The TPM driver by default allocates a 4K transfer buffer on the stack,
which leads to lots of fun on boards with 2K or 3K stack sizes. On
RK3288 this ends up writing over random memory sections which dependent
on the memlayout of the day might contain timestamp data (no big deal)
or page tables (-> bad time).
This patch fixes the problem by reducing the buffer size to slightly
above 1K, which still seems to work as far as I can tell. There was
already some really odd code that #undef'ed this value and redefined it
with the lower number in one .c file (unfortunately not the one with the
buffer declaration), with no explanation whatsoever... I'm removing that
and just assume the smaller value will be fine for everything.
BRANCH=veyron
BUG=None
TEST=Booted Pinky and Falco.
Change-Id: I440a5662b41cbd8b7becab3113262e1140b7f763
Signed-off-by: Stefan Reinauer <reinauer@chromium.org>
Original-Commit-Id: 3d3288041b6629b7623b9d58816e782e72836b81
Original-Change-Id: Idf80f44cbfb9617c56b64a5c88ebedf7fcb4ec71
Original-Signed-off-by: Julius Werner <jwerner@chromium.org>
Original-Reviewed-on: https://chromium-review.googlesource.com/236976
Original-Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-on: http://review.coreboot.org/9481
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
e6cf3c6f78