel source, and did the following:

cp arch/ia64/config/sn2_defconfig .config
make oldconfig
make vmlinux

And it resulted with a bug check at very early stage of boot. The boot log says:

Linux version 2.6.14.2 (root@linux-sgi) (gcc version 3.2.3 20030502 (Red Hat Linux 3.2.3-24)) #3 SMP Thu Nov 17 19:25:56 PST 2005 EFI v1.02 by SGI: SALsystab=0x3002514a60 ACPI 2.0=0x30027a1950
Number of logical nodes in system = 16
Number of memory chunks in system = 16
SAL 2.9: SGI SN2 version 4.4
SAL Platform features: ITC_Drift
SAL: AP wakeup using external interrupt vector 0x12
No logical to physical processor mapping available
kernel BUG at arch/ia64/sn/kernel/setup.c:468!
swapper[0]: bugcheck! 0 [1]
....

2.6.12 boots just fine with the default sn2 config file.

#1 lore raw
Quoted 11 lines · Kenneth W

Is there something special I need to do to boot 2.6.14 on SGI altix machine? I took 7 day old 2.6.14.2 kernel source, and did the following:

cp arch/ia64/config/sn2_defconfig .config
make oldconfig
make vmlinux

And it resulted with a bug check at very early stage of boot. The boot log says:

Quoted 10 lines

Linux version 2.6.14.2 (root@linux-sgi) (gcc version 3.2.3 20030502 (Red Hat Linux 3.2.3-24)) #3 SMP Thu Nov 17 19:25:56 PST 2005 EFI v1.02 by SGI: SALsystab=0x3002514a60 ACPI 2.0=0x30027a1950
Number of logical nodes in system = 16
Number of memory chunks in system = 16
SAL 2.9: SGI SN2 version 4.4
SAL Platform features: ITC_Drift
SAL: AP wakeup using external interrupt vector 0x12
No logical to physical processor mapping available
kernel BUG at arch/ia64/sn/kernel/setup.c:468!
swapper[0]: bugcheck! 0 [1]

You need a newer version of SAL for kernel 2.6.14. From memory you need at least SAL 4.30, but check with your SGI support person.

#2 lore raw
In reply to Chen, Kenneth W
Quoted 3 lines · Kenneth W

Is there something special I need to do to boot 2.6.14 on SGI altix machine? I took 7 day old 2.6.14.2 kernel source, and did the following:

Give this a try. Line numbers may be slightly off - I have a different version of the tree but this _should_ apply. Untested ...

Index: linux/include/asm/sn/sn_sal.h
=================================--- linux.orig/include/asm/sn/sn_sal.h 2005-11-03 18:07:57.000000000 -0600
+++ linux/include/asm/sn/sn_sal.h 2005-11-17 22:28:22.607430600 -0600
@@ -1018,6 +1018,24 @@ ia64_sn_get_sn_info(int fc, u8 *shubtype
ret_stuff.v2 = 0;
SAL_CALL_NOLOCK(ret_stuff, SN_SAL_GET_SN_INFO, fc, 0, 0, 0, 0, 0, 0);

+/***** BEGIN HACK - temp til old proms no longer supported ********/
+ if (ret_stuff.status = SALRET_NOT_IMPLEMENTED) {
+ int nasid = get_sapicid() & 0xfff;;
+#define SH_SHUB_ID_NODES_PER_BIT_MASK 0x001f000000000000UL +#define SH_SHUB_ID_NODES_PER_BIT_SHFT 48
+ if (shubtype) *shubtype = 0;
+ if (nasid_bitmask) *nasid_bitmask = 0x7ff;
+ if (nasid_shift) *nasid_shift = 38;
+ if (systemsize) *systemsize = 10;
+ if (sharing_domain_size) *sharing_domain_size = 8;
+ if (partid) *partid = sn_partition_serial_number;
+ if (coher) *coher = nasid >> 9;
+ if (reg) *reg = (HUB_L((u64 *) LOCAL_MMR_ADDR(SH1_SHUB_ID)) & SH_SHUB_ID_NODES_PER_BIT_MASK) >>
+ SH_SHUB_ID_NODES_PER_BIT_SHFT;
+ return 0;
+ }
+/***** END HACK *******/
+
if (ret_stuff.status < 0)
return ret_stuff.status;

Signature · 4 lines
Thanks

Jack Steiner ([email protected])          651-683-5302
Principal Engineer                      SGI - Silicon Graphics, Inc.