Quoting Andrew Morton (akpm@xxxxxxxx): > So we've probably lost the info which will tell us how the problems > started. A serial console would be nice. This is from serial console. Unfortunatly it's a portmaster2 which has no buffer. What i do is start a screen, do a telnet to the portmaster and attach the serial console. I can't go back anyfurther than this since the output of the new kernel booting replaced it. Mayby i can tell screen to have more lines of history ?! > Are all the failures due to the aic79xx driver failing in this manner? If > not then please report the different failures separately, thanks. So far, all new kernels indeed barfed with scsi errors. It somehow seems to miss a scsi event and a chain of events occur which also sometimes lead to shutdown ethernet interfaces. This is the difference in IRQ settings i talked about earlier: Linux 2.6.14-rc2-git7 (root@newsgate) (gcc 4.0.2 ) #1 1CPU irq 0: 3736032 timer irq 12: 3 irq 1: 8 i8042 irq 16: 4519576 aic79xx irq 4: 449 serial irq 17: 34901765 aic79xx, eth3 irq 9: 0 acpi irq 18: 89100231 acenic vs Linux 2.6.12-mm1 (root@newsgate) (gcc 3.3.6 ) #1 Mon Jun 20 11:13:18 CEST 2005 1CPU [newsgate.(none)] irq 0: 3822422 timer irq 12: 3 irq 1: 8 i8042 irq 16: 82787217 acenic irq 4: 353 serial irq 17: 4194166 aic79xx irq 9: 0 acpi irq 18: 28628408 aic79xx, eth3 -- Police aren't effective because of their uniforms, badges, guns, or nightsticks, they're effective because of their radios [seen on /.] - : send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
