Skip calling vbt_get() if FSP is not supposed to run GOP.
Change-Id: I6b8cd3646ffcd6df39229d4e36b315dfb7a8c859
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36111
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
The current DCACHE_BSP_STACK_SIZE is set to 128KiB for CML & ICL when
FSP uses the same stack provided by coreboot. This patch updates it to
129KiB since the default value of DCACHE_BSP_STACK_SIZE must be
the sum of FSP-M stack requirement (128KiB) and CB romstage
stack requirement (~1KiB).
BUG=b:140268415
TEST=Build and boot CML-Hatch.
Change-Id: Icedff8b42e86dc095fb68deb0b8f80b2667cfeda
Signed-off-by: V Sowmya <v.sowmya@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36032
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Maulik V Vaghela <maulik.v.vaghela@intel.com>
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Let the mainboard decide whether to let coreboot load the verb table.
Change-Id: I8f05ac02f690a43ada470916f5292b83aeaa8a4f
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35274
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Make the driver work with stepping=0.
Change-Id: Id0961369b9cc9cfe1b0c09ebc50e6966ccd2e919
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35273
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Prepare for products that can use any DRAM for TSEG.
Include an ID for data pointed to by an ACPI BERT table. This region's
only requirement is it is marked reserved.
Change-Id: Ia6518e881b0add71c622e65572474e0041f83d61
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36115
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Clarify names as I2C2, etc. Use iomap.h defines for base addresses.
Update IRQs.
Change-Id: I3800592e4b0bcb681d0dcf24f69e269f845be025
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34915
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add a function to uart.c to ensure the right IOMux settings are
programmed for the console UART. Update Kconfig to reflect the
new addresses.
Give the user the ability to downclock the UARTs' refclock to
1.8342MHz.
Add the abiltiy to use an APU UART at a legacy I/O address.
Update the AOAC register configuration for the two additional
UARTs.
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Change-Id: I74579674544f0edd2c0e6c4963270b442668e62f
Reviewed-on: https://review.coreboot.org/c/coreboot/+/33767
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add Picasso's Device ID and default filename.
Only a single Device ID is documented for Picasso so remove the oprom
remapper function.
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Change-Id: Iaf43d7c8da41beb05b58c494f0a6814f8f571b18
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34422
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Picasso's SATA controller operates only in AHCI mode. Remove the
Kconfig symbols previously used to select between other possibilities.
Change-Id: Iaeb8b4a2540e976d2e7361faf8c6d261e60398fd
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36113
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
SATA is no longer defined in AOAC so remove its definitions.
Change-Id: Ief0ab6b5f69f2d17c11d8e2ee40941ac56c077f6
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36112
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Picasso's FCH has many similarities to Stoney Ridge, so few changes
are necessary. The most notable changes are:
* Update the index values for the C00/C01 interrupt routing
* FORCE_STPCLK_RETRY is not present
* PCIB is not defined
* FCH MISC Registers 0xfed80e00 numbering has changed
* C-state base moves from PM register to MSR
* Add option to determine the intended MUX settion for LPC vs. eMMC
* Remove the LEGACY_FREE option
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Change-Id: I69dfc4a875006639aa330385680d150331840e40
Reviewed-on: https://review.coreboot.org/c/coreboot/+/33770
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
The standard PCI register space for D14F0 is accessible at 0xfed80000.
Add functions for use as helpers.
Change-Id: Icbf5bdc449322c3f5e59e6126d709cb2808591d5
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34914
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
The LPC-ISA bridge supports two ranges for SuperIO control registers.
Add a generic function to allow a mainboard to enable the appropriate
range. Provide #define values that are more descriptive than the
register's field names.
Change-Id: Ic5445cfc137604cb1bb3ee3ea4c3a4ebdb9a9cab
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35271
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Change to the appropriate device IDs. Remove the ehci resource
call. Remove overcurrent settings, as this will be passed to
AGESA in later change.
Remove unused USB2 ACPI name assignment.
Change-Id: Ic287a05b30ca03e3371cc4a30aaa93b236c6d3fb
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34420
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add Kconfig options and Makefile command line options to generate
the amdfw.rom image. A new intermediate image is introduced, which
is the initial BIOS image the PSP places into DRAM prior to
releasing the x86 reset. The amd_biospsp.img is a compressed
version of the romstage.elf program pieces.
Additional details of the PSP items are not public information. See
NDA document PID #55758.
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Change-Id: Ib5e393e74ed60e968959012b6275686167a2d78a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/33764
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Fix compiler warning by adding an additional check for the fastpath
memset.
Change-Id: I9a80438995bafe7e436f3fe2180b8c9574eeff23
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35682
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
It is similar to X200s, with U-series CPU, slightly different gpio
setup, no docking support, and no superio chip.
Tested:
- CPU Core 2 Duo U9400
- Slotted DIMM 4GiB*2 from samsung
- Camera
- pci-e slots
- sata and usb2
- libgfxinit-based graphic init
- NVRAM options for North and South bridges
- Sound
- Thinkpad EC
- S3
- Linux 4.19.67-2 within Debian GNU/Linux stable, loaded from
Linux payload (Heads) and Seabios.
TODO: repurpose and/or rename flag H8_DOCK_EARLY_INIT (introduced in
CB:4294 ) for h8-using devices without a dock.
Change-Id: Ic6a6059ccf15dd2e43ed4fc490c1d3c36aa1e817
Signed-off-by: Bill XIE <persmule@hardenedlinux.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36093
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
This commit creates a dood variant for Octopus. The initial settings
override the baseboard was copied from variant bobba.
BUG=b:141960652
BRANCH=octopus
TEST=emerge-octopus coreboot
Change-Id: Id8852e1f04f4356fac5445f6da6d56d430c88ad0
Signed-off-by: Tim Chen <tim-chen@quanta.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35768
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marco Chen <marcochen@google.com>
Reviewed-by: Karthik Ramasubramanian <kramasub@google.com>
Reviewed-by: Martin Roth <martinroth@google.com>
* Add cbfsoption --ibb to mark files as IBB
* Will be used by "Legacy FIT TXT" boot
Change-Id: I83313f035e7fb7e1eb484b323862522e28cb73d4
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31497
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com>
* Mark files in CBFS as IBB (Initial BootBlock)
* Will be used to identify the IBB by any TEE
Change-Id: Idb4857c894b9ee1edc464c0a1216cdda29937bbd
Signed-off-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/29744
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This corrects the LPC/eSPI lock bit from bit 2 to bit 1 in accordance
with doc#332691-003EN and doc#334819-001.
Change-Id: I45335909b1f2b646e4fafedd78cb1aaf7052d60c
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36100
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
This port is based on NCT6776
Change-Id: Ib8d64e8faa74802ab0213d87881e57d4d9bd1c35
Signed-off-by: Pavel Sayekat <pavelsayekat@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35028
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
FBG-1701 revision 1.3 will use Kingston onboard memory.
Add Kingston SPD file.
When Samsung memory configuration is disabled use
cpld_read_pcb_version() for using correct SPD data.
BUG=N/A
TEST=Boot and verified on Facebook FBG-1701 revision 1.3
Change-Id: I2e1d1b933d5a49a7005685ed530c882429019027
Signed-off-by: Frans Hendriks <fhendriks@eltan.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35792
Reviewed-by: Wim Vervoorn
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Change d3a1a417 "src/superio: Remove unused superio chips" removed
all unused devices except for ones used on mainboards still under
review. The SMSC 1036 was inadvertenly also removed as well. This
device is used in debug cards that may be connected to AMD CRBs.
This patch restores the smsc1036 directory as-is and then corrects
the following lint messages.
* WARNING: Prefer 'unsigned int' to bare use of 'unsigned'
* ERROR: else should follow close brace '}'
* WARNING: braces {} are not necessary for single statement blocks
Change-Id: I851826e12032f802b9b2ff86d5a0eb99871bee6d
Signed-off-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36119
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Move all code involving the cpld to a single file.
Rename mainboard_read_pcb_version() to cpld_read_pcb_version().
BUG=N/A
TEST=tested on fbg1701 board
Change-Id: I9ee9a2c605e8b63baa7d64af92f45aa07e0d9d9e
Signed-off-by: Wim Vervoorn <wvervoorn@eltan.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36095
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The change applies the DPTF parameters.
BUG=b:142849037
TEST=build and verified by thermal team
Change-Id: I5da8d373f38d23929ffec95bc1c9e942f131297f
Signed-off-by: Wisley Chen <wisley.chen@quantatw.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36098
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Fagerburg <pfagerburg@chromium.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
The DRAM calibration blob can be compressed using pre-RAM algorithm
(currently LZ4), which will save ~12ms in boot time.
On Kodama, boot time difference:
Before: 1,082,711
After: 1,070,309
BUG=b:139099592,b:117953502
TEST=build and boot, cbfstool coreboot.rom print -v (see dram compressed)
BRANCH=kukui
Change-Id: Ic3bd49d67ee6f80a0e4d8f6945744642611edf64
Signed-off-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36054
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Chrome OS firmware images have moved bitmap resources from GBB into CBFS
for a long time, so the GBB should only hold firmware keys and HWID,
that is usually less than 10k.
ARM boards usually limit GBB to 0x2f00 (see gru, cheza and kukui) but
many recent x86 simply copy from old settings and may run out of space
when we want to add more resources, for example EC RO software sync.
Note, changing the GBB section (inside RO) implies RO update,
so this change *must not* be cherry-picked back to old firmware
branches if some devices were already shipped.
BRANCH=none
BUG=None
TEST=make # board=darllion,hatch,kahlee,octopus,sarien
Change-Id: I615cd7b53b556019f2d54d0df7ac2723d36ee6cf
Signed-off-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36088
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Mathew King <mathewk@chromium.org>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
The two GPIO functions variant_gpio_table() and
variant_early_gpio_table() provide the pointer to the variants GPIO
table for late and early GPIO init. As these functions are variant
dependent the keyword __weak must not be used as otherwise the linker
might choose the tables from the baseboard.
This patch removes the __weak definition making these functions
overriding the general ones in baseboard/gpio.c.
Change-Id: Ic7fc816d40cb112d7ab51089c3962a77798c08a8
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36094
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
To make data flow more explicit, global variable 'impedance' is replaced
with a local variable, which is passed as a function argument.
BUG=none
BRANCH=kukui
TEST=Krane boots correctly
Change-Id: I0f6dacc33fda013a3476a10d9899821b7297e770
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35766
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
The patch adds config MT8183_DRAM_DVFS to enable DRAM calibration with
multiple frequencies to support DVFS switch.
BUG=b:80501386,b:142358843
BRANCH=kukui
TEST=Boots correctly on Kukui
Change-Id: I97c8e513dc3815a2d62b2904a246a1d8567704a4
Signed-off-by: Yu-Ping Wu <yupingso@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35555
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
This patch supports voltage adjustment for each DRAM frequency, which is
neccesary to support DVFS switch.
BUG=b:80501386,b:142358843
BRANCH=none
TEST=Boots correctly and stress test pass on Kukui.
Change-Id: I9539473ff708f9d0d39eb17bd3fdcb916265d33e
Signed-off-by: Huayang Duan <huayang.duan@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35017
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
DRAM DVFS needs to be calibrated with different vddq voltages to get
correct parameters.
A new API is added to allow changing vddq voltage.
BUG=b:80501386
BRANCH=none
TEST=measure vddq voltage with multimeter
Change-Id: I5f0d82596a1709bf0d37885f257646133f18f210
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35147
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
DRAM DVFS needs to be calibrated with different vdram1 voltages to get
correct parameters.
A new API is added to allow changing vdram1 voltage.
BUG=b:80501386
BRANCH=none
TEST=measure vdram1 voltage with multimeter
Change-Id: Ia15ab3a2e1668e5b4873d317b57a38ebee037709
Signed-off-by: Hsin-Hsiung Wang <hsin-hsiung.wang@mediatek.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/33186
Reviewed-by: Hung-Te Lin <hungte@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Most coreboot debug messages are sent to UART and cbmem console, and we
also want to collect DRAM calibration module output, especially for
cbmem console (so we can see the logs after kernel is up).
Instead of sharing whole cbmem/cbtable/cbmemconsole implementations, we
want to simplify that by a simple function pointer so output can be
preserved by do_putchar, which internally sends data to all registered
consoles (usually cbmem console and UART).
BUG=b:139099592
TEST=make; boots properly for full-k, with and without serial console.
BRANCH=kukui
Change-Id: I1cf16711caf3831e99e17b522b86694524425116
Signed-off-by: Hung-Te Lin <hungte@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36056
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Yu-Ping Wu <yupingso@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Previously if .xcompile was missing, abuild would silently ignore the
error. With https://review.coreboot.org/c/coreboot/+/34241 we now check
the return code so abuild started failing.
We should generate the .xcompile if it doesn't exist. The Makefile will
handle that so we include it as the first Makefile.
We then need to override the default target so we don't use the one from
the Makefile.
BUG=b:112267918
TEST=ran abuild and made sure it generated a .xcompile in the root.
Change-Id: I79ded36d47b0219d0b126adff80a57be1c2bdf07
Signed-off-by: Raul E Rangel <rrangel@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35137
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add an entry for mb/supermicro/x11-lga1151-series and add myself to the
list of maintainers.
Change-Id: I634d251b4323c4f05edd553a9fa82e0f8c53773b
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/36099
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
AGESA and binaryPI set the whole CACHE_ROM_SIZE to WRPROT during the
romstage and do not reference the CONFIG_XIP_ROM_SIZE symbol.
Change-Id: I548b9c9066d825c2f03749353b9990b2efddfd9c
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35825
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested with CONFIG_H8_FN_KEY_AS_VBOOT_RECOVERY_SW to switch between
RW_A and recovery, which works quite well as a developer mode to test
RW_A with the COREBOOT slot as a fallback.
Change-Id: I9d524988e991457032f63a947606d1b3581de5e7
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/35959
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>
Marshall Dawson