In order for the cpu code to start SMM relocation 2 new
functions are added to be shared:
- void smm_initiate_relocation_parallel()
- void smm_initiate_relocation()
The both initiate an SMI on the currently running cpu.
The 2 variants allow for parallel relocation or serialized
relocation.
BUG=chrome-os-partner:22862
BRANCH=None
TEST=Built and booted rambi using these functions.
Change-Id: I325777bac27e9a0efc3f54f7223c38310604c5a2
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/173982
Provide a common entry point for bringing up the APs
in parallel. This work is based off of the Haswell one
which can be moved over to this in the future. The APs
are brought up and have the BSP's MTRRs duplicated in
their own MTRRs. Additionally, Microcode is loaded before
enabling caching. However, the current microcode loading
support assumes Intel's mechanism.
The infrastructure provides a notion of a flight plan
for the BSP and APs. This allows for flexibility in the
order of operations for a given architecture/chip without
providing any specific policy. Therefore, the chipset
caller can provide the order that is required.
BUG=chrome-os-partner:22862
BRANCH=None
TEST=Built and booted on rambi with baytrail specific patches.
Change-Id: I0539047a1b24c13ef278695737cdba3b9344c820
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/173703
Haswell was the original chipset to store the cache
in another area besides CBMEM. However, it was specific
to the implementation. Instead, provide a generic way
to obtain the location of the ramstage cache. This option
is selected using the CACHE_RELOCATED_RAMSTAGE_OUTSIDE_CBMEM
Kconfig option.
BUG=chrome-os-partner:23249
BRANCH=None
TEST=Built and booted with baytrail support. Also built for
falco successfully.
CQ-DEPEND=CL:172643
CQ-DEPEND=CL:*146397
CQ-DEPEND=CL:*146398
CQ-DEPEND=CL:*146435
CQ-DEPEND=CL:*146445
Change-Id: I70d0940f7a8f73640c92a75fd22588c2c234241b
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/172602
Reviewed-by: Stefan Reinauer <reinauer@google.com>
The access to control registers were scattered about.
Provide a single header file to provide the correct
access function and definitions.
BUG=chrome-os-partner:22991
BRANCH=None
TEST=Built and booted using this infrastructure. Also objdump'd the
assembly to ensure consistency (objdump -d -r -S | grep xmm).
Change-Id: Iff7a043e4e5ba930a6a77f968f1fcc14784214e9
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/172641
Reviewed-by: Stefan Reinauer <reinauer@google.com>
There are ARM systems which are essentially heterogeneous multicores where
some cores implement a different ARM architecture version than other cores. A
specific example is the tegra124 which boots on an ARMv4 coprocessor while
most code, including most of the firmware, runs on the main ARMv7 core. To
support SOCs like this, the plan is to generalize the ARM architecture so that
all versions are available, and an SOC/CPU can then select what architecture
variant should be used for each component of the firmware; bootblock,
romstage, and ramstage.
BUG=chrome-os-partner:23009
TEST=Built libpayload and coreboot for link, pit and nyan. Booted into the
bootblock on nyan.
BRANCH=None
Change-Id: I22e048c3bc72bd56371e14200942e436c1e312c2
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://chromium-review.googlesource.com/171338
Reviewed-by: Gabe Black <gabeblack@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
The Exynos family and most ARM products are SoC, not just CPU.
We used to put ARM code in src/cpu to avoid polluting the code base for what was
essentially an experiment at the time. Now that it's past the experimental phase
and we're going to see more SoCs (including intel/baytrail) in coreboot.
BUG=none
TEST=emerge-daisy chromeos-coreboot-snow;
emerge-peach_pit chromeos-coreboot-peach_pit
Change-Id: I5ea1f822664244edf5f77087bc8018d7c535f81c
Reviewed-on: https://chromium-review.googlesource.com/170891
Tested-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Ronald Minnich <rminnich@chromium.org>
Commit-Queue: Hung-Te Lin <hungte@chromium.org>
The ARM Makefile was copied from x86 and then modified, and as a result it
was carrying a lot of baggage. On top of that, the extra complication made it
inflexible, and we need a lot of flexiblity in order to support the fact that
the Tegra124 starts on an ARMv4 coprocessor instead of one of the ARMv7 main
CPUs.
BUG=None
TEST=Built and booted on pit. With this and other changes, built and booted
into the bootblock on nyan.
BRANCH=None
Change-Id: Ia6ddc27619bdb51e152ad0c628ad6f3037c103ce
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://chromium-review.googlesource.com/171017
Reviewed-by: Ronald Minnich <rminnich@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
The INTERMEDIATE variable was used to hook dd-ing the BL1 into the image for
Exynos SOCs, but we can do that directly without having a special hook.
BUG=None
TEST=Built for pit and snow and verified that the BL1 was still copied into
the image as expected.
BRANCH=None
Change-Id: I434506b52ca4ea1d01e25a785cbfe66dfdea21c4
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://chromium-review.googlesource.com/170921
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
Peppy had some issues with FUI. We decided it was time to create
peppy-specific gma.c and i915io.c files. Using yabel and the i915tool,
we generated a replay attack, then interpolated against the slippy
i915io.c to get something working.
Also, in preparation for moving code out of the mainboard gma.c to
generic driver code, we got rid of some hardcodes in the mainboard
gma.c that have no business being there. The worst were the
computation of gmch_[m,n] and it turns out that we had some
long-standing bugs related to confusion about 'bpp'. I've killed the
word bpp everywhere I could because there are at least 3 things that
correspond to bpp. We now have framebuffer, pipe, and panel bpp. The
names are long because I want to avoid all the mistakes we've all been
making in the last year :-) Sadly, that means a lot of changes not just
peppy-related, but they are simple and in a good cause.
The test pattern generation is driven by a global variable in
mainboard/peppy/gma.c. I've found in the past that it's very useful
to have a function like this available, as one can activate it while
using a jtag debugger: halt at the right place in ramstage, set the
variable to 1, continue. It's not enough code to worry about always
including.
The last hard-codes for M and N registers are gone, and the function
to set from generic intel_dp.c code works. To avoid screen trash on a
dev mode boot, which we liked but nobody else did :-), we now take the
time to put a pleasing background color that sort of doubles as a
power LED.
Rough timing is ramstage start is at 2.2, and dev setup is done at
3.3. These new platforms are depressingly slow to boot. Rom init alone
is taking 1.9 seconds. 13 years ago it was 3 seconds from power on to bash
prompt. These CPUs are at least 10x faster and take much longer to get going.
Future work, once we get this through, is to move more functions to the
intel driver, and combine the mainboard i915io.c into the mainboard gma.c.
That separation only existed because i915io.c was generated by a tool, and it
had lots of ugliness. Most ugliness is gone.
BUG=None
TEST=build and boot on peppy and get a screen, in both dev and normal modes.
BRANCH=None
Change-Id: I6a6295b423a41e263f82cef33eacb92a14163321
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://chromium-review.googlesource.com/170013
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Commit-Queue: Ronald Minnich <rminnich@chromium.org>
Tested-by: Ronald Minnich <rminnich@chromium.org>
Reviewed-by: Furquan Shaikh <furquan.m.shaikh@gmail.com>
I just spent half a day (including the time to implement a stack dumper)
to figure out that I am reading from a NULL pointer. A problem this
simple should be more easy to catch. Let's mark the address range below
SRAM as uncached so that the MMU can yell at you right away for being
the bad programmer you are when you access a NULL pointer.
BUG=None
TEST=Manual
Change-Id: I4a3a13f75bf21b25732be2ecb69d47503eff1b53
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/170112
Reviewed-by: Ronald Minnich <rminnich@chromium.org>
The source file reset.c, present in both the exynos5250 and 5420 directories,
is not being built for either SOC. Let's get rid of the clutter.
BUG=None
TEST=Built for snow and pit.
BRANCH=None
Change-Id: Iab4c7982a271d08cbaf3207b6f5431f0ef52697e
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://chromium-review.googlesource.com/170402
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: David Hendrix <dhendrix@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
The UART / serial console is put in retention state by kernel during suspend /
resume path, which caused Coreboot not able to print any messages during resume.
Sending values to the padret_uart_opt inside PMU may release UART, but that may
also cause unexpected output when kernel is back. However, it's still very
helpful when we are debugging suspend/resume inside Coreboot.
To get UART message on resume, call wakeup_enable_uart() in boot block or
romstage (before console_init).
BUG=none
TEST=emerge-peach_pit chromeos-coreboot-peach_pit
Change-Id: Ib5759cb402c6e018d9dba14fad8b61f6a1b1a265
Reviewed-on: https://chromium-review.googlesource.com/170440
Tested-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Commit-Queue: Hung-Te Lin <hungte@chromium.org>
This patch enhances the armv7 exception handlers in Coreboot and
libpayload to show the correct SP and LR registers from the aborted
context, and also dump a part of the current stack. Since we cannot
access the banked registers of SVC mode from a different exception mode,
it changes Coreboot (and its payloads) to run in System mode instead. As
both modes can execute all privileged instructions, this should not have
any noticeable effect on firmware operation (please correct me if I'm
wrong!).
BUG=None
TEST=Cause a data_abort in Coreboot or depthcharge. Marvel at the
beautifully displayed stack dump that helps you debug the abort.
Change-Id: I0e04f47619e55308f7da4a3a99c9cae6ae35cc30
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/170045
Reviewed-by: Gabe Black <gabeblack@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Trustzone needs to be initialized/disabled both on boot and on wake, so it
needs to be done before ramstage which doesn't run on wake. cpu.c isn't
compiled into romstage and fixing that causes other problems, so the trustzone
functions were split out.
BUG=chrome-os-partner:22759
TEST=Built and booted on for snow, pit and kirby. Successfully slept and woke
on each.
BRANCH=None
Change-Id: I8fc630237ebec1f02a91600f8baf3d4e9ea66d0e
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://chromium-review.googlesource.com/169817
Reviewed-by: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
The CPU_ADDR_BITS was being unconditionally set.
Don't do that.
BUG=None
BRANCH=None
TEST=Built bayleybay and can now set CPU_ADDR_BITS properly.
Change-Id: Idbc63328fade8f5f05f7f46282139b86e6694989
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/169711
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Only have the SPI driver produce output when CONFIG_DEBUG_SPI is
enabled.
BRANCH=none
TEST=booted on Pit, saw SPI messages from spi_claim_bus() go away
BUG=none
Change-Id: Ib911182b9b6a85e1a7b308f03c9b1bdba2fa0bf4
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://chromium-review.googlesource.com/170026
Tested-by: Stefan Reinauer <reinauer@chromium.org>
Reviewed-by: David Hendrix <dhendrix@chromium.org>
That extra struct is not needed, we already defined it earlier on.
Also fix coding style in the file.
BRANCH=none
TEST=boot tested on pit
BUG=none
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Change-Id: I586d290f2f3ba2f44aca7fdee400b88547465599
Reviewed-on: https://chromium-review.googlesource.com/169780
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Commit-Queue: Stefan Reinauer <reinauer@google.com>
Tested-by: Stefan Reinauer <reinauer@google.com>
This patch adds support for the DesignWare3 USB 3.0 DRD controller and
PHY to the Exynos5250 and Exynos5420 CPUs. It also adds code to the
Google Snow, Pit and Kirby boards to turn these controllers on where
applicable.
BUG=chrome-os-partner:21969
TEST=Manual
Change-Id: Idcca627363a69f1d65402e1acb9a62b439f077ff
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/169452
Reviewed-by: Stefan Reinauer <reinauer@google.com>
timer initialization is the first thing happening in
the Exynos CPU's bootblock code. Hence we don't need
to keep track of it in several places, and we don't
need to do it over and over again (e.g. in each stage)
BUG=none
BRANCH=none
TEST=boot tested on pit
Change-Id: I7bd9a0b7930fc9c37faabd62e3eecc3e5614a879
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://chromium-review.googlesource.com/168994
Reviewed-by: Stefan Reinauer <reinauer@chromium.org>
Commit-Queue: Stefan Reinauer <reinauer@chromium.org>
Tested-by: Stefan Reinauer <reinauer@chromium.org>
Right now some console specific objects are included
in the bootblock even if CONFIG_BOOTBLOCK_CONSOLE is
disabled while others are not. Make all of them conditional
and also fix a preprocessor misuse in bootblock_simple.c
and a stray (useless) die() in the Exynos wakeup code that
made inclusion of those files necessary.
BRANCH=none
BUG=none
TEST=boot tested on pit
Change-Id: Ia7f9d17654466f199b0e13afbdc9e14c9706530f
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://chromium-review.googlesource.com/168772
Reviewed-by: David Hendrix <dhendrix@chromium.org>
This patch cleans out a lot of unused variables in the
ARM Kconfig files and introduces CONFIG_RAMSTAGE_BASE
which is similar to CONFIG_RAMBASE on x86.
This gets rid of the hard coded assumption that on ARM
coreboot is always executed at the lowest DRAM address.
But in fact, this might not be true because we might want
coreboot to live at the end of RAM, or in SRAM
BUG=none
BRANCH=none
TEST=successfully booted on pit.
Change-Id: I03e992645f9eb730e39a521aa21f702959311f74
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://chromium-review.googlesource.com/168645
Reviewed-by: David Hendrix <dhendrix@chromium.org>
Tested-by: David Hendrix <dhendrix@chromium.org>
All this samsung_get_base_address_of_device_with_a_really_long_name()
boilerplate makes my eyes bleed... I think there are so much cleaner
ways to do this. Unfortunately changing this ends up touching nearly
every Exynos5 file, but I hope you agree that it's worth it (and the
sooner we get it over with, the better... I can't bring myself to make
another device fit into that ugly scheme).
This also removes the redundant EXYNOS5 base address definitions from
the 5420 directory when there are EXYNOS5420 ones, to avoid complete
confusion. The new scheme tries to use EXYNOS5 for base addresses and
exynos5 for types that are common between the two processors, and
EXYNOS5420/exynos5420 for things that have changes (although I probably
didn't catch all differences).
BUG=None
TEST=Try to write some code that flips a single PMU register bit without
getting super frustrated with all the extra stuff you need to set up.
Change-Id: I87e58434490ed55a9bbe743af1f9bf2520dec13f
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/167579
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Reviewed-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: ron minnich <rminnich@chromium.org>
Since we're now supporting ARMv7 relocations, we can enable
rmodule support on Exynos 5420. This does not automatically
enable relocatable ramstage.
Signed-off-by: Stefan Reinauer <reinauer@google.com>
BRANCH=none
TEST=needs more changes
BUG=none
Change-Id: Ic3af1eabb3b816944587a46409224f778d941b8a
Reviewed-on: https://chromium-review.googlesource.com/167403
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: Stefan Reinauer <reinauer@google.com>
Commit-Queue: Stefan Reinauer <reinauer@google.com>
We were running this loop 100 times with 5 ms delays. Change it
to run 500 times with 1 ms delays, which gives us the same
overall timeout but lets us bail out a bit sooner -- in practice,
at most, 4 ms sooner but every bit counts. Note, however, that
the tighter timing does reduce opportunities for threading. There
is a non-obvious set of tradeoffs on timeouts.
BUG=None
TEST=build and boot with this test and note we still have graphics
BRANCH=None
Change-Id: I4af671c2a791aa92e446e66ac2fe5710d1e6aa4c
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://chromium-review.googlesource.com/167387
Reviewed-by: David Hendricks <dhendrix@chromium.org>
Commit-Queue: ron minnich <rminnich@chromium.org>
Tested-by: ron minnich <rminnich@chromium.org>
This cleans up a few minor things (mostly #defines) of the memory code
for exynos5420, pit, and kirby. Specifically:
- CONCONTROL.empty is read-only, so don't try to set it and also
get rid of the unneeded DMC_CONCONTROL_EMPTY_ENABLE #define.
- MEMBASECONFIG* overlaps members of the mem_timings struct and
are mainboard-dependent anyway, so get rid of 'em.
- DMC_MEMCONTROL_TP_DISABLE corresponds to a reserved bit. It may
have been deprecated.
- Same with TIMING* #defines.
- Clarify DDR_MODE_* usage and use mem->mem_type when appropriate.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on pit and kirby
Change-Id: Ideb21efcc97b24f7e115e90051c20daef4480f17
Reviewed-on: https://chromium-review.googlesource.com/167500
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: ron minnich <rminnich@chromium.org>
This updates DRAM usage for Kirby so that we can actually
use the available 3.5GB:
- The chips on Kirby have 16 row address lines.
- CONFIG_DRAM_SIZE_MB should be 3584 (4096-512).
- We use 2 DMC channels on Kirby (each with 2GB).
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=chrome-os-partner:22144
BRANCH=none
TEST=built and booted (sorta) on Kirby, mem test succeeds
Change-Id: I86d1a96d0d1a028587f7655f8de5a2e52165e9d2
Reviewed-on: https://chromium-review.googlesource.com/167489
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: ron minnich <rminnich@chromium.org>
membaseconfig0/1 are utterly dependent on the mainboard's particular
DRAM setup. This defines their values in the mem_timings struct for
pit and kirby.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=booted on pit, ran /usr/local/opt/punybench/bin/memcpy_test -b,
tested on kirby in follow-up patch
Change-Id: Ifd782d1229b2418f8ddbf0bcb3f45cc828ac34b0
Reviewed-on: https://chromium-review.googlesource.com/167488
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: ron minnich <rminnich@chromium.org>
This changes the number of chip selects that we configure from 2 to 1.
On current setups with (x16 memory 4Gbit chips) that means that we're
at 2GByte.
Technically we should add a second setting in the ares_ddr3_timings
and select between the two of the based on board strappings. That
would make the CONFIG_RUN_TIME_BANK_NUMBER would work properly. I've
changed the ddr3_mem_ctrl_init() so it should handle that, but I'm not
actually doing the board strapping read right now.
This change means that accesses to 0xA0000000 - 0xFFFFFFFF on 2G
systems will no longer put the system in a messed up state (leading to
a hang). It also prevents some of the weird boot behavior that we've
seen that comes and goes depending on U-Boot alignment. See
<http://crosbug.com/p/20577>.
This patch was ported from: https://gerrit.chromium.org/gerrit/66117
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=booted on pit, ran suspend_stress_test --memory_check ran for
100 iterations and reboot seemed fine.
Change-Id: Ib4cfe420aac30bd817438f06d01e8671afc4a27d
Reviewed-on: https://chromium-review.googlesource.com/167210
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: ron minnich <rminnich@chromium.org>
This minor refactoring patch changes the signature of all limited cache
invalidation functions in Coreboot and libpayload from unsigned long to
void * for the address argument, since that's really what you have in
95% of the cases and I think it's ugly to have casting boilerplate all
over the place.
CQ-DEPEND=CL:167358
BUG=chrome-os-partner:21969
TEST=Make sure all payloads still compile cleanly when this and
dependent changes are in.
Change-Id: Ic9d3b2ea70b6aa8aea6647adae43ee2183b4e065
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/167338
Fine tuning DDR timings value for better stability
* Changed Data Driver Strength from 34 ohms to 30 ohms, expected to
enhance signal integrity.
* Changed DQ signal from 0xf to 0x1f000f, to keep default value safe.
* Changed mrs[2] and added new mrs direct command for setting WL/RL
without resetting DLL.
* Added explicit reset value write in phy_con0 instead of just setting
a bit, to ensure that reset happens.
* Added DREX automatic control for ctrl_pd in none read memory state.
This is ported from: https://gerrit.chromium.org/gerrit/61405
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=booted on pit, suspend_stress_test --memory_check -c10 passed
Change-Id: I59e96e6dede7b49c6572548aca664d82ad110bb1
Reviewed-on: https://chromium-review.googlesource.com/66995
Reviewed-by: ron minnich <rminnich@chromium.org>
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
For many boards, the EDID is known and is set in the ramstage. Reading
the EDID is slow and if we have it we do not want to reread it.
If the raw_edid struct member is non-null, skip reading the EDID.
BUG=None
TEST=Build and boot and verify that the panel works
BRANCH=None
Change-Id: I63fb11aa90b2f739a351cdc3209faac2713ea451
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://chromium-review.googlesource.com/167116
Reviewed-by: Gabe Black <gabeblack@google.com>
Tested-by: ron minnich <rminnich@chromium.org>
Commit-Queue: Ronald Minnich <rminnich@google.com>
This patch flushes the caches and disables the MMU before resuming.
c32b9b3 ("Set up caching in the bootblock.") had a bug where the
dcache and MMU remained enabled in the resume path. This caused
the machine to hang on resume. However, other bugs were preventing
us from testing this properly earlier on so it went unnoticed until
now.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=Built and booted on pit, powerd_dbus_suspend works again,
"suspend_stress_test --memory_check" ran 10 times without problems.
Change-Id: Ib1774f09d286a4d659da9fc2dad1d7a6fc1ebe5e
Reviewed-on: https://chromium-review.googlesource.com/67007
Reviewed-by: ron minnich <rminnich@chromium.org>
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
This patch intends to remove all code which enables hardware read
leveling. We need to disable h/w read leveling because new ASV table
is merged in kernel (which is based on the new characterization
condition) and new characterization environment has h/w read leveling
disabled, so we should also disable this. Also, disabling h/w read
leveling improves the MIF LVcc value (LVcc value is the value at which
DDR will fail to work properly), improve LVcc means we have enough
voltage margin for MIF. When h/w leveling is enabled, we have almost
zero volatge margin.
This was ported from: https://gerrit.chromium.org/gerrit/66070
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=booted on pit, suspend_stress_test --memory_check -c10 passed
Change-Id: Id0a2d77e6214325f226d51ae08464b39424cea83
Reviewed-on: https://chromium-review.googlesource.com/66994
Reviewed-by: Gabe Black <gabeblack@chromium.org>
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
This patch moves around some of the existing Exynos5 USB 2.0 PHY code
to make it cleaner in preparation of the 3.0 PHYs. It moves the VBUS
GPIOs (which are completely board-specific) into the mainboard code and
makes sure to only initialize PHYs on the boards that actually need
them. It also removes the USB 3.0 PLL hack that was needed on Snow from
the Pit and Kirby boards (which do not have that PLL anymore).
BUG=chrome-os-partner:21969
TEST=Manual
Change-Id: Ia35f47a765acff60481f0907f7448ec4f78e0937
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/66887
Reviewed-by: Stefan Reinauer <reinauer@google.com>
The old ddr3_mem_ctrl_init() for exynos5420 had hardcoded constants
for accessing directcmd registers. Modify to use #defines where
possible.
This is ported from: https://gerrit.chromium.org/gerrit/#/c/65616
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on Pit
Change-Id: I01567fc6941608a570832de97259c55e84942d01
Reviewed-on: https://gerrit.chromium.org/gerrit/66789
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
As per hardware recommendation, CKE PAD retention release must
happen just before gate leveling enable and only in case of resume.
Hence, this patch moves pad retention release from dmc_common.c to
dmc_init_ddr3_exynos5420.c. In addition to this we are providing
125 (+3 extra being safe) times auto refresh to DRAM by sending
REFA direct command. This is required because when CKE PAD retention
release happens, self refresh mode of DDR3 is disabled.
Hence, auto refresh 125 times.
This is ported from https://gerrit.chromium.org/gerrit/#/c/65573
Note: Since WAKEUP_DIRECT does not go thru memory init, it should be
safe to move CKE PAD retention out of bootblock.c.
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on Pit
Change-Id: Idec5d6fbbe3c6344d47401ba7203079c52a9b866
Reviewed-on: https://gerrit.chromium.org/gerrit/66788
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Set the CLK_DIV_CPERI1 value as recommended by the
0.02 UM section 7.9.1.25.
This suggests to use 0x3F3F0000 as the value to be
set to save power.
This is ported from https://gerrit.chromium.org/gerrit/#/c/64905
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on pit
Change-Id: I89a6a72d20374a513019a272628a05e139b31773
Reviewed-on: https://gerrit.chromium.org/gerrit/66787
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
This patch augments the alternative CBFS media source implementation for
Exynos5250 and Exynos5420 to allow booting from SDMMC devices (such as
an SD or uSD card reader, if available). It also moves MMC
initialization for the Snow, Pit and Kirby boards from romstage to
ramstage (mainboard_init) to prevent it from interfering with the IROM
during SDMMC boot.
BUG=chrome-os-partner:18733
TEST=cros_write_firmware -i /build/<board>/firmware/image.bin -w sd:.
(or simply dd if=<imagefile> of=/dev/sdX seek=1)
Change-Id: Ic4adef80c28262d084a53c28ec59aa7ac3af50c8
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/66154
The DWMMC controller internally divided clock by values in CLKSEL registers,
so we must adjust MMC clock for that.
Added a divide function to stdlib.h which should be useful in other contexts.
BUG=none
TEST=emerge-peach_pit chromeos-coreboot-peach_pit # and boots successfully.
Change-Id: I44f55b634cfc6fd81d76631595b6928c862a219f
Signed-off-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/66657
Commit-Queue: Ronald G. Minnich <rminnich@chromium.org>
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Tested-by: Ronald G. Minnich <rminnich@chromium.org>
Reviewed-by: David Hendricks <dhendrix@chromium.org>
Apparently the IROM doesn't like data caches... the recently added
dcache-in-bootblock makes A-A booting fail, and flushes/invalidations
alone don't seem to fix it. It's pretty fast anyway, so we just disable
the cache again for the duration of the IROM call.
Also removes a superfluous invalidation line from the bootblock code...
dcache_mmu_enable/disable already take care of that.
BUG=chrome-os-partner:18733
TEST=cros_write_firmware -i <coreboot image> -b peach_pit (with CL:65276)
Change-Id: I35580d15664c7b4197d4ed14028720147adbf918
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/66602
Reviewed-by: Gabe Black <gabeblack@chromium.org>
Reviewed-by: David Hendricks <dhendrix@chromium.org>
This improves firmware boot time substantially. Because cbmem isn't available
yet, we need to allocate some space in sram for the ttb. Doing cache
initialization in the bootblock means we can implement this once per CPU
instead of once per mainboard.
BUG=chrome-os-partner:19420
TEST=Built and booted on pit. Built and booted on snow.
BRANCH=None
Change-Id: Iad339de24df8ec2e23f91fe7bf57744e4cc766c5
Signed-off-by: Gabe Black <gabeblack@google.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/65938
Reviewed-by: David Hendricks <dhendrix@chromium.org>
Commit-Queue: Gabe Black <gabeblack@chromium.org>
Tested-by: Gabe Black <gabeblack@chromium.org>
Shorten a few delays, and make some delays shorter but let the
loops have a higher termination count (i.e. give it the same
amount of time to warm up, but check more frequently).
BUG=None
TEST=Build and boot many times, graphics is fine
BRANCH=None
Change-Id: Id9fe846ae3a8d792b14d62aea4e98d8aad05be43
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/66156
Commit-Queue: Ronald G. Minnich <rminnich@chromium.org>
Tested-by: Ronald G. Minnich <rminnich@chromium.org>
Reviewed-by: Stefan Reinauer <reinauer@google.com>
Call thread_yield_microseconds in udelay. This works with and without
COOP_MULTITASKING enabled.
BUG=None
TEST=Build and boot with multitasking on and off, it works.
BRANCH=None
Change-Id: Ib3eab00d1630dc4daada850e7458ab89702d1864
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/66327
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Commit-Queue: Ronald G. Minnich <rminnich@chromium.org>
Tested-by: Ronald G. Minnich <rminnich@chromium.org>
On x86, cpu_info lives at the top of stack. Make the arm do that as
well, as the threading model needs that and so will multicore support.
As part of this change, make the stack size a power of 2.
Also make it much smaller -- 2048 bytes is PLENTY for ram stage.
Note that the small stack size is counterintuitive for rom stage. How
can this work in rom stage, which needs a HUGE stack for lzma? The
main use of STACK_SIZE has always been in ram stage; since 2002 or so
it was to size per-core stacks (see, e.g.,
src/arch/x86/lib/c_start.S:.space CONFIG_MAX_CPUS*CONFIG_STACK_SIZE
and, more recently, thread stacks. So, we define the STACK_TOP for rom
and ram stage, but the STACK_SIZE has no real effect on the ROM stage
(no hardware red zones on the stack) and hence we're ok with actually
defining the "wrong" stack size. In fact, the coreboot_ram ldscript
for armv7 sizes the stack by subtracting CONFIG_STACK_BOTTOM from
CONFIG_STACK_TOP, so we replicate that arithmetic in bootblock.inc
Observed stack usage in ramstage:
BS: BS_PAYLOAD_LOAD times (us): entry 1 run 153887 exit 1
Jumping to boot code at 23104044
CPU0: stack: 02072800 - 02073000, lowest used address 020728d4, stack used: 1836 bytes
entry = 23104044
Which means we do need 2K, not 1K.
BUG=None
TEST=Build and boot and verify that cpu_info returns a correct value
BRANCH=None
Change-Id: I1a21db87081597efe463095bfd33c89eba1d569f
Signed-off-by: Ronald G. Minnich <rminnich@gmail.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/66135
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Tested-by: Ronald G. Minnich <rminnich@chromium.org>
Commit-Queue: Ronald G. Minnich <rminnich@chromium.org>
BRANCH=none
TEST=none
BUG=none
Change-Id: I1144e9d6d6c4278842fdd36743c8a88555f81707
Signed-off-by: Stefan Reinauer <reinauer@google.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/65912
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Commit-Queue: Stefan Reinauer <reinauer@chromium.org>
Tested-by: Stefan Reinauer <reinauer@chromium.org>
A few curly braces on the wrong line.
BUG=None
TEST=None
Change-Id: I4ddac4476c6509dc1716e8c1915fbdb67d346786
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/66153
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
This patch ports the USB A-A firmware upload functionality from
exynos5250 over to exynos5420. Essentially just like a conflicless
cherry-pick of 9e69421f5f. It also fixes
the exact same bug with SPI initialization for Pit and Kirby.
BUG=None
TEST=None (CBFS offsets different so I'll need to update CWF once more)
Change-Id: Ief0ed54c0beb2701e51201041f9bc426b2167747
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/65751
Reviewed-by: David Hendricks <dhendrix@chromium.org>
In UM ver0.02, 600MHz clock PMS values differs from what is programed
currently. Though this also results in 600MHz clock, but it is better to
match what UM says. This patch chnage this as per UM
This is ported from https://gerrit.chromium.org/gerrit/#/c/65106/3
(Note: we already used the correct 600MHz value for KPLL)
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on Pit
Change-Id: I6786815ab33427a23436e6ee37295f6c37dcd3d5
Reviewed-on: https://gerrit.chromium.org/gerrit/65726
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Tested-by: Ronald G. Minnich <rminnich@chromium.org>
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
lets enable memory controller internal clock gating for ddr3.
with these bits enabled we save some power out of ddr3.
This is ported from https://gerrit.chromium.org/gerrit/#/c/60774
Signed-off-by: David Hendricks <dhendrix@chromium.org>
BUG=none
BRANCH=none
TEST=built and booted on pit
Change-Id: I2f9b0d78483b3ea7441f54a715c7c1e42eda3f7f
Reviewed-on: https://gerrit.chromium.org/gerrit/65728
Reviewed-by: Ronald G. Minnich <rminnich@chromium.org>
Commit-Queue: David Hendricks <dhendrix@chromium.org>
Tested-by: David Hendricks <dhendrix@chromium.org>
