The STATESTS register is a very simple read/write-1-clear status
register. OS drivers have to read and clear it all the time with-
out any quirk handling. So it seems unlikely that this sequence
was ever necessary for any coreboot-supported chip.
More likely, that sequence was copied from the dance around the
reset bit when Poulsbo support was added in commit be61a17351
("Support Intel SCH (Poulsbo) and add iwave/iWRainbowG6 board
which uses it.").
TEST= Verbs were loaded correctly on off-tree HP ProBook 450 G3.
Change-Id: I1fbea8ffb71a2fcb4ce5f42b3cb8f816ec336c5b
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89653
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The azalia_set_bits() function does some sequencing specific to the
CRST# bit. Name it accordingly and adapt its signature.
The while loop was also changed to match the other functions (e.g.
wait_for_ready()).
TEST= Verbs were loaded correctly on off-tree HP ProBook 450 G3.
Change-Id: I5ac766f2ddf8b48b436a54469815bf799ae31d52
Signed-off-by: Nico Huber <nico.h@gmx.de>
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89652
Reviewed-by: Alicja Michalska <ahplka19@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The spec[1] says a codec is allowed to take up to 521us before sig-
naling an initialization request. Our original SB600 implementation
had a 1ms delay here since commit 4505948fae ("Use the correct device
for switching on HDA.")
Most codecs are a lot faster, which is probably why nobody noticed
the missing delay. For instance, the Realtek ALC272 datasheet spe-
cifies a 1 frame (1/48kHz) maximum[2]. It doesn't hurt, though, to
be correct here. We have a lot longer delays around.
[1] High Definition Audio Specification 1.0a: "4.3 Codec Discovery"
[2] ALC272 datasheet: "9.2.1. Link Reset and Initialization Timing"
TEST= Verbs were loaded correctly on off-tree HP ProBook 450 G3.
Change-Id: Ifd3357758fb3678e60b4c6edcfbdb60b3bda9746
Signed-off-by: Nico Huber <nico.h@gmx.de>
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89651
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Alicja Michalska <ahplka19@gmail.com>
Parse the supported voltages from the DDR3 SPD and populate the
corresponding fields in CBMEM_ID_MEMINFO to make sure the SMBIOS
type 17 tables report the actual supported voltages of the DIMM.
Change-Id: I35af7c23f285af10b607a80eab7f4d9df664b3fd
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/90395
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The current implementation of the HDA verb table has been shown to have
some problems. The primary issue is that it requires the programmer to
keep track of the amount of verbs that are going to be loaded. While
this may sound simple, in practice there have been numerous cases where
this "count field" has been forgotten to be updated or miscounted. In
the case where the "count field" is incorrect, coreboot will start
looking for codecs in some random memory location, essentially making
loading further codecs impossible.
Another issue is the "count field" treats 4 32-bit values as a single
group, therefore the amount of verbs in the table must be a multiple
of 4. This makes intuitive sense when only using the AZALIA_PIN_CFG() or
AZALIA_SUBVENDOR() macros. However, once the verb table requires
"controls" that use < 4 verbs (e.g. "Coefficient Index"), we need to add
padding values to ensure the alignment is correct. This adds unnecessary
verbs to the table which can further lead to unnecessary processing.
Therefore, in this change we proprose a solution by separating the codec
entries in the verb table into structures, which allows us to separate
the verbs into an array and automatically calculate the "count field"
using the ARRAY_SIZE() macro. It also makes iteration and access to
member fields easier. We also now count the verbs and not 4 32-bit
groups, eliminating the aforementioned alignment issue.
Additionally, this change also changes the way coreboot searches for
entries in the verb table. Before, we searched the table for only a
matching vendor ID, but now we search for a matching vendor ID and
codec address pair. This allows a mainboard to be able to correctly load
multiple audio codecs that use the same chips.
To make reviewing this large rework easier, we temporarily keep both
implementations (legacy and reworked) and allow boards to choose which
implementation to use by selecting a Kconfig. Newer boards are
discouraged from using the legacy implementation, as it is not selected
by default. This allows us to slowly change the codebase instead of
changing everything at once.
TEST=
1. Timeless build with AZALIA_USE_LEGACY_VERB_TABLE=y produces
identical binaries (with INCLUDE_CONFIG_FILE=n)
2. HP ProBook 450 G3 using reworked verb table was able to load all
verbs successfully.
Change-Id: Ib16237de89956405afa3be5b4e3f64a4d62e6a48
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88656
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add a new Kconfig which tells the mainboard is using the "legacy" verb
table implementation. This is only needed during the transition towards
the reworked implementation, and will be removed once completed.
The Kconfigs were added manually and not with a script.
Change-Id: I3806cc8df4e244ee6542ad6796ccd9e36de557e5
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89174
Reviewed-by: Erik van den Bogaert <ebogaert@eltan.com>
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
I found a typo in the comments, so I fixed it.
It seems that 'this operations' is a typographical error.
Change-Id: Ic272e122ce180dee8c0516ecea27cd10932c2363
Signed-off-by: NyeonWoo Kim <knw0507@naver.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/89282
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
One of my previous commits attempted to simplify azalia_codec_init(),
but resulted in illogical code which also introduced a bug in certain
cases.
To summarize, codec_is_operative() tells the controller to get the
vendor ID of a specific codec. While doing so, this also checks how the
controller and codec respond to see if they are functioning. However, we
read the response in azalia_codec_init(). Therefore, these functions
must be called sequentially in order to initialize the codecs correctly.
In certain cases, we would attempt to read the response without
requesting the vendor ID in the first place. This possibly caused
these verbs to not get loaded at all.
These are the areas affected by the bug:
- northbridge/intel/haswell/minihd.c
- soc/intel/broadwell/minihd.c
TEST=Verbs were loaded on HP ProBook 450 G3
Fixes: 516d05f43d ("device/azalia: Separate codec checking and initialization")
Change-Id: I82ada9e6eca0539b854b5bc61f6f7a88ffd1cdc5
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88918
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This is a purely cosmetic change to make things slightly more easier to
read. We also only add the IDs which are actively used in the codebase.
TEST=Timeless build produces identical binaries
Change-Id: I4ec0a570020059c85768bab913dff1ba1977e9f9
Signed-off-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88917
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <patrick.rudolph@9elements.com>
dev->path.gicc_v3.mpidr is an unsigned long long, so the format
specifier should be %llx, not %x. Keep the minimum 2 digit output.
BUG=CID 1611971
Change-Id: I126b0281efcba2c3e41cf6da4d006b8d2eb7215b
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88769
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Rename config options 'USE_DDRx' to 'DRAM_SUPPORT_DDRx' to make them
less clunky, and in preparation to expand their use inside SoC code.
Change-Id: Ie6edd730c5cbad679a90fcf7989a942d9b2dd3d8
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88520
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: <yuchi.chen@intel.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
This is fairly intuitive upon thinking about it, SeaBIOS has neither
long mode nor PAE page tables, but make it obvious to developers,
and let users know this.
Change-Id: I769c1bdb9d7ea78d56455d125adf3d9bf07a1211
Signed-off-by: Benjamin Doron <benjamin.doron@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/88453
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
I've found some TODOs in comment in device_util.c,
so I replaced them with descriptions for readability.
Change-Id: I429ce0b2a1b56a60fdb4127591ec219768c2c044
Signed-off-by: NyeonWoo Kim <knw0507@naver.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/87907
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This ensures that bridge windows allocate enough space to cover
SR-IOV BARs. Without this Linux will print messages, these messages
may differ depending on the kernel version used.
Debian GNU/Linux 12 (kernel 6.1.0-28-amd64):
pci 0000:06:00.0: BAR 7: no space for [mem size 0x00200000 64bit pref]
pci 0000:06:00.0: BAR 7: failed to assign [mem size 0x00200000 64bit
pref]
Ubuntu 22.04.5 LTS (kernel 6.8.0-52-generic):
pci 0000:06:00.0: VF BAR 0 [mem size 0x00200000 64bit pref]: can't
assign; no space
pci 0000:06:00.0: VF BAR 0 [mem size 0x00200000 64bit pref]: failed to
assign
TEST=Raptorlake-P
Change-Id: Ib169efe5a6b998a8342a895f1456a280669c719d
Signed-off-by: Stephen Douthit <stephend@silicom-usa.com>
Signed-off-by: Harrie Paijmans <hpaijmans@eltan.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/34620
Reviewed-by: Frans Hendriks <fhendriks@eltan.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Currently there are 3 GDTs (Global Descriptor Tables) being used on x86:
- preRAM (gdt_init.S)
- SMM (smm_stub.S)
- RAM (c_start.S)
They have different layouts and thus different offsets for the segments
being used in assembly code. Stop using different GDT segments and
ensure that for ROM (preRAM + SMM) and RAM (ramstage) the segments
match. RAM will have additional entries, not found in pre RAM GDT,
but the segments for protected mode and 64-bit mode now match in
all stages.
This allows to use the same defines in all stages. It also drops the
need to know in which stage the code is compiled and it's no longer
necessary to switch the code segment between stages.
While at it fix the comments in the ramstage GDT and drop unused
declarations from header files, always set the accessed bit and drop
GDT_CODE_ACPI_SEG.
Change-Id: I208496e6e4cc82833636f4f42503b44b0d702b9e
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/87255
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Commit a959f0ad76 ("allocator_v4: Disable top-down allocation for
EDK2") disabled top-down allocation for edk2 due to a bug which broke
display init with Intel IGD. A workaround has been implemented in
MrChromebox's fork (and others) while the issue is being resolved
upstream, so re-enable top-down allocation unless upstream edk2
is being used.
TEST=build/boot various Google mainboards with edk2 payload selected.
Change-Id: I0e9b0d02bbf30878aef37a97d6a743a402700fc7
Signed-off-by: Matt DeVillier <matt.devillier@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/87504
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Sean Rhodes <sean@starlabs.systems>
Stop using magic values and use defines for Global Descriptor Table
(GDT) offsets. Use the existing defines from the corresponding headers.
Change-Id: I40c15f6341bdef9cd457619ec81e7ac624ec2d63
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/87254
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Previously only tested on server platforms - it wasn't working correctly
on consumer platforms due to missing boolean.
This patch fixes it, which makes resource allocator use uint64 instead
of uint32. Thanks to that, modern GPUs like Intel Arc or Radeon RX now
work correctly with ReBAR enabled, and correctly initialize the
framebuffer in payload (i.e EDK2) after initializing the OpROMs.
Example of issue caused by resource allocator using uint32 (Intel Arc
A580):
[ERROR] Resource didn't fit!!!
[ERROR] PCI: 00:01:00.0 10 prefmem64 size: 0x0000800000 not assigned
[ERROR] PCI: 00:03:00.0 18 prefmem64 size: 0x0200000000 not assigned
(Followed by Linux reporting that BAR space was limited to 256MB, which
severely hindered the performance).
TESTed on Intel Tiger Lake-H (mb/erying/tgl) with Intel Arc A580 and AMD
Radeon RX7800XT.
Change-Id: Ia17b3312016409d8fd6bcce4321481a7b7e35ce5
Signed-off-by: Alicja Michalska <alicja.michalska@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86131
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Maximilian Brune <maximilian.brune@9elements.com>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
The arch include files are overshadowed by PSP verstage include files.
The reason is that psp_verstage implements its own set of inb() and
outb() functions, which use a runtime configurable IO base address
instead of a built time constant.
But this works at the moment only because of the order in which the
include files are added. Since that is very error prone, this patch
introduces another solution to the problem.
Signed-off-by: Maximilian Brune <maximilian.brune@9elements.com>
Change-Id: I16fa4a4cb5168024aaef30119e9aa8a34dbaacbe
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86874
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Move PCI Option ROM handling code into device/pci_rom.c as it's
already using a majority of functions within this file.
Change-Id: I50fc3bf45a1ab6572ab031b9e24ca2f882a13aad
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86733
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Ana Carolina Cabral <ana.cpmelo95@gmail.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Recent AMD iGPUs are not VGA compatible anymore, thus they don't
identify themself as "VGA compatible" anymore by the PCI class code.
Since the PCI VGA Option ROM code assumes it only runs on VGA
compatible devices, relax the ACPI code part to handle display devices
as well. In order to run a VBIOS in coreboot it still must be VGA
compatible, but for ACPI table generation, where no code is run, it's
not necessary any more.
The new code allows to use Linux's amdgpu driver on AMD/glinda.
TEST: On amd/birman+ the amdgpu kernel drivers starts and dmesg shows:
[ 3.010224] [drm] amdgpu kernel modesetting enabled.
The coreboot log shows:
[INFO ] CBFS: Found 'pci1002,150e.rom' @0x10a40 size 0x4400 in mcache @0x1b7dd184
[DEBUG] In CBFS, ROM address for PCI: 00:02:00.0 = 0xff012a6c
[DEBUG] Class Code mismatch ROM 00030000, dev 00038000
[DEBUG] Copying non-VGA ROM image from 0xff012a6c to 0x000d0000, 0x4400 bytes
[...]
Copying initialized VBIOS image from 0x000d0000
[DEBUG] ACPI: * VFCT at 1b5cb960
Change-Id: I623cd80b45b148b91f2796b22a589bbede0feeeb
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86386
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Currently the VBIOS is placed somewhere in DRAM when necessary.
While generating ACPI tables the code attempts to find the VBIOS
by looking at "known" places.
Simplify the code and keep track of the VBIOS using a pointer in
struct device by filling it in pci_rom_load().
The following patches will reuse this pointer to generalize the
code even more.
Change-Id: Ib27d30e3b07740d6098d4f7a5c4f5d8bce976f00
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86385
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Do not assume that a VBIOS has been run when loaded below 1MiB.
On recent AMD platforms the VBIOS is loaded into the C/D-segment,
but it's not run as CONFIG_VGA_ROM_RUN is not set.
Since commit 5f5aa79 "device/pci_rom: Move VBIOS checksum fix" the
VBIOS has a valid checksum in ati_rom_acpi_fill_vfct(), thus it's
not possible to tell if it has been run or ATOMBIOS tables have
been modified.
Update the debug message to avoid confusion.
Change-Id: I63289ecf2c212f3d95e022e8c47dcd0ac610d970
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86732
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Before I got the following error:
[ERROR] DDR5 speed of 3750 MHz is out of range
tested: glinda based mainboard
Change-Id: I141f63c4fc505a9e16eed132a9a550441f4ad68d
Signed-off-by: Maximilian Brune <maximilian.brune@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86543
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Alicja Michalska <ahplka19@gmail.com>
Reviewed-by: Andy Ebrahiem <ahmet.ebrahiem@9elements.com>
Reviewed-by: Marvin Drees <marvin.drees@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Naresh Solanki <naresh.solanki@9elements.com>
Move the VBIOS checksum code into the soc/amd folder, as it's
specific to AMD's FSP. The code now fixes the VBIOS in place
instead only fixing it for the VFCT table.
TEST: VBIOS has correct checksum after loading in BS_DEV_RESOURCES.
VBIOS checksum is invalid entering graphics_dev_init().
VBIOS checksum is correct leaving graphics_dev_init().
Change-Id: I63aaaefaf01ea456e2ed39cd0891e552a7fb5135
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86384
Reviewed-by: Ana Carolina Cabral <ana.cpmelo95@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
The Option ROM contains lots of 16bit values that are being used,
thus use the 16bit endianness conversion function over the 32bit
variant to avoid confusion.
TEST: Still works on amd/birman+.
Change-Id: I571be97a930ad018e1d1316117cefe5bd1c68f9b
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/86383
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Andy Ebrahiem <ahmet.ebrahiem@9elements.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Increase the char buffer size to fit all characters that are printed
into it by the snprintf() call below.
Change-Id: Ib153e1d02a08b2551dad5b51c4c88bf0bb606af3
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85807
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nicholas Chin <nic.c3.14@gmail.com>
Match function name "init" with "remove" by renaming all
*_breakpoint_disable() to *_breakpoint_remove().
Change-Id: Id3da25cfa6fc0594887f3112e269e57e8ecb32b3
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85540
Reviewed-by: Jérémy Compostella <jeremy.compostella@intel.com>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
On platforms with multiple IOAPICs the GSI base must not be
linear, which is currently assumed by acpi_create_madt_ioapic_from_hw().
Integrate the existing struct device DEVICE_PATH_IOAPIC type and allow
to assign custom GSI bases for each IOAPIC. Write out the IOAPIC devices
into the MADT table if any.
For now, since no platform adds IOAPIC devices, the existing behaviour
remains the same. Allows to get rid of soc_get_ioapic_info().
Change-Id: Ie13d4f5c4f0704f0935974f90e5b7cf24e94aab3
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85226
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Add const qualifier for input of dev_is_active_bridge so that
dev_is_active_bridge could be used for both struct device * input
and const struct device * input.
TESTED=Build and boot on intel/avenuecity CRB
Change-Id: Ia4231534c87cd13d4e6e4d606733f9eb11221ac1
Signed-off-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/85150
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Add a helper function to identify PCI IOAPICs.
Will be used in the following commits.
Change-Id: Ibe50934260b025575440fd52eace73fe2327a193
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/84849
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Add a helper method to read the PME capability.
Will be used in the following commit.
Change-Id: Id1fdc98c9ce86d3ddf8056bb609afc58008cf2e9
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/84793
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Subrata Banik <subratabanik@google.com>
Add and use a new helper function to determine if a device is
1) a PCIe device
2) it's mark hot-plug capable
Change-Id: I61cc013844024b43808cd2f054310cb6676ba69e
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/84792
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
The spec tells us to clear the busy bit manually after a timeout. Do
that and wait immediately, to detect further issues early. Also fix
some related comments and prints: Failures shouldn't be debug messa-
ges. And we are talking to the PIO interface of the controller, not
the codec. So this was never about the codec being ready.
Change-Id: I4b737f8259157c01bfcd9e6631cc15d39c653d06
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83592
Reviewed-by: Nicholas Sudsgaard <devel+coreboot@nsudsgaard.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This allows to configure a default screen rotation in 90-degree
steps. The framebuffer contents will then be displayed rotated,
by the same amount in the other direction; i.e. if you turn the
screen to the left, the picture has to be rotated to the right
to accommodate.
This is only supported by libgfxinit from Skylake / Apollo Lake
on (earlier GPUs didn't support the 90-degree steps anyway) and
it only works with the linear-framebuffer option.
Change-Id: Iac75cefbd34f28c55ec20ee152fe67351cc48653
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/38922
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Rename the 'domain' element of the 'domain_path' struct to 'domain_id'
to clarify that this element is the domain ID.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Suggested-by: Martin Roth <gaumless@gmail.com>
Change-Id: I3995deb83a669699434f0073aed0e12b688bf6e7
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83677
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
To avoid having constructs like 'dev->path.domain.domain' in the SoC
code, create the 'dev_get_domain_id' helper function that returns the
domain ID of either that device if it's a domain device or the
corresponding domain device's domain ID, and use it in the code.
If this function is called with a device other than PCI or domain type,
it won't have a domain number. In order to not need to call 'die',
'dev_get_domain_id' will print an error and return 0 which is a valid
domain number. In that case, the calling code should be fixed.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I3d79f19846cea49609f848a4c42747ac1052c288
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83644
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Move is_domain0 and is_dev_on_domain0 from the Intel Xeon SP code to the
common coreboot code so that it can be used elsewhere in coreboot too,
and while moving also implement it as functions instead of macros which
is more in line with the rest of helper functions in that new file.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I954251ebc82802c77bf897dfa2db54aa10bc5ac4
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83642
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Shuo Liu <shuo.liu@intel.com>
Reviewed-by: Paul Menzel <paulepanter@mailbox.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
In the HD Audio Specification Rev. 1.0a, every bitfield in the GCAP
register is RO (Read Only). However, it is known that in some Intel
PCHs (e.g 6-series and 7-series, documents 324645 and 326776), some
of the bitfields in the GCAP register are R/WO (Read / Write Once).
GCAP is RO on 5-series PCHs; 8-series and 9-series PCHs have a lock
bit for GCAP elsewhere.
Lock GCAP by reading GCAP and writing back the same value. This has
no effect on platforms that implement GCAP as a RO register or lock
GCAP through a different mechanism.
Change-Id: Id61e6976a455273e8c681dbeb4bad35d57b1a8a2
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83218
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Use the existing `azalia_enter_reset()` function instead of explicitly
clearing the bit (and having to explain in a comment what this means).
Change-Id: I04924e68420a93a1ad46f5a7ab359e38c0f7e210
Signed-off-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83217
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Reviewed-by: Elyes Haouas <ehaouas@noos.fr>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
The QEMU Bochs display driver and the QEMU Firmware Configuration
interface code (in the qemu-i440fx mainboard dir) were written for x86.
These devices are available in QEMU VMs of other architectures as well,
so we want to port them to be independent from x86.
The main problem is that the drivers use x86 port I/O functions to
communicate with devices over PCI I/O space. These are currently not
available for ARM* and RISC-V, although it is often still possible to
access PCI I/O ports over MMIO through a translator.
Add implementations of port I/O functions that work with PCI I/O space
on these architectures as well, assuming there is such a translator at a
known address configured at build-time.
Change-Id: If7d9177283e8c692088ba8e30d6dfe52623c8cb9
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80372
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
