Hi,
the attached patch sets up the stack pointer in the c7 CAR-post routine, to make function calls work in the small area after CAR and before coreboot_ram. For sufficient high loglevels, there are two printk_*s in generic CAR code that made the system reboot. (with my preceding patch, it crashes for all loglevel configurations and an issue like this one would have been found directly)
Signed-off-by: Patrick Georgi patrick.georgi@coresystems.de
the attached patch sets up the stack pointer in the c7 CAR-post routine, to make function calls work in the small area after CAR and before coreboot_ram. For sufficient high loglevels, there are two printk_*s in generic CAR code that made the system reboot. (with my preceding patch, it crashes for all loglevel configurations and an issue like this one would have been found directly)
Signed-off-by: Patrick Georgi patrick.georgi@coresystems.de
The stack pointer is a magic value to me. Could you add a comment stating why you chose 0x400000? Maybe that will help someone decide whether it needs to change later for suspend-to-RAM.
Acked-by: Myles Watson mylesgw@gmail.com
Thanks, Myles
Hi,
For suspend/resume one would need to satisfy:
Either do not corrupt any memory or corrupt memory which is between _RAMBASE -MEM_TOPK. Now it is assumed that the CAR does not get flushed back to memory.
Rudolf
-
Myles Watson -
Patrick Georgi -
Rudolf Marek