This change is necessary as an additional 2 MB is
allocated as reserved memory for TZ_APPS exposed
in qseecom driver and we must notify the same
boundary as FDT_HIGH to uboot.
Without this change, the board gets stuck in uboot
with starting kernel message during boot up.
Change-Id: I6653d3c3dbac97cdf38a41d84d0b401d462cf5ce
Signed-off-by: speriaka <speriaka@codeaurora.org>
Using of Dcache invalidate all resulted in stack which
was dirty being invalidated. Hence the return address stored
in stack wasn't flushed and this caused issues. Reverting to
invalidate_dcache_range.
Change-Id: I382e07a3a81885ab02c9eae38196d3aa3ba8b086
Signed-off-by: Manoharan Vijaya Raghavan <mraghava@codeaurora.org>
During kernel bootup, dcache is enabled to reduce boot time.
When CONFIG_SYS_DCACHE_OFF macro is defined, dcache will not
be enabled.
Change-Id: I1830b467f2948250f5e7a676dd0f16a025e7b229
Signed-off-by: Sasirekaa Madhesu <smadhesu@codeaurora.org>
Software SPI reset mode can be trigger from 4-Byte Address mode
in flash W25Q256JW.
Change-Id: I3f8ec46c2732c07a7af0cc7331102334413067ba
Signed-off-by: Santan Kumar <santank@codeaurora.org>
Enable or disable of cache functions provided by board.
This should override the default functions, hence should not be
__weak function definition in board command files.
Change-Id: I676c1f05975c723a88a0cbd2596e907fcf089eff
Signed-off-by: Manoharan Vijaya Raghavan <mraghava@codeaurora.org>
This kw issue occurred with below message:
"Use of free memory (access) - possible".
In order to fix this issue, we have removed all
dereferenced buffer after freeing the buffer..
Change-Id: I080c45b17397333c3f897b960bdf6fca01b7a53e
Signed-off-by: Avinash Pandey <avinasv@codeaurora.org>
U-Boot now has dcache turned on. This requires cache flushing to
be performed before executing smc instrution. There are calls which
are made to TZ with huge buffer (such as kernel image for authentication).
Flushing and invalidating by address for such huge buffer will have
a performance impact, as the size of buffer is vastly bigger than
the cache size itself. Hence performing a flush of the entire cache
as opposed to performing cache flush with a start and end address.
Change-Id: If90a49d240ecf021ac9714025b6de0cce3fb9fc8
Signed-off-by: Manoharan Vijaya Raghavan <mraghava@codeaurora.org>
Since OCR value is changed,1ms delay is added to
give cards time to respond.
Change-Id: I18bddbc9d01ab2c62529c9f2065331f83b7ecca9
Signed-off-by: Antony Arun T <antothom@codeaurora.org>
Kw issues occurred with below message:
"Array 'buf'of size 256 may use index value(s) -2..0".
In order to fix this issue, error message will be
returned with array index out of index if buffer
size isless than 0 or if it increases buffer size.
Change-Id: I5a7885b0ee2fd6e9c0ac3083b28366f1ae5ceaee
Signed-off-by: Avinash Pandey <avinasv@codeaurora.org>
This change modifies the size of IMEM region during dump
collection in secured and non-secured boots.
Change-Id: If418a353ff8fa37bdad20730f9dfcc46cddd5491
Signed-off-by: Pavithra Palanisamy <pavip@codeaurora.org>
As 52Mhz clk does not have 50% dutycycle, setting
48MHz clk for mmc data transfer mode
Change-Id: Id9c0ce07fe652df7d575c5ea11f1d83eab0fb24c
Signed-off-by: Rajkumar Ayyasamy <arajkuma@codeaurora.org>
In non-secured boards the dump save size is set to
be the size of RAM buffer. Since the name field of
dumpinfo was same for both secured and non-secured
boards,the dump save size for secured board is also
set to RAM buffer size which causes data abort during
secure crashdump collection. This patch retains the
actual save size of secure board.
Change-Id: Ib26fc4730355a7f0cb131dee61983a15c1825f6f
Signed-off-by: Sasirekaa Madhesu <smadhesu@codeaurora.org>
