Ok - The problem is the module_init entry point for mptbase.ko is not being called when the driver is statically compiled into the kernel. Maybe when I moved registering the pci ids to the scsi llds(e.g. mptspi and mptfc) that mptbase.ko doesn't have its module_init called. The panic in mpt_interrupt occurs when the ISR is calling the callback handlers when cb_idx is 0xFF, and MptCallbacks is pointing to NULL. The cb_idx would of been initialized to something less than 0xFF had the module_init entry point been called. I guess I will have to have the scsi lld drivers registering the pci ids call mptbase.c module_init function, and maintain a flag so its not called more that once. Eric Moore On Thursday, April 28, 2005 3:21 AM, Andrew Morton wrote: > > Andrew Morton <akpm@xxxxxxxx> wrote: > > > > I guess I'll test this on x86 next. > > It crashed, strangely. See > http://www.zip.com.au/~akpm/linux/patches/stuff/dsc02504.jpg. > > It appears to have jumped directly to address zero from > handle_IRQ_event. > Disabling CONFIG_PM didn't help. No shared IRQs on this box. > > Config is at http://www.zip.com.au/~akpm/linux/patches/stuff/config > - : 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
