-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hello,
I still did not find time to fix the cbmem stuff to work on K8, making resume always fail. It is broken since stephan checked in the cbmem stuff.
Also, because kconfig has RAMBASE and RAMTOP on 1M - 2M adding the e820 reserved region when HAVE_ACPI_RESUME is there forces the grub to fail with "inconsistent filesystem" - it took me 3hours to find this out. It was 31M-32M hole for suspend/resume without memory backup. But for some reason this change was not correctly copied to kconfig, maybe the HAVE_MOTHERBOARD_RESOURCES was missing too making the mistake hard to find ;)
Anyway, disabling the suspend/resume now. Its broken. Either cbmem stuff needs to be fixed, or old functionality for suspend/resume with hole must be restored, and for this board a hole must reside on 31MB or above.
Signed-off-by: Rudolf Marek r.marek@assembler.cz
Thanks Rudolf
On 2/21/10 11:08 PM, Rudolf Marek wrote:
Hello,
I still did not find time to fix the cbmem stuff to work on K8, making resume always fail. It is broken since stephan checked in the cbmem stuff.
Also, because kconfig has RAMBASE and RAMTOP on 1M - 2M adding the e820 reserved region when HAVE_ACPI_RESUME is there forces the grub to fail with "inconsistent filesystem" - it took me 3hours to find this out. It was 31M-32M hole for suspend/resume without memory backup. But for some reason this change was not correctly copied to kconfig, maybe the HAVE_MOTHERBOARD_RESOURCES was missing too making the mistake hard to find ;)
Anyway, disabling the suspend/resume now. Its broken. Either cbmem stuff needs to be fixed, or old functionality for suspend/resume with hole must be restored, and for this board a hole must reside on 31MB or above.
Signed-off-by: Rudolf Marek r.marek@assembler.cz
Sorry for leaving this broken for so long... can't fix it, I don't have the hardware and it's not sold anymore....
Acked-by: Stefan Reinauer stepan@coresystems.de
Am 21.02.2010 23:08, schrieb Rudolf Marek:
Anyway, disabling the suspend/resume now. Its broken. Either cbmem stuff needs to be fixed, or old functionality for suspend/resume with hole must be restored,
Could you go to some detail as how cbmem must be fixed for your board? I'd prefer that over special solutions for single boards.
Signed-off-by: Rudolf Marek r.marek@assembler.cz
Committed in r5148.
Thanks, Patrick