Use global variables to provide mainboard USB settings, and have the
northbridge code copy it into the `pei_data` struct. For now.
To minimize diffstat noise, this patch does not reindent the now-global
mainboard USB configuration arrays. This is cleaned up in a follow-up.
Change-Id: I273c7a6cd46734ae25b95fc11b5e188d63cac32e
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50538
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Was copy-pasted from bd82x6x and no mainboard actually needs it.
The few globals moved outside the GNVS will be removed, relocated or
replaced with acpigen later.
Change-Id: I590a355f1bd1e54365b2e329cfdc62384446a15c
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49280
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Comparing against MRC, looks like the values for TA3 and TA4 are
backwards. All of them. Thus, correct the tables accordingly.
Tested on Acer G43T-AM3, DDR3-1066 and CL = 8 now works.
Change-Id: I2c99502b8f105c77098c888b024a4c3c2c8877d4
Tested-by: Michael Büchler <michael.buechler@posteo.net>
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49388
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Michael Büchler <michael.buechler@posteo.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Fix a typo and do some style improvements.
Change-Id: Ibc7e1869faa6b9ae12a51b1c3d209bbd8e54b0d2
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49402
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Using MCHBAR32_AND_OR() in these two cases changes the order of
additions slightly. Originally, the MCHBAR offset and the base
register offset (0x5a4/0x5b4) were added first. Due to the added
parentheses in the register macros, now the complete register
offset is calculated first and then added to MCHBAR. Associativity
tells us that this doesn't change the result.
Changes in the resulting binary were verified manually on the
object file.
Change-Id: Id10882225c8e82b02583aa73e73d661c25abdef9
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50355
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Clean up cosmetics after refactoring the code. Reflow long lines and
align values in the tables, and also remove a now-unnecessary scope.
Tested with BUILD_TIMELESS=1, Asus P5QL PRO remains identical.
Change-Id: I2712c1ad5404d6968d18d762e6048c5da120ff78
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49400
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
The first RCOMP group (data) is programmed differently, and has its own
tables. Remove the unused first index from the other tables, and adjust
the loop bound accordingly. Cosmetics are cleaned up in a follow-up.
Tested on Asus P5QL PRO (DDR2), still boots.
Change-Id: I3010acbd00f762c91aebeaf1625ed7543b14bf74
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49399
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
The RCOMP data group is special and is programmed differently. Prepare
to simplify the code by programming it outside of the loop. Subsequent
commits will simplify the logic even further, then clean up cosmetics.
The special DDR3 case in the loop overwrites the command group strength
multiplier value. It doesn't need to be programmed for each RCOMP group.
Add a comment to justify not programming this register while programming
the settings for the RCOMP data group.
Tested on Asus P5QL PRO (DDR2), still boots.
Change-Id: I5c2484f48e3c07e8e787b1894932e342e8e8a75c
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49398
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
These settings can be programmed with a single register write. Factor
the writes out into a single function to avoid some redundancy.
Tested on Asus P5QL PRO, still boots.
Change-Id: I3a08c255dd2b0deae650c7fe2ba4e1f4d1cef581
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49396
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
MRC uses an incorrect mask when programming this register, but the reset
default value is zero and it is only programmed once. As it makes no
difference, we can safely use the correct mask. Document this difference
in a comment to indicate the deviation from MRC behavior is intentional.
The default value for this register was dumped from Asus P5QL PRO.
Change-Id: I93b0c382f76e141b319414258e40a8bfe6c7848a
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49391
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Consistently use commas after the last element of arrays, and also align
columns of values and comments. Remove `MHz` units from DDR speed values
to avoid confusion, as the memory's actual clock speed is half of these.
Tested with BUILD_TIMELESS=1, Asus P5QL PRO remains identical.
Change-Id: Id13022483c6221ce87d21dd21a5cfe4317a55ccd
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49390
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Time has proven this statement to be unnecessary. Uncommenting it would
not have any effect on the existing code, thus remove it completely.
Change-Id: Iff4cdd71435e4fd69d4f3284e9fb2830fdd5b173
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49389
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
These accessors can be reused for several other northbridges.
Tested with BUILD_TIMELESS=1, Roda RK9 remains identical.
Change-Id: Ia16ccc63dddebf938f4e9a7f5518e4d25d3e7e66
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49748
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Drop casts to prevent pointer arithmetic and for consistency with other
platforms. These macros will be factored out in a subsequent commit.
Change-Id: I959e7378a8bf46fd1772192090a751d7a2f6f470
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49747
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Guarding the MCHBAR macro breaks reproducibility, but should not have
any functional impact.
Change-Id: I8be8d7b8a0f289d2be76d3dec43999f6b42e3265
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49746
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
It was commented that the need for the delay was mainly related
to external displays and only with VBIOS execution. Move the
delay such that it is done only when we actually need to execute
the VBIOS aka option rom.
A delay is currently only defined for librem/purism_bdw in
its Kconfig. As the description of the issue sounds like it
would equally happen on other platforms when VBIOS is involved,
promote the Kconfig visible option to global scope.
Change-Id: I4503158576f35057373f003586bbf76af4d59b3d
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48787
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Create `FIXED_RCBA_MMIO_BASE` and use it everywhere, except in cases
where a pointer cast would be necessary. Instances in Sandy Bridge MRC
code were left as-is intentionally, so as not to collide with another
cleanup patch train.
Tested with BUILD_TIMELESS=1, these boards remain identical:
- Asus P8Z77-V LX2
- Packard Bell MS2290
Change-Id: I642958fbd6f02dbf54812d6a75d6bc3087acc77a
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50036
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This partially reverts:
- Commit 77d3b655ed
- Commit 487c1a24f5
- Commit 875c21f491
- Commit c4d1b47ad9
- Commit b96c358751
- Commit 9cbf26d18e
It is intentional to use <device/pci_ops.h> whenever one needs to use
PCI config access. The bootblock.c files needing I/O config do not need
to be an exception to this.
Change-Id: Ifba05717dad404a844618815c5347a05e07a3362
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50231
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: HAOUAS Elyes <ehaouas@noos.fr>
Note that bootblock.c originally wrote a reserved bit of the PCIEXBAR
register. The `length` bitfield was set to 0, so assume 256 busses.
Moreover, the ASL reservation for MMCONFIG was only for 64 busses.
Change-Id: I7366a5096aacd92401535be020358447650b4247
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49759
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Angel Pons