This resolves an issue on Shanghai dual CPU configurations where
the APs on node 0 would not start. Single CPU configurations are
unaffected by this issue.
TEST: Booted KFSN4-DRE with dual Opteron 8389 CPUs and verified
proper BSP/AP start and microcode patch levels.
Change-Id: I0f5d4e0e356c6bd64e324b4399ef43b400ecab0c
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8397
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Improve documentation of lock down MSRs in finalize().
Most of these aren't documented in public MSRs.
Change-Id: I4fc47bb9b71bdd7907aae65fc18b419a17ae8547
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Reviewed-on: http://review.coreboot.org/8294
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Peter Stuge <peter@stuge.se>
It doesn't provide any useful information.
Change-Id: I13e68d443bbcadea45b8fbcc262ceb9deb3e2e61
Signed-off-by: Alexander Couzens <lynxis@fe80.eu>
Reviewed-on: http://review.coreboot.org/8380
Tested-by: build bot (Jenkins)
Reviewed-by: Nicolas Reinecke <nr@das-labor.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Peter Stuge <peter@stuge.se>
There is no auto-detection for AGP type and we have no hardware that
selects this. Furthermore, we treat AGP bridges just like PCI bridges,
there is no optimisation for higher bandwidth.
Change-Id: I4fe87c83411643cb9b8d3216f2af07bf098174d3
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8367
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
By adding a description to ME_PATH it becomes visible
and editable in menuconfig.
Change-Id: I8c2f6a30c10f16b19f3667263db02c93688c9f8f
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: http://review.coreboot.org/8398
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
The original firmware has a 2 MiB BIOS region in both 4 MiB and 8 MiB
flash variants. Let's allow using the whole region instead of the
gm45 default of 1 MiB.
Change-Id: I2d8a04bcb992bf2e8e15890a5c6719810b1cf405
Signed-off-by: Michał Masłowski <mtjm@mtjm.eu>
Reviewed-on: http://review.coreboot.org/8392
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Fix a mistake that led to an invalid 0ms latency in the automatically
generated PowerNow! ACPI _PSS objects.
TEST: Booted FreeBSD and Linux and verified correct latency values.
Found-by: Coverity Scan
Change-Id: I03cecab694708136dc555ca2af7ee9a0bf9be5af
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8376
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Because the pointer to the FSP HOB list is now being saved, we can
use that to find the top of usable memory. This eliminates the need
to hardcode the size of the FSP reserved memory area.
Tested on minnowboard max for baytrail.
The HOB structure used does not seem to be present for the rangeley
or ivybridge/pantherpoint FSPs. At the very least, the GUID is not
documented in the integration guides.
Change-Id: I643e57655f55bfada60075b55aad2ce010ec4f67
Signed-off-by: Martin Roth <gaumless@gmail.com>
Reviewed-on: http://review.coreboot.org/8308
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Aaron Durbin <adurbin@google.com>
The "used" attribute was added in commit 27cf2472 which caused these
warnings to start appearing when using the standard coreboot GCC
toolchain:
{standard input}: Assembler messages:
{standard input}:96: Warning: ignoring changed section type for .car.global_data
{standard input}:96: Warning: ignoring changed section attributes for
.car.global_data
The # at the end of the section name causes the assembler to
ignore everything following the name. I verified that the resulting
binaries are the same with and without the #.
Change-Id: Iaac8042533842ed887f33895f083b613a18f496a
Signed-off-by: Martin Roth <gaumless@gmail.com>
Reviewed-on: http://review.coreboot.org/8301
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@google.com>
Now that BROKEN_CAR_MIGRATE is fixed we can stash these in CAR.
Change-Id: I49c31b91f34d415778797d08a347a51dbef797e3
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8024
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <gaumless@gmail.com>
The port is based on the x230 / t530.
Tested - is in active use.
Change-Id: Ic5ccfe70343e8aef3465690edce9cdebf153a44d
Signed-off-by: Nicolas Reinecke <nr@das-labor.org>
Reviewed-on: http://review.coreboot.org/8359
Tested-by: build bot (Jenkins)
Reviewed-by: Philipp Deppenwiese <zaolin@das-labor.org>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
cmos_init() had layers of preprocessor directives, which resulted in
a complete mess. Refactor it to make use of the IS_ENABLED() macro.
This improves readability significantly.
One of the changes is to remove in inline stub declaration of
(get|set)_option. Although that provided the ability for the compiler
to optimize out code when USE_OPTION_TABLE is not selected, there is
no evidence that such savings are measureable.
Change-Id: I07f00084d809adbb55031b2079f71136ade3028e
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8306
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
This macro is controversial for arches where the bits are numbered
MSb first, though we don't support such an arch. We've seen this macro
creep into our tree in different places, so provide it in one place.
Change-Id: I86cd8a16420f34ef31b615aec4e0f7bd3191ca35
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8280
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@gmail.com>
Reviewed-by: David Hendricks <dhendrix@chromium.org>
TEST: Booted KFSN4-DRE with on-board XGI Volari Z9s
Initial text from coreboot appeared, and the Linux
console was displayed immediately at the start of
kernel initialization. After boot was complete
the text mode console continued to behave normally.
SeaBIOS does not currently make use of the legacy
VGA text-mode display.
Change-Id: I2177a1d00e6f07db661dd99fe0184e2c228404d1
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8360
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
The indentations of #define are not consistent in chip.h. Update to make
all #define indentations being aligned and put them after the variable
declaration.
Change-Id: I37550acac18bac3efddb580ef6b956be0e2b357a
Signed-off-by: York Yang <york.yang@intel.com>
Reviewed-on: http://review.coreboot.org/8333
Reviewed-by: Dave Frodin <dave.frodin@se-eng.com>
Tested-by: build bot (Jenkins)
1) Save the pointer to the FSP HOB list to low memory at address 0x614.
This is the same location as CBMEM_RESUME_BACKUP - the two aren't used
in the same platform, so overlapping should be OK. I didn't see any
documentation that actually said that this location was free to use, and
didn't need to be restored after use in S3 resume, but it looks like
the DOS boot vector gets loaded juat above this location, so it SHOULD
be ok. The alternative is to copy the memory out and store it in cbmem
until we're ready to restore it.
2) When a request for the pointer to a CAR variable comes in, pass back
the location inside the FSP hob structure.
3) Skip the memcopy of the CAR Data. The CAR variables do not
get transitioned back into cbmem, but used out of the HOB structure.
4) Remove the BROKEN_CAR_MIGRATE Kconfig option from the FSP platform.
Change-Id: Iaf566dce1b41a3bcb17e4134877f68262b5e113f
Signed-off-by: Martin Roth <gaumless@gmail.com>
Reviewed-on: http://review.coreboot.org/8196
Reviewed-by: Aaron Durbin <adurbin@google.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Add native XGI Z9s framebuffer support to coreboot
XGI initialization code largely taken from Linux 3.18.5
TEST: Booted KFSN4-DRE with XGI Volari Z9s into SeaBIOS
with SeaVGABIOS enabled. Text appeared correctly on screen
and interaction with graphical comboot menu was successful.
However, Linux cleared the framebuffer on boot, rendering the
screen useless until Linux loaded its native xgifb driver.
Change-Id: I606a3892849fc578b0c4d74536aec0a0adef3be3
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8331
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
This copies what was done in southbridge/amd/agesa in:
commit 56f46d8 agesa/family15tn: Switch to per-device ACPI
TEST: amd/lamar.
Change-Id: Id8890ccd4a1ea783ad4740333ae6b061b6bbd7fc
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8288
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
This adds the PCI and interrupt related definitions
for the bolton specific features.
Change-Id: Ia6530c57ec5a4a5c4525bfbae0eb5db04c0bef9e
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8286
Tested-by: build bot (Jenkins)
Reviewed-by: Marc Jones <marc.jones@se-eng.com>
Tests on CPUID are valid regardless of revision.
Change-Id: I5a3a01baca2c0ecfb018ca7965994ba74889a2e2
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8337
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Keep the value that is listed first, it also the one with a more
recent change.
Change-Id: I0336c962544d75f94512563c08f280aa43c7a175
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/8336
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
The amd/lamar mainboard is the first mainboard to use the f81216h.
These changes are needed to fix up commit 27a63d77 so the
Super I/O builds and functions correctly.
- Wrong #include name
- Removed global variable in romstage
- Missing "case" in switch()
Change-Id: I1b2058a915b776664fba14e4341e8a410b50330f
Signed-off-by: Dave Frodin <dave.frodin@se-eng.com>
Reviewed-on: http://review.coreboot.org/8255
Tested-by: build bot (Jenkins)
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Include microcode updates in CBFS for every CPU revision the platform
can support, as changing to different CPU revision should not require
a coreboot rebuild.
This increases CBFS usage from 2 kB to 14 kB.
Change-Id: I6bf90221a688f1a54e49641ce3ba378c5bf659f9
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: http://review.coreboot.org/4521
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Nicolas Reinecke <nr@das-labor.org>
This option is now deperecated by loading microcode updates from cbfs.
Remove this option in anticipation of implementing CBFS loading for
AMD cpus. Removing it beforehand results in less patch overhead.
Change-Id: Ibdef7843db686734e2b6b1568692720fb543b240
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8322
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Use HT MSI Mapping capability register at 0xe0 in CK804 HT device to enable
HT MSI Mapping so that MSIs work. Prior to this change PCIe devices
downstream of the CK804 with MSI enabled would fail to actually assert
their interrupt. Tested on msi/ms7135 and winent/mb6047 running Debian
GNU/Linux 7.0 (wheezy).
Change-Id: I5e0dc8b352f3d04e3b16b899af11d2b908a82850
Signed-off-by: Jonathan A. Kollasch <jakllsch@kollasch.net>
Reviewed-on: http://review.coreboot.org/8276
Tested-by: build bot (Jenkins)
Reviewed-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Although bool normally belongs in stdbool.h, for our use cases,
providing these definitions in stdint.h is acceptable.
Change-Id: I1d0ca1018efacc27d7a4a72aa452912e004401f9
Signed-off-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-on: http://review.coreboot.org/8279
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@gmail.com>
Rangeley POST-GOLD 2 FSP added PCIe ports de-emphasis configuration
by UPD input. Update UPD_DATA_REGION structure for matching up this
FSP change.
PcdCustomerRevision is a debugging aid that will be output to debug
message in FSP. When needed, it can be customized by BCT tool for tracking
BCT configurations.
Change-Id: I6d4138c9d8bbb9c89f24c77f976dbc760d626a9b
Signed-off-by: York Yang <york.yang@intel.com>
Reviewed-on: http://review.coreboot.org/8107
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <gaumless@gmail.com>
This patch clears the CAR area. The FSP loads the entire CAR area with a
pattern instead of clearing it. At least the cbmem area needs to be
cleared or cbmem will not use it.
Change-Id: I829ddc26133353a784dfc01729af9b3bf427e889
Signed-off-by: Martin Roth <gaumless@gmail.com>
Reviewed-on: http://review.coreboot.org/8195
Reviewed-by: Aaron Durbin <adurbin@google.com>
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Test:
Single Opteron 2419 with 1GB RAM in slot A1
Booted Ubuntu Linux 14.04 and verified all 6 cores were visible
Brief stress test of all 6 cores simultaneously
Verified proper ACPI power states for all 6 cores
Change-Id: I1e598e36f9eaed5ba8a18b9c62ceedee16870f15
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8311
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins)
Move mapping tables to struct and prevent OOB array
access that was crashing the APs during CAR initialization.
Change-Id: I9e2554b50ad60a8d02ef4bd3fbee6fddb238d83f
Signed-off-by: Timothy Pearson <tpearson@raptorengineeringinc.com>
Reviewed-on: http://review.coreboot.org/8310
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@gmail.com>
Reviewed-by: Alexandru Gagniuc <mr.nuke.me@gmail.com>
Timothy Pearson