Skip to contentSkip to conversation

Random IP27 Tests on 4.3.0-rc3

DISCUSSION Joshua Kinard 1 participant 1 message Oct 4, 2015 MIPS

lore
Share & tools
lore
#0 lore raw

Has anyone else tested the below various kernel modes on IP27 hardware? Same observations?

CONFIG_SGI_SN_M_MODE=y
- Boots, works, until the hardlock bug bites.

CONFIG_SGI_SN_N_MODE=y
- Appears to boot, but crashes early. This message is left on the MSC console:
2A 000: *** TLB Refill Exception on node 0
2A 000: *** EPC: 0xa80000000001d890 (0xa80000000001d890)
2A 000: *** Press ENTER to continue.

CONFIG_MAPPED_KERNEL=y
CONFIG_REPLICATE_KTEXT=y
- Does not boot, give this classic (and annoying) error:
Cannot load bootp():.
Text start 0x1c000, size 0x1965a70 doesn't fit in a FreeMemory area. Unable to execute bootp():: not enough space

CONFIG_REPLICATE_EXHANDLERS=y
- Appears to boot, but crashes early. MSC error:
2A 000: *** TLB Refill Exception on node 0
2A 000: *** EPC: 0x0 (0x0)
2A 000: *** Press ENTER to continue.

CONFIG_NUMA=y is true for all tests above, because that seemingly staves off the hardlock bug a little longer....though mdraid seems to really exacerbate the thing now.

--J