Accesses to architectural registers should be really fast -- they're
just registers, after all. In fact, the arm64 architecture uses them for
some timing-senstive uses like the architectural timer. A read should be:
one instruction, no data dependencies, done.
However, our current coreboot framework wraps each of these accesses
into a separate function. Suddenly you have to spill registers on a
stack, make a function call, move your stack pointer, etc. When running
without MMU this adds a significant enough delay to cause timing
problems when bitbanging a UART on SDM845.
This patch replaces all those existing functions with static inline
definitions in the header so they will get reduced to a single
instruction as they should be. Also use some macros to condense the code
a little since they're all so regular, which should make it easier to
add more in the future. This patch also expands all the data types to
uint64_t since that's what the actual assembly instruction accesses,
even if the register itself only has 32 bits (the others will be ignored
by the processor and set to 0 on read). Arm regularly expands registers
as they add new bit fields to them with newer iterations of the
architecture anyway, so this just prepares us for the inevitable.
Change-Id: I2c41cc3ce49ee26bf12cd34e3d0509d8e61ffc63
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/27881
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
When we first created the arm64 port, we weren't quite sure whether
coreboot would always run in EL3 on all platforms. The AArch64 A.R.M.
technically considers this exception level optional, but in practice all
SoCs seem to support it. We have since accumulated a lot of code that
already hardcodes an implicit or explicit assumption of executing in EL3
somewhere, so coreboot wouldn't work on a system that tries to enter it
in EL1/2 right now anyway.
However, some of our low level support libraries (in particular those
for accessing architectural registers) still have provisions for
running at different exception levels built-in, and often use switch
statements over the current exception level to decide which register to
access. This includes an unnecessarily large amount of code for what
should be single-instruction operations and precludes further
optimization via inlining.
This patch removes any remaining code that dynamically depends on the
current exception level and makes the assumption that coreboot executes
at EL3 official. If this ever needs to change for a future platform, it
would probably be cleaner to set the expected exception level in a
Kconfig rather than always probing it at runtime.
Change-Id: I1a9fb9b4227bd15a013080d1c7eabd48515fdb67
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/27880
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
CNTFRQ_EL0 is a normal AArch64 architectural register like hundreds of
others that are all accessed through the raw_(read|write)_${register}()
family of functions. There's no reason why this register in particular
should have an inconsistent accessor, so replace all instances of
set_cntfrq() with raw_write_cntfrq_el0() and get rid of it.
Change-Id: I599519ba71c287d4085f9ad28d7349ef0b1eea9b
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/27947
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Rotor is dead, long live [PROJECT NAME REDACTED]!
Change-Id: Ia9308944257255e077a44c1df262c7f49c69890c
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/27964
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add 3 new Kconfig options:
DRAM_PART_NUM_IN_CBI
DRAM_PART_NUM_ALWAYS_IN_CBI
DRAM_PART_IN_CBI_BOARD_ID_MIN
These control whether to 1. attempt to use CBI at all 2. always use cbi
and 3. conditionally use cbi based on board id. The intent is that the
MIN variant would be used for the tranisition period then cut over to
ALWAYS after full transition. Since multiple OEMs have different
schedules these options are there to bridge the gap. yorp. bip, and
octopus build targets would never flip DRAM_PART_NUM_IN_CBI, but in case
someone does the MIN values are 255 to always take the old path.
BUG=b:112203105
TEST=Set correct part number on phaser during testing.
Change-Id: If9a0102806d78e89330b42aa6947d503a8a2deac
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/27946
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Within procedure arch_write_tables, the pointer "rom_table_end" is updated
every time a table is created. However, after creating last table, pointer
rom_table_end is not used, though it is updated. Add a "(void)rom_table_end;"
at the end to avoid the static analysis error.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I8a34026795c7f0d1bb86c5f5c0469d40aa53994a
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27958
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Within procedure save_bsp_msrs, the structure pointer "msr_entry" is updated
every time procedure save_msr() is called. However, after the last call of
save_msr(), "msr_entry" is not used, thus causing a static analysis error.
Add a "(void)msr_entry;" at the end to avoid the static analysis error.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: If0fb336fbf49eec3da255fadbe38b3a38768d0cf
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27956
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Careena EVT SanDisk EMMC sku has high fail rate of 0x5B reboot failure.
It'll need to increase 1.8V EMMC CLK/CMD, Data driving strength for
this issue.
CLK[6:4]
CMD,DATA[3:1]
original register value: 0x6B
enhanced: 0x7F
BUG=b:111964336
BRANCH=master
TEST=emerge-grunt coreboot
Change-Id: I3db38ff12c566c258895c6643008a0472ca528bb
Signed-off-by: Kevin Chiu <Kevin.Chiu@quantatw.com>
Reviewed-on: https://review.coreboot.org/27816
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
In procedure spi_flash_cmd_erase(), parameter "len" is not validated and
could lead to the return of an invalid (non-initialized) value. Validate
the parameter early on.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I0b5129a15c9e0ea45f4dba4ab0729196cb64699b
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27952
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
In procedure allocate_cpu_devices(), if structure pointer new is null skip
using the pointer. Add a "continue;" to skip using the pointer.
The issue was found by static analysis tool.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I7011fbfa0725f22a6dfbca6752e668eddac3463c
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27951
Reviewed-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The Micron material that was broken has long since been fixed that
required this option. glkrvp had these stale entries and were
subsequently copied to octopus. Remove the need for this option.
BUG=b:35581751
Change-Id: Id73584367c2ad0e4958b5ea0f04a28e5fc82d085
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/27959
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Justin TerAvest <teravest@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
If TSEG_BASE is not TSEG_SIZE aligned the SMRR settings are invalid, therefore
guard against this.
Change-Id: I48f55cdac5f4b16b9a8d7a8ef3a84918e756e315
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/27870
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
RK3288 has always been notoriously low on SRAM, to the point where its
boards have less than 100 bytes left in both their bootblock/verstage
sections. This becomes a problem every time we try to add a tiny amount
of code to common coreboot interfaces that are included in them.
This patch manages to add another KB to each, one from the CBMEM console
(which now might get cut off a bit, but that's life) and one by moving
the TTB_SUBTABLES to PMUSRAM. PMUSRAM is a weird world where write
accesses must always be exactly 4 bytes long or they hang the CPU, so we
mostly ignore it... but thankfully, page table entries are exactly 4
bytes long and that's the only thing we write to this region, so it
works out in this case.
Change-Id: I5aecd66db40b3f52299b270322b8c8784dbe7e6f
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/27950
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Use bitfields to pack the struct more tightly.
Change-Id: If1e7a5a3a9504327f987403ec0a7b79b2383792a
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/27815
Reviewed-by: Julius Werner <jwerner@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The implementation of wpcd376i in coreboot is based on the
superiotool output which apparently was incorrect. This
patch refines the implementation to match the datasheet.
Change-Id: I0108e912dc4f603276074f0999c6d3146c3b13f9
Signed-off-by: Stefan Tauner <stefan.tauner@gmx.at>
Reviewed-on: https://review.coreboot.org/27857
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Within procedure cea_hdmi_block, the variable "b" is used as an index into
a buffer of EDID bytes. At the end, it's incremented but not used, thus
causing a static analysis error. Add a "(void)b;" at the end to avoid the
static analysis error.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: Ibd0b4a21bf82fcc46a627bc75564a850b7374989
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27929
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Skylake SoC code now sets the icc_max based on the CPU SKU, so we
should not hard-code it in the device tree.
BUG=b:110890675
BRANCH=None
TEST=boots on atlas
Change-Id: I7eb3499b7bea9ab2c49e1f299e2dbb688c8d1c33
Signed-off-by: Caveh Jalali <caveh@chromium.org>
Reviewed-on: https://review.coreboot.org/27791
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Gaggery Tsai <gaggery.tsai@intel.com>
In procedure tpm_unmarshal_response(), variable "rc" is used early to
decide if it should return NULL. Later however, the code proceeds to its
end even if one subroutine reports error. If "rc" is not 0, report that
there was a partial error in the procedure.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I7575bc75104fd97f138224aa57561e68f6548e58
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27931
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
BUG=none
TEST=compiled on grunt and made sure USB still works in depthcharge
Change-Id: I972f4604bb5ff3838cb15f323c5a579ad890ecf5
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27883
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
If a target is interrupted in the middle of writing an output file, the
file could be left in a corrupt state. A subsequent make invocation will
see the file as up to date and can cause very confusing errors.
BUG=b:112267918
TEST=Made a target fail before completion and verified make deleted the
output file.
Change-Id: I865827ea769b4dffa638d4324fc7284f6cb2ddc0
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27885
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Add the ability to specify the fwid version via a file instead of
via config. This makes it so when doing an incremental build all
objects are not invalidated when bumping the fwid.
The coreboot ebuild will create this file to pass the latest version.
BUG=b:112267918
TEST=ran dmidecide -t 0 and verified version was present
Change-Id: I955106efd648a75a1311f24ede46bd238d1517e0
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27884
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
This configures the ACPI FADT perferred power management profile to
PM_MOBILE instead of PM_DESKTOP.
I'm not sure what impact this actually has. I just noticed the other
boards have it set.
BUG=b:110971913
TEST=Made sure SYSTEM_TYPE_LAPTOP shows up in coreboot.config
Change-Id: Iea1b8359b80d167e69745358f543f025713294ba
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27930
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
By setting this register in bootblock AmdInitEnv will no longer trigger
a reset in romstage. This fixes a few vboot test failures and also
speeds up boot time.
BUG=b:111610455
TEST=Built grunt and made sure bootblock only happens once on cold boot,
and S3 resume.
Change-Id: Ie19f7a14deaef45ac63156bec6946273c1b9447e
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/27876
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add a function to provide a rudimentary dump of the Machine Check
Architecture registers. These values survive a warm reset.
BUG=b:65445599
TEST=Verify on a Grunt having propensity for #MC errors
Change-Id: Ib6875cabe3041e65c811d8b2232f7ac6bedd1a02
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/27926
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Extend the existing reset handling features in Stoney Ridge to plan for,
and recognize, warm resets. The ColdRstDet bit is always zero on a cold
reset, and is intended as a mechanism for the BIOS to determine the type
of a reset that occurred.
Set ColdRstDet=1 after all cores have been initialized, so that any
subsequent reset may be identified as warm/cold. A later patch will check
the value during mp_init.
Change-Id: I90255918de03018c9f090bff1e56a8bda5e7365e
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/27924
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Use the value discovered in the MCG_CAP[Count] for the number of MCA
status registers to clear. The generations should have the following
number of banks:
* Family 10h: 6 banks
* Family 12h: 6
* Family 14h: 6
* Family 15h: 7
* Family 16h: 6
Change-Id: I0fc6d127a200b10fd484e051d84353cc61b27a41
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/27923
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Remove for() braces from around single lines. Remove extra blank lines.
This cleans up checkpatch problems in a subsequent patch.
Change-Id: I329ac03365e51799581c56eed27ee54de6826f14
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/27935
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Change the defined name of MCI_STATUS (i.e. MCi_STATUS) to reflect its
MC0_STATUS address.
Change-Id: I97d2631a186965bb8b18f544ed9648b3a71f5fb0
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/27922
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
The DRAM part number can be stored in the CBI data. Therefore, add
support for fetching the DRAM part number from CBI.
BUG=b:112203105
TEST=Fetched data from CBI on phaser during testing.
Change-Id: Ia721c01aab5848ff36e11792adf9c494aa25c01d
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/27945
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Justin TerAvest <teravest@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
The current call for saving dimm info passed the lpddr4_cfg and
memory sku id. In order to prepare decoupling the part number
from lpddr4_cfg provide a new API, save_lpddr4_dimm_info_part_num(),
which explicitly takes the part number. The previous API now
uses the new one internally.
BUG=b:112203105
Change-Id: Ieadf452b6daa3231a0c5e3be61b0603b40d0fff2
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/27944
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Justin TerAvest <teravest@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
nWR (Write-Recovery for AutoPre-charge commands), the programmed value of nWR
is the number of clock cycles the LPDDR4-SDRAM device uses to determine the
starting point of an internal Pre-charge operation after a Write burst with
AP (auto-pre-charge) enabled. For >2133MHz speed parts the nWR needs to
be set to 24 clock cycles. The nWR field, though, is only in the GLK
FSP, so just update that field conditionally based on the GLK Kconfig
option.
BUG=b:112062440
TEST= build test
Change-Id: I1147538f72f4e2f14e32f3657c05f1f505a56fbf
Signed-off-by: Ravi Sarawadi <ravishankar.sarawadi@intel.com>
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/27850
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
In procedure generate_cpu_entries(), the code was copied from code that
could change variables "plen" and "pcontrol_blk" based on number of cores.
This is not the case with stoneyridge (2 cores only), and there's no need
to use the variables. Remove them and replace with fixed values.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I0258b19960b050e8da9d218ded3f1f3bfccad163
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27877
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
In procedure pci_get_resource, when setting an IO mapped base address,
variable attr is &= with PCI_BASE_ADDRESS_IO_ATTR_MASK. However, in this
particular code flow variable attr is not used later. Remove the line.
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: Ia4fdda1be92d22017a7a913a911db15aaa440b69
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27928
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
The variable "begin" is extracted from the structure, but 4 lines below
it's overwritten with "end - size". This causes a static build scan error
that should be fixed. Remove the initial assignment of variable "begin".
BUG=b:112253891
TEST=Build and boot grunt.
Change-Id: I0a265747e61289f045c5cac09e40478bd31e16fc
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/27886
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Julius Werner