It was originally designed such that if usbdebug_init() was called
before cbmem_initialize(), it would fetch the already-initialized
state from CBMEM. This changed when cbmem_find() behaviour changed
to require cbmem_initialize() to be called first. As a result,
ramstage had to reinitialize all of the EHCI controller and USB
endpoints on entry. This was slow, specially with AMD hardware
where one can scan USB ports to probe for the debug dongle.
For postcar and ramstage, move usbdebug entry such that it is
triggered from CBMEM_INIT_HOOK instead of console_init().
Side-effect of this is usbdebug console shows 'coreboot-xxx ...
starting...' line only for romstage.
Initialisation for usbdebug is never done in postcar. If you have
USBDEBUG_IN_ROMSTAGE=n, postcar will not have console output on
usb either.
While at it, fix also some other __PRE_RAM__ cases to ENV_ROMSTAGE
and alike.
Change-Id: If8ab973321a801abc5529f3ff68c5dccae9512ad
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/21443
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
d07f377872