From AMD USB phy specialist recommended that for DB port2 (type-A), port3 (type-C C1)
the most effective corrections for the depressed eye are
tx_rise_tune=0x0
tx_pre_emp_amp_tune=0x3
tx_fsls_tune = 0x3
BUG=b:173476380
BRANCH=zork
TEST=1. emerge-zork coreboot
2. pass USB 2.0 SI eye diagram verification
Change-Id: Ib31c5d55e30b958d3e552e8d0b4a160947444636
Signed-off-by: Kevin Chiu <kevin.chiu@quantatw.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49826
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
From AMD USB phy specialist recommended that for DB port2 (type-A), port3 (type-C C1)
the most effective corrections for the depressed eye are:
tx_rise_tune=0x0
tx_pre_emp_amp_tune=0x3
tx_fsls_tune = 0x3
BUG=b:165209698
BRANCH=zork
TEST=1. emerge-zork coreboot
2. pass USB 2.0 SI eye diagram verification
Change-Id: I80afd6bf1257b9a72d0d7651b48d243ebaf5de2f
Signed-off-by: Kevin Chiu <kevin.chiu@quantatw.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49824
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Rob Barnes <robbarnes@google.com>
The only difference is an additional include that is no longer needed.
Change-Id: I0053d03aa4d05f5c0fa833d8634419b6667e38a7
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49832
Reviewed-by: EricR Lai <ericr_lai@compal.corp-partner.google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
There's no need to repeat the same code on every board.
Change-Id: I2e19decfe8609fa644e609673a56ee5109bafefa
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49831
Reviewed-by: EricR Lai <ericr_lai@compal.corp-partner.google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This change uses the newly added meminit block driver and updates TGL
SoC and mainboard code accordingly.
TEST=Verified that UPDs are configured correctly with and without this
change.
Change-Id: I6d58cd6568b7bbe03c4e3011b2301209893e85a9
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49042
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: EricR Lai <ericr_lai@compal.corp-partner.google.com>
This change adds support for a common block memory driver that can be
used for performing the required operations to read SPD data for
different memory channel DIMMs. This data can then be used by the SoC
code to populate different memory related UPDs.
Most recent Intel platforms follow a similar pattern for configuring
FSP-M UPDs for initializing memory. These platforms use one of the
following topologies:
1. Memory down
2. DIMM modules
3. Mixed
Thus, SPD data is either obtained from CBFS (for memory down topology)
or from on-module EEPROM (for DIMM modules). This SPD data read from
CBFS or EEPROM is then passed into FSP-M using SPD UPDs for different
channels/DIMMs as per the memory organization.
Similarly, DQ/DQS configuration is accepted from mainboard and passed
into FSP-M using UPDs as per the FSP-M/MRC organization of memory
channels.
Different memory technologies on a platform support physical channels
of different widths. Since the data bus width is fixed for a platform,
the number of physical channels is determined by data bus width /
physical channel width. The number of physical channels are different
depending upon the size of physical channel supported by the memory
technology. FSP-M for a platform uses the same set of UPDs for
different memory technologies and aims at providing maximum
flexibility. Thus, the platform code needs to format mainboard inputs
for DQ, DQS and SPD into the UPDs appropriately as per the memory
technology used by the board.
Example: DDR4 on TGL supports 2 physical channels each 64-bit
wide. However, FSP-M UPDs assume channels 16-bit wide. Thus, FSP-M
provides 16 UPDs for SPDs (considering 2 DIMMs per channel and 8
channels with each channel 16-bit wide). Hence, for DDR4, only the SPD
UPDs for MRC channel 0 and 4 are supposed to be used.
This common driver allows the SoC to define the attributes of the
platform:
1. DIMMS_PER_CHANNEL: Maximum DIMMs that are supported per channel by
any memory technology on the platform
2. DATA_BUS_WIDTH: Width of the data bus.
3. MRC_CHANNEL_WIDTH: Width of the channel as used by the MRC to
define UPDs.
In addition to this, the SoC can define different attributes of each
memory technology supported by the platform using `struct
soc_mem_cfg`:
1. Number of physical channels
2. Physical channel to MRC channel mapping
3. Masks for memory down topologies
Using the above information about different memory technologies
supported by the platform and the mainboard configuration for SPD,
the common block memory driver reads SPD data and provides pointers to
this data for each dimm within each channel back to the SoC code. SoC
code can then use this information to configure FSP-M UPDs
accordingly. In addition to that, the common block driver also returns
information about how the channels are populated so that the SoC code
can use this information to expose DQ/DQS information in FSP-M UPDs.
This driver aims at minimizing the effort required for supporting
different memory technologies on any new Intel SoC by reducing per-SoC
effort to a table of configurations rather than having to implement
similar logic for each SoC.
BUG=b:172978729
Change-Id: I256747f0ffc49fb326cd8bc54a6a7b493af139c0
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49040
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: EricR Lai <ericr_lai@compal.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Creating named objects within a method is highly inefficient. So, pass a
reference to the OperationRegion field that needs to be updated instead.
Change-Id: I88272fc5cbe35427ccb5fc50789d47b66ace88fe
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46967
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
None of the mainboards have the magic SpeedStep device, so the C-state
generation function bails out without doing anything. Moreover, this
code is broken and was copied from Sandy Bridge. Thus, drop it.
Change-Id: I580157ee33c599af5fc48b06eeb39cb32c9831ec
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49806
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
When SerialIO devices are disabled, their _STA method evaluates to 0,
which means the device is not present. It is expected that OSPM would
not attempt to change the power state of a device that is not present.
Lynxpoint does not have these checks, thus remove them from Broadwell.
Also remove the now-unused Arg1 parameter to avoid warnings from IASL.
Change-Id: Ic5e999ac1171ce49db66bec45c58d8aa5711ec53
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46966
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Michael Niewöhner <foss@mniewoehner.de>
Unlike Ivy Bridge series, there isn't a method to flash coreboot
internally when running vendor firmware (yet). Until someone finds a way
to bypass flash protections, the first flash has to be done externally.
Change-Id: Idaff264f2b7277516d69d1323f1a0c885b28c3db
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49848
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Kevin Chiu