Some methods like discover_402x assume an clear state.
Should fix fallback attempt raminit failures.
Change-Id: I7a6fe044c17f5e0dbfa0e9b9d2aed0c3b6ae3972
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17471
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
By shifting the `chan` right instead of left, values were always taken
from the DIMMs of the first channel. The diff-stat also looks like an
improvement.
Change-Id: I605eb4f9b04520c51eea9995a2d4a1f050f02ecc
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/17587
Tested-by: build bot (Jenkins)
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Damien Zammit <damien@zamaudio.com>
If smbus_read_byte returned an error when reading the DIMM size,
this value would be used as an offset into an array.
Check for the error, and set the DIMM size to 0 if there's
a problem.
Addresses coverity issue 1229658 - Negative array index read
Signed-off-by: Martin Roth <martinroth@google.com>
Change-Id: I6461a0fae819dd9261adbb411c4bba07520d076d
Reviewed-on: https://review.coreboot.org/17485
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.h@gmx.de>
RW flag was added to spi_slave structure to get around a requirement on
some AMD flash controllers that need to group together all spi volatile
operations (write/erase). This rw flag is not a property or attribute of
the SPI slave or controller. Thus, instead of saving it in spi_slave
structure, clean up the SPI flash driver interface. This allows
chipsets/mainboards (that require volatile operations to be grouped) to
indicate beginning and end of such grouped operations.
New user APIs are added to allow users to perform probe, read, write,
erase, volatile group begin and end operations. Callbacks defined in
spi_flash structure are expected to be used only by the SPI flash
driver. Any chipset that requires grouping of volatile operations can
select the newly added Kconfig option SPI_FLASH_HAS_VOLATILE_GROUP and
define callbacks for chipset_volatile_group_{begin,end}.
spi_claim_bus/spi_release_bus calls have been removed from the SPI flash
chip drivers which end up calling do_spi_flash_cmd since it already has
required calls for claiming and releasing SPI bus before performing a
read/write operation.
BUG=None
BRANCH=None
TEST=Compiles successfully.
Change-Id: Idfc052e82ec15b6c9fa874cee7a61bd06e923fbf
Signed-off-by: Furquan Shaikh <furquan@chromium.org>
Reviewed-on: https://review.coreboot.org/17462
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Time spent in printk() is highly unpredictable, depending of the
enabled consoles. If only CBMEM console is enabled, debugstring
is repeated tens of times, consuming preram_cbmem_console storage.
Change-Id: I2b0d9bd11c294d988a0eb84b90e77d5cc7f1f848
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17516
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Make MMCONF_SUPPORT selected with MMCONF_SUPPORT_DEFAULT.
Platforms that remain to have explicit MMCONF_SUPPORT are
ones that should be converted.
Change-Id: Iba8824f46842607fb1508aa7d057f8cbf1cd6397
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17527
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
When MRC cache is available, first read only the SPD unique
identifier bytes required to detect possible DIMM replacement.
As this is 11 vs 256 bytes with slow SMBus operations, we save
about 70ms for every installed DIMM on normal boot path.
In the DIMM replacement case this adds some 10ms per installed DIMM
as some SPD gets read twice, but we are on slow RAM training boot path
anyways.
Change-Id: I294a56e7b7562c3dea322c644b21a15abb033870
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17491
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
For S3 resume path SPD is only used for DIMM replacement detection.
As this detection already fails in the case of removal/insertion of
same DIMM, we can rely on cbmem_recovery() failure alone to force
system reset in case someone accidentally does DIMM replacements while
system is suspend-to-ram stage.
Skipping DIMM replacement detection allows skipping slow SPD loading,
thus reducing S3 resume path time by 80ms for every installed DIMM.
Change-Id: I4f2838c05f172d3cb351b027c9b8dd6543ab5944
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17490
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Take the timestamp before SPD loading takes place, for easier
comparison against MRC blob performance and followup changes
will optimize some of the slow SPD/SMBus operations.
Change-Id: I50b5a9d02d2caf4c63e1a4025544131a085b8fb6
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17489
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Switch to use CRC of unique identifier section SPD[117..127],
remaining area of SPD data is ignored.
Change-Id: If4b43183f99f5f911ae6c311b43c29a72b9922e2
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17487
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Also memset info.dimm as it contains decoded SPD timings
used to calculate common timings.
Tested manually on Lenovo T420.
Change-Id: I659e5bc2a6cbadd9539931ee00ddea0a5253295f
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17473
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
No need to find the same CMD rate for all channels.
Allow different CMD rates for every channel.
Tested on Lenovo T420 with different modules on each channel.
No regressions found.
Change-Id: I7036275ae89335dd3549ec392fa64824355b3cbf
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17472
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Use register names found on forums.corsair.com.
No functionality changed.
Change-Id: Ibaede39a24e8df1c4d42cb27986ab66174b7d45b
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17400
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Locking the PLL again once it's locked doesn't work.
The MRC doesn't do this, for some reason.
Remove fallback attempts of lowering DDR frequency.
Change-Id: Iccb54fa7d7357a22182dd26bd5b49c4073c04dc9
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17399
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
As documented in DDR3 spec for MR2 the CWL is based on DDR frequency.
There's no to little difference for most memory modules operating at DDR3-1333.
It might fix problems for memory modules that operate at a higher frequency and
memory modules with low CL values should work even better.
Tested on Lenovo T420 with DDR3-1333 CL9 and DDR3-1600 CL11.
No regressions found.
Change-Id: Ib90b5de872a219cf80b4976b6dfae6bc02e298f4
Signed-off-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-on: https://review.coreboot.org/17389
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Don't use scratchpad registers when we have romstage_handoff
to pass S3 resume flag. Also fixes console log from reporting
early in ramstage "Normal boot" while on S3 resume path.
Change-Id: I5b218ce3046493b92952e47610c41b07efa4d1de
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17455
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Adapt implementation from haswell to prepare for removal of HIGH_MEMORY_SAVE
and moving on to RELOCATABLE_RAMSTAGE. With the change, CBMEM and SMM regions
are set to WRBACK with MTRRs and romstage ram stack is moved to CBMEM.
Also fixes regression of slower S3 resume path after commit
9b99152 intel/sandybridge: Use common ACPI S3 recovery
Skipping low memory backup and using stage cache for ramstage decreases
time spent on S3 resume path by 50 ms on samsung/lumpy.
Change-Id: I2afee3662e73e8e629188258b2f4119e02d60305
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/15790
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
Replace the use of the old device_t definition inside
northbridge/intel/i855.
Change-Id: Iae66d1ef838095a560868d9c9ff81f4208f814f1
Signed-off-by: Antonello Dettori <dev@dettori.io>
Reviewed-on: https://review.coreboot.org/17314
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
This is more consistent with other Intel GMCH code.
Change-Id: I7bfaa79b9031e2dcc5879a607cadacbdd22ebde7
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17405
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.h@gmx.de>
Have a common romstage.c file to prepare CAR stack guards.
MTRR setup around cbmem_top() is somewhat northbridge specific,
place stubs under northbridge for platrform that will move
to RELOCATABLE_RAMSTAGE.
Change-Id: I3d4fe4145894e83e5980dc2a7bbb8a91acecb3c6
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/15762
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Fix regression, S3 resume not working on sandy/ivy after commit
9d6f365 ACPI S3: Remove HIGH_MEMORY_SAVE where possible
There is some 20ms delay with ACPI S3 wakeup time due to MTRR setup
being done after the backup copy. Moving to RELOCATABLE_RAMSTAGE fixes
this delay by removing need of this backup entirely.
Change-Id: Ib72ff914f5dfef8611f5f6cf9687495779013b02
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/15248
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This change may slow down the raminit by maximum 200usec,
but reuses the lapic udelay definition.
Change-Id: I60a68f8a7911b257c0eecda96f7c5bf302bb51ed
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17152
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
No boards left in the tree for this northbridge.
Change-Id: Id45da11b9d78cbd6bd50acb5a3c6c3c270f9020e
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/17281
Tested-by: build bot (Jenkins)
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
These settings should be always made by the firmware, no matter if we
set up graphics or not. It looks like Linux doesn't even know these
registers.
The values are taken from the PRMs for Sandy Bridge and Ivy Bridge [1,
2]. They match the settings that were done in the native graphics path
for Ivy Bridge. I expect the differences to be an update (i.e. the set-
tings we did on the Sandy Bridge path were just outdated). Also, these
settings affect the PCH and not the CPU which are independent from each
other.
[1] Intel® OpenSource HD Graphics Programmer’s Reference Manual (PRM)
Volume 3 Part 3: PCH Display Registers (SandyBridge)
Doc Ref #: IHD-OS-V3 Pt3 – 05 11
https://01.org/sites/default/files/documentation/snb_ihd_os_vol3_part3.pdf
[2] Intel ® OpenSource HD Graphics Programmer’s Reference Manual (PRM)
Volume 3 Part 4: South Display Engine Registers (Ivy Bridge)
Doc Ref #: IHD-OS-V3 Pt 4 – 05 12
https://01.org/sites/default/files/documentation/ivb_ihd_os_vol3_part4.pdf
Change-Id: I83cc90c7558b93273a727f332fb0d8ced47ed70e
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/17073
Tested-by: build bot (Jenkins)
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Those registers are only used on more recent Intel platforms featuring a
PCH. The DP registers on G4X hardware are at a different offset.
Change-Id: I4660e547426ccec0b2095d897e4a8c86e0acf41e
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17111
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.h@gmx.de>
Those registers are only used on more recent Intel platforms featuring a
PCH. The DP registers on G4X hardware are at a different offset.
Change-Id: Ib49e54d4e7d6595dc09fb1be35ac8178b80c7f71
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17110
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.h@gmx.de>
Also drop an odd comment about the resource allocator which seems to
work fine, with the right id.
Change-Id: I9099211fe946c28f90dd7730345b81a3f7f6f545
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/17095
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
This fixes an instability on 945gc where the IGD completely locks
up the system, when for instance tasked to do something with
compositing (like GNOME or GDM).
TESTED on ga-945gcm-s2l and d945gclf
TEST: launch GDM (gnome display manager)
Change-Id: Iec49bccf3e3164df9dc1e0b54460a616fe92e04d
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17094
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins)
The cross clocking of 800MHz FSB CPU with 667MHz RAM was incorrect.
The result is that 800MHz FSB CPUs now properly work with 667MHz RAM.
Value taken from vendor bios on ga-945gcm-s2l and suggested by Haouas
Elyes.
Change-Id: I56c12af50c75a735af0150a4e7bce4faacc93648
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17038
Tested-by: build bot (Jenkins)
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Previously the 945gc raminit only worked for 533MHz FSB CPUs.
This extends the tRD_Mclks in drt0_table for other FSB speeds. The values are
taken from the vendor bios of Gigabyte ga-945gcm-s2l.
The result is that 1067MHz FSB CPUs now boot without problems.
800MHz FSB cpus still don't get past romstage.
Change-Id: I13a6b97d2e580512155edf66c48405a153121957
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17034
Tested-by: build bot (Jenkins)
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Reviewed-by: Nico Huber <nico.h@gmx.de>
This removes writes to FDI related registers since there is no FDI
link on these targets. This is likely a remainder from copying code from
later targets.
TESTED on Thinkpad x200 (gm45)
Change-Id: Id67fdc999185fa184a9ff0e5c3fc9bced04131ad
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16993
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins)
This issue was found by Coverity Scan, CID 1364118.
Change-Id: Iba3c0f4f952729d9e0987d928b63ef8b8fe8841e
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16992
Reviewed-by: Martin Roth <martinroth@google.com>
Tested-by: build bot (Jenkins)
According to: "Intel ® 965 Express Chipset Family and Intel ® G35 Express
Chipset Graphics Controller PR" the p2 divisor needs to be 10 when the
dotclock is below 225MHz and 5 when its above 225MHz.
Change-Id: I363039b6fd92051c4be4fdc88788f27527645944
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16991
Tested-by: build bot (Jenkins)
Reviewed-by: Nico Huber <nico.h@gmx.de>
According to "G45: Volume 3: Display Register Intel ® 965G Express
Chipset Family and Intel ® G35 Express Chipset Graphics Controller" the
VSYNC end should start at bit 16. This is also how Linux (at least 4.4)
sets this register, which can be seen with intel-gpu-tools.
TESTED on Lenovo thinkpad X60 (it does not change anything).
Change-Id: Ie222ac13211a91c4fbc580e2bf9de0d973ea9a3a
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/17015
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Alexander Couzens <lynxis@fe80.eu>
Some devices have no LVDS output but if no VGA is connected or
no EDID can be found, it will try to init LVDS.
This patch detects the presence of an LVDS panel and makes sure that
LVDS is not initialized when it is absent.
Change-Id: Ie15631514535bab6c881c1f52e9edbfb8aaa5db7
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16513
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
This reuses linux code (at least 4.1) to compute the graphic clock
divisors for LVDS displays on the gm45 northbridge.
The divisors m1, m2, n, p1, p2 need to be such that
"BASE_FREQUECY * (5 * (m1 + 2) + (m2 + 2)) / (n + 2)
/ (p1 * p2)" is as close as possible to the target_frequency.
On g4x hardware the BASE_FREQUENCY is 96000kHz.
This potentially increases LVDS display compatibility.
Change-Id: I2323af5756431e89769f95059790f5a922af14b4
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16741
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
MHCBAR(CLKCFG) was previously incorrectly written by the
sdram_program_memory_frequency function which required falsely
limiting the max dram frequency for 945GC.
TESTED on Intel d945gclf (memclock 667 and fsb 533) and
Gigabyte ga-945gcm-s2l (memclock 667 and fsb 1067)
Change-Id: I520efd69fa09fc9fde87c5301fd81121fde6a700
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/16940
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
