MT8192 devapc supports remapping domains.
There may be different domain bit for different subsys.
For example, domain bit in INFRA is 4-bit, while in MMSYS,
domain bit is 2-bit. For INFRA master to access MM registers,
the domain bit will change from 4 to 2 and need to be remapped.
In this patch we have remapped:
1. TINYSYS (3-bit to 4-bit)
- domain 3 to domain 3
- others to domain 15
2. MMSYS slave (4-bit to 2-bit)
- domain X to domain X, for X = 0 ~ 3
- others to domain 0
Change-Id: Id10a4c0bdf141cc76a386159896c861d0dc302aa
Signed-off-by: Nina Wu <nina-cm.wu@mediatek.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49790
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Move the initialization from bootblock to romstage for following reasons:
- Follow MT8183 initialization sequence.
- PMIC and RTC functions are only called after verstage.
- Reduce bootblock size.
- PMIC initialization setting is complex and may need to be changed by
an RW firmware update.
TEST=boot to kernel successfully
Change-Id: I3e4c3f918639590ffc73076450235771d06aae91
Signed-off-by: Yidi Lin <yidi.lin@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51409
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Xi Chen <xixi.chen@mediatek.com>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
We saw EXT_PMIC_EN1 and PPVAR_DVDD_PROC_BC power off sequence
failure, and after checking MT6315 MT6315 PMIC protection key
summary.xlsx and MT6315 Top and CLK programming guide.docx,
we found there are something wrong about the sequence of magic
key protection flow and clk setting. Update correct initial
flow.
BUG=b:179000151
BRANCH=none
TEST=boot asurada correctly
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Change-Id: I1b7f970a44904fda09a97f4064eef7c95feefad7
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51245
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The initial settings for MT6315 were not applied correctly
because the setup process didn't specify correct slave id
(incorrectly always sending 0), and may cause failure in
power off sequence.
BUG=b:179000151
BRANCH=none
TEST=boot asurada correctly
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Change-Id: Ifd04da8ac55bcc9f9fdbc088d430522c2725ad47
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51056
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Move PRERAM_CBMEM_CONSOLE to SRAM L2C and increase its size from 15K to
400K. With this change, most part of the DRAM full calibration log can
be stored in CBMEM console.
BUG=b:181933863
TEST=emerge-asurada coreboot
TEST=Hayato boots
BRANCH=none
Change-Id: I896884d298e197149f75865e9d00579124a34404
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51174
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Many header files and helper macros have been moved to the
common folder and we want to use them in mt8173/mt8183
DRAM calibration code.
Signed-off-by: Xi Chen <xixi.chen@mediatek.com>
Change-Id: Ifa483dcfffe0e1383cb46811563c90f0ab484d5d
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51224
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Mediatek has released the reference implementation for DRAM
initialization in vendorcode/mediatek/mt8192/dramc (CB:50294)
so we want to use it to replace the derived calibration code
in soc folder.
Signed-off-by: Xi Chen <xixi.chen@mediatek.com>
Change-Id: I2b2f41d774c6b85f106867144fb0b29a4a1bdfcf
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51125
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
This patch changes the memlayout macro infrastructure so that the size
of a region "xxx" (i.e. the distance between the symbols _xxx and _exxx)
is stored in a separate _xxx_size symbol. This has the advantage that
region sizes can be used inside static initializers, and also saves an
extra subtraction at runtime. Since linker symbols can only be treated
as addresses (not as raw integers) by C, retain the REGION_SIZE()
accessor macro to hide the necessary typecast.
Signed-off-by: Julius Werner <jwerner@chromium.org>
Change-Id: Ifd89708ca9bd3937d0db7308959231106a6aa373
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49332
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
For MT8192 MCUs, replace LZMA compression with LZ4 to speed up boot
process. The loading (plus decompression) time of mcupm.bin and sspm.bin
is consistently reduced by 8ms, respectively.
BUG=b:177389446
TEST=emerge-asurada coreboot
TEST=Hayato booted up
BRANCH=none
Change-Id: Ida35e7f6e0572ad43082e53bcc69bc708cf7da44
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50278
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
From the output of 'objdump -x dram.elf', the DRAM blob needs 222K
memory, but currently only 208K is reserved for it. Since MT8192 has 1MB
SRAM L2C, increase SRAM_L2C_END to 0x00300000, and reorganize regions in
SRAM_L2C to have larger DRAM_INIT_CODE (256K). The size of
OVERLAP_DECOMPRESSOR_VERSTAGE_ROMSTAGE is also increased to 252K.
BUG=b:170687062
TEST=emerge-asurada coreboot
TEST=Asurada booted successfully
BRANCH=none
Cq-Depend: chrome-internal:3568265
Change-Id: I062f00739b72cf6b1bb7ac3318b91721fbe226cc
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50017
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Raise little CPU to 2GHz at romstage to improve boot time.
BUG=b:177389446
TEST=observe boot time by `cbmem`
Before: 1,062,359 us
After: 907,458 us
Signed-off-by: Yidi Lin <yidi.lin@mediatek.com>
Change-Id: I723a916d7f708627525ef11e3c5ea0b381f269aa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49935
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Some efuse settings would not be applied automatically, so we need
set the settings manually. The low power consumption would not be
optimal without correct efuse settings.
BUG=b:172636735
BRANCH=none
TEST=see 'pmic_efuse_setting: Set efuses in 11 msecs'
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Change-Id: Ideb862c3cb0f1fee183804aed74fcf141bf1f5df
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49006
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
For pq module size registers such as DISP_AAL_SIZE, the high bits
should be HSIZE, while low bits should be VSIZE. Fix the incorrect
settings for these registers where width and height are reversed.
According to MediaTek, there is no practical impact on mt8183 devices,
but it's still nice to get this fixed to avoid future confusion.
BUG=b:171167210
TEST=none
BRANCH=kukui
Change-Id: I4b6aedf9a3ca133fcbe9cb88b99a13d228233e24
Signed-off-by: Yu-Ping Wu <yupingso@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46626
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
We need to write some special values to key protection registers before
applying init_setting table and lp_setting table to PMIC. Otherwise,
those settings won't take effect.
After applying init_setting table and lp_setting table, we lock the
settings by writing zero to key protection registers.
Reference datasheet: MT6359_PMIC_Data_Sheet_V1.5.docx, RH-D-2018-0205.
BUG=b:172636735
BRANCH=none
TEST=boot asurada correctly
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Change-Id: I593d4e02bf0b62ac297957caf4ae1c1837f1f38d
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48954
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
We found that the switch frequency of vgpu is at 4~5Mhz with high
current case (> 3.5A) and is at 2.5Mhz with low current case(< 2.8A).
The switch frequency of vgpu should be kept at 2.5Mhz.
The root cause is that phase config of vcore is not disabled, it will
affect the switch frequency of vgpu. Corret the phase setting at
initialization.
BUG=b:172636735
BRANCH=none
TEST=boot asurada correctly
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Change-Id: I48d3729302de9e3343dce79fe6f5ed045d0296a5
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49005
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
After calibration, we can get ddr vendor id or density info from
MR5 or MR8, this helps to make sure the DDR HW is as we expected.
Signed-off-by: Huayang Duan <huayang.duan@mediatek.com>
Change-Id: Ie62948368716d309aab8149372b2b6093fc33552
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44712
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
SoC will transmit the EoTp (End of Transmission packet) when
MIPI_DSI_MODE_EOT_PACKET flag is set.
Enabling EoTp will make the line time larger, so the hfp and
hbp should be reduced to keep line time.
BUG=b:168728787
BRANCH=kukui
TEST=Display is normal on Kukui
Signed-off-by: Shaoming Chen <shaoming.chen@mediatek.corp-partner.google.com>
Change-Id: Ifadd0def13cc264e9d39ab9c981fbdc996396bfa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48868
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
The flash controller driver can be shared among mt8173 and mt819x.
TEST=boot to kernel on Asurada
boot to kernel on Hana (w/o BL31)
Change-Id: I4e5213563189336496122a0f2d8077b3e5245314
Signed-off-by: Yidi Lin <yidi.lin@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48972
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
To get better PI linearity, perform 8 phase calibration to do
MCK 0/180/45 training and select the best PI settings.
Signed-off-by: Huayang Duan <huayang.duan@mediatek.com>
Change-Id: Ib4ccaa8d43b8382cbc64cf82de86ad1ac16cb89a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44709
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
eint event mask register is used to mask eint wakeup source on mt8192.
All wakeup sources are masked by default. Since most MediaTek SoCs do
not have this design, we can't modify the kernel eint upstream driver to
solve the issue 'Can't wake using power button (cros_ec) or touchpad'.
So we add a driver here to unmask all wakeup sources.
BUG=b:169024614
Signed-off-by: G.Pangao <gtk_pangao@mediatek.com>
Change-Id: I8ee80bf8302c146e09b74e9f6c6c49f501d7c1c4
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46409
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add rtc MT6359P driver for rtc init and rtc eosc calibration. Refactor
mt8173 and mt8183 code by extracting common API. Move rtc_read and
rtc_write to each SoC folder, because mt8173 and mt8183 access rtc via
pmic wrapper, while mt8192 accesses it via pmif.
Reference datasheet:
Document No: RH-D-2018-0101.
Signed-off-by: Yuchen Huang <yuchen.huang@mediatek.corp-partner.google.com>
Change-Id: I57d6738fdec148c7458b2024a0a8225415ca2f3e
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46395
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Add basic devapc (device access permission control) drivers.
DAPC driver is used to set up bus fabric security and data protection
among hardwares. DAPC driver groups the master hardwares into different
domains and gives secure and non-secure property. The slave hardware can
configure different access permissions for different domains via DAPC
driver.
Change-Id: I2ad47c86b88047c76854a6f8a67b251b6a9d4013
Signed-off-by: Nina Wu <nina-cm.wu@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46402
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Memory PLL is used to provide the basic clock for dram controller
and DDRPHY. PLL must be initialized as predefined way.
First, enable PLL POWER and ISO, wait at least 30us, release ISO, then
configure PLL frequency and enable PLL master switch.
At last, enable control ability for SPM to switch between active and
idle when system is switched between normal and low power mode.
TEST=Confirm Memory PLL frequency is right by frequency meter
Signed-off-by: Huayang Duan <huayang.duan@mediatek.com>
Change-Id: Ieb4e6cbf19da53d653872b166d3191c7b010dca6
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44702
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
