Fix booting issues on google/kahlee introduced by CB:51723.
Update use inital apic id in smm_stub.S to support xapic mode error.
Check more bits(LAPIC_BASE_MSR BIT10 and BIT11) for x2apic mode.
TEST=Boot to OS and check apicid, debug log for CPUIDs
cpuid_ebx(1), cpuid_ext(0xb, 0), cpuid_edx(0xb) etc
Signed-off-by: Wonkyu Kim <wonkyu.kim@intel.com>
Change-Id: Ia28f60a077182c3753f6ba9fbdd141f951d39b37
Reviewed-on: https://review.coreboot.org/c/coreboot/+/52696
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The existing helpers for reading/writing MSRs (rdmsr, wrmsr) require use
of the struct `msr_t`, which splits the MSR value into two 32 bit parts.
In many cases, where simple 32 bit or 64 bit values are written, this
bloats the code by unnecessarly having to use that struct.
Thus, introduce the helpers `msr_read` and `msr_write`, which take or
return `uint64_t` values, so the code condenses to a single line or two,
without having to deal with `msr_t`.
Example 1:
~~~
msr_t msr = {
.lo = read32((void *)(uintptr_t)0xfed30880),
.hi = 0,
};
msr.lo |= 1;
wrmsr(0x123, msr);
~~~
becomes
~~~
uint32_t foo = read32((void *)(uintptr_t)0xfed30880);
msr_write(0x123, foo | 1)
~~~
Example 2:
~~~
msr_t msr = rdmsr(0xff);
uint64_t msr_val = (msr.hi << 32) | msr.lo;
~~~
becomes
~~~
uint64_t msr_val = msr_read(0xff);
~~~
Change-Id: I27333a4bdfe3c8cebfe49a16a4f1a066f558c4ce
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/52548
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Use the same stack location during relocation as for the permanent
handler.
When the number of CPUs is too large the stacks during relocation
don't fit inside the default SMRAM segment at 0x30000. Currently the
code would just let the CPU stack base grow downwards outside of the
default SMM segment which would corrupt lower memory if S3 is
implemented.
Also update the comment on smm_module_setup_stub().
Change-Id: I6a0a890e8b1c2408301564c22772032cfee4d296
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51186
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Implement x2apic mode as existing code only supports apic mode.
Use info from LAPIC_BASE_MSR (LAPIC_BASE_MSR_X2APIC_MODE) to check
if apic mode or x2apic mode and implement x2apic mode according to
x2apic specfication.
Reference:
https://software.intel.com/content/www/us/en/develop/download/intel-64-architecture-x2apic-specification.html
BUG=None
BRANCH=None
TEST=boot to OS and check apic mode
cat /proc/cpuinfo | grep "apicid"
ex) can see apicid bigger than 255
apicid : 256
apicid : 260
Signed-off-by: Wonkyu Kim <wonkyu.kim@intel.com>
Change-Id: I0bb729b0521fb9dc38b7981014755daeaf9ca817
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51723
Reviewed-by: Ravishankar Sarawadi <ravishankar.sarawadi@intel.com>
Reviewed-by: Jamie Ryu <jamie.m.ryu@intel.com>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The permanent handler module argument 'save_state_size' now holds the
meaning of the real save state size which is then substracted from the
CPUs save state 'top' to get the save state base.
TESTED with qemu Q35 on x86_64 where the stub size exceeds the AMD64
save state size.
Change-Id: I55d7611a17b6d0a39aee1c56318539232a9bb781
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50770
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Remove variables that are either constants or are just assigned but
not used.
Change-Id: I5d291a3464f30fc5d9f4b7233bde575010275973
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50784
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
With the smm_module_loaderv2 the save state map is not linear so copy
a map from ramstage into the smihandler.
TESTED on QEMU q35: Both SMMLOADER V1 and V2 handle save states properly.
Change-Id: I31c57b59559ad4ee98500d83969424e5345881ee
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50769
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The parameters that the permanent handler requires are pushed directly
to the permanent handlers relocatable module params.
The paremeters that the relocation handler requires are not passed on
via arguments but are copied inside the ramstage. This is ok as the
relocation handler calls into ramstage.
Change-Id: Ice311d05e2eb0e95122312511d83683d7f0dee58
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50767
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
struct smm_loader_params is a struct that is passed around in the
ramstage code to set up either the relocation handler or the permanent
handler. At the moment no parameters in the stub 'smm_runtime' are
referenced so it can be dropped. The purpose is to drop the
smm_runtime struct from the stub as it is already located in the
permanent handler.
Change-Id: I09c1b649b5991f55b5ccf57f22e4a3ad4c9e4f03
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50766
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Keep a copy of start32_offset into ramstage to avoid needing to pass
arguments, calling from assembly. Doing this in C code is better than
assembly.
Change-Id: Iac04358e377026f45293bbee03e30d792df407fd
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50765
Reviewed-by: Eugene Myers <cedarhouse1@comcast.net>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This is only consumed by the stub and not by the relocation handler or
the permanent handler, so move it out of the runtime struct.
Change-Id: I01ed0a412c23c8a82d88408be058a27e55d0dc4d
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50762
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
These stub params need to be synced with the code in smm_stub.S and
are consumed by both the smmloader and smmloader_v2. So it is better
to have the definition located in one place.
Change-Id: Ide3e0cb6dea3359fa9ae660eab627499832817c9
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50761
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
CB:49896 added support in `intel_microcode_find()` to cache the found
microcode for faster subsequent accesses. This works okay when the
function succeeds in finding the microcode on BSP. However, if for any
reason, `cpu_microcode_blob.bin` does not contain a valid microcode
for the given processor, then the logic ends up attempting to find
microcode again and again every time it is called (because
`ucode_updates` is set to NULL on failed find, thus retriggering the
whole find sequence every time). This leads to a weird race condition
when multiple APs are running in parallel and executing this
function.
A snippet of the issues observed in the scenario described above:
```
...
microcode: Update skipped, already up-to-date
...
Microcode header corrupted!
...
```
1. AP reports that microcode update is being skipped since the current
version matches the version in CBFS (even though there is no matching
microcode update in CBFS).
2. AP reports microcode header is corrupted because it thinks that the
data size reported in the microcode is larger than the file read from
CBFS.
Above issues occur because each time an AP calls
`intel_microcode_find()`, it might end up seeing some intermittent
state of `ucode_updates` and taking incorrect action.
This change fixes this race condition by separating the logic for
finding microcode into an internal function `find_cbfs_microcode()`
and maintaining the caching logic in `intel_microcode_find()` using a
boolean flag `microcode_checked`.
BUG=b:182232187
TEST=Verified that `intel_microcode_find()` no longer makes repeated
attempts to find microcode from CBFS if it failed the first time.
Change-Id: I8600c830ba029e5cb9c0d7e0f1af18d87c61ad3a
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51371
Reviewed-by: Patrick Rudolph
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add a helper function mp_run_on_all_aps, it allows running a given
func on all APs excluding the BSP, with an added provision to run
func in serial manner per AP.
BUG=b:169114674
Signed-off-by: Aamir Bohra <aamir.bohra@intel.com>
Change-Id: I74ee8168eb6380e346590f2575350e0a6b73856e
Reviewed-on: https://review.coreboot.org/c/coreboot/+/51271
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
The new name is more consistent with the rest of the MSR definitions.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I5666d9837c61881639b5f292553a728e49c5ceb2
Reviewed-on: https://review.coreboot.org/c/coreboot/+/50855
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Add code to collect all required information and generate ACPI CRAT
table entries. Publish tables generated from cb, rather than use the
tables created by FSP binary.
BUG=b:155307433
TEST=Boot trembyle and compare coreboot generated tables with tables
that FSP published previously.
BRANCH=Zork
Change-Id: If64fd624597b2ced014ba7f0332a6a48143c0e8c
Signed-off-by: Jason Glenesk <jason.glenesk@amd.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/47727
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
This MSR will be used in the follow-up changes.
Change-Id: Ia6f74861502d4a9f872b2bbbab2e5f1925a14c4d
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/49044
Reviewed-by: Lance Zhao
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
All platforms moved to initialise GNVS at the time
of SMM module loading.
Change-Id: I31b5652a946b0d9bd1909ff8bde53b43e06e2cd9
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48699
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Move CPUID definitions out of msr.h into new cpuid.h header.
BUG=b:155307433
BRANCH=Zork
Change-Id: I2ed5e0a5a6dbdb38fce8bf3e769f680330718653
Signed-off-by: Jason Glenesk <jason.glenesk@amd.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/48533
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
It is too easy to confuse those with IA32_SMRR_PHYS_x registers.
Change-Id: Ice02ab6c0315a2be14ef110ede506262e3c0a4d5
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46896
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This prototype will be used outside of soc/amd.
Change-Id: Icc69cf8a910764b27edf64f0f527b8f6a9013121
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45813
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Currently coreboot has limited use for the SMM save state. Typically
the only thing needed is to get or set a few registers and to know
which CPU triggered the SMI (typically via an IO write). Abstracting
away different SMM save states would allow to put some SMM
functionality like the SMMSTORE entry in common places.
To save place platforms can select different SMM save sate ops that
should be implemented. For instance AMD platforms don't need Intel SMM
save state handling.
Some platforms can encounter CPUs with different save states, which
the code then handles at runtime by comparing the SMM save state
revision which is located at the same offset for all SMM save state
types.
Change-Id: I4a31d05c09065543424a9010ac434dde0dfb5836
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44323
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Clarify what the function does by renaming it from do_lapic_init() to
lapic_virtual_wire_mode_init().
Change-Id: Ie4430bf0f6c6bf0081b6aaeace351092bcf7f4ac
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/47020
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Marshall Dawson <marshalldawson3rd@gmail.com>
Reviewed-by: Nico Huber <nico.h@gmx.de>
The "Nominal Performance" is not the same as the "Guaranteed
Performance", but is defined as the performance a processor can deliver
continously under ideal environmental conditions.
According to edk2, this is the "Maximum Non-Turbo Ratio", which needs to
be read from MSR_PLATFORM_INFO instead of IA32_HWP_CAPABILITIES.
Correct the entry in the CPPC package.
Test: dumped SSDT from Supermicro X11SSM-F and checked decompiled
version
Change-Id: Ic2c27fd3e14af18aa4101c0acd7a5ede15d1f3a9
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46464
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Rework the code moved to common code in CB:46274. This involves
simplification by using appropriate helpers for MSR and CPUID, using
macros instead of plain values for MSRs and cpu features and adding
documentation to the header.
Change-Id: I7615fc26625c44931577216ea42f0a733b99e131
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46588
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Move a whole bunch of copy-pasta code from soc/intel/{bdw,skl,cnl,icl,
tgl,ehl,jsl,adl} and cpu/intel/{hsw,model_*} to cpu/intel/common.
This change just moves the code. Rework is done in CB:46588.
Change-Id: Ib0cc834de8492d59c423317598e1c11847a0b1ab
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46274
Reviewed-by: Felix Singer <felixsinger@posteo.net>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Deduplicate code by using the new common cpu code implementation of
AES-NI locking.
Change-Id: I7ab2d3839ecb758335ef8cc6a0c0c7103db0fa50
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46278
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Simplify the AES-NI code by using msr_set and correct the comment.
Change-Id: Ib2cda433bbec0192277839c02a1862b8f41340cb
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46275
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Copy the AES-NI locking function to common cpu code to be able to reuse
it.
This change only copies the code and adds the MSR header file. Any
further rework and later deduplication on the platforms code is done in
the follow-up changes.
Change-Id: I81ad5c0d4797b139435c57d3af0a95db94a5c15e
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46272
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
msr_set_bit can only set single bits in MSRs and causes mixing of bit
positions and bitmasks in the MSR header files. Thus, replace the helper
by versions which can unset and set whole MSR bitmasks, just like the
"and-or"-helper, but in the way commit 64a6b6c was done (inversion done
in the helper). This helps keeping the MSR macros unified in bitmask
style.
In sum, the three helpers msr_set, msr_unset and msr_unset_and_set get
added.
The few uses of msr_set_bit have been replaced by the new version, while
the used macros have been converted accordingly.
Change-Id: Idfe9b66e7cfe78ec295a44a2a193f530349f7689
Signed-off-by: Michael Niewöhner <foss@mniewoehner.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/46354
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This allows to remove some assembly code.
Tested with QEMU Q35 to still print the revision correctly.
Change-Id: I36fb0e8bb1f46806b11ef8102ce74c0d10fd3927
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44319
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The used assembler code only works on x86_32, but not on x86_64.
Use the inline functions to provide valid rdtsc readings on both
x86_32 and x86_64.
Tested on Lenovo T410 with additional x86_64 patches.
Change-Id: Icf706d6fb751372651e5e56d1856ddad688d9fa3
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45702
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Add code to generate p-state and c-state SSDT objects to coreboot.
Publish objects generated in native coreboot, rather than the ones
created by FSP binary.
BUG=b:155307433
TEST=Boot morphius to shell and extract and compare objects created in
coreboot with tables generated by FSP. Confirm they are equivalent.
BRANCH=Zork
Change-Id: I5f4db3c0c2048ea1d6c6ce55f5e252cb15598514
Signed-off-by: Jason Glenesk <jason.glenesk@amd.corp-partner.google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/45340
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Update the COS mask calculation to accomodate the RW data as per SoC
configuration. Currently only one way is allocated for RW data and
configured for non-eviction. For earlier platform this served fine,
and could accomodate a RW data up to 256Kb. Starting TGL and JSL, the
DCACHE_RAM_SIZE is configured for 512Kb, which cannot be mapped to a
single way. Hence update the number of ways to be configured for non-
eviction as per total LLC size.
The total LLC size/ number of ways gives the way size. DCACHE_RAM_SIZE/
way size gives the number of ways that need to be configured for non-
eviction, instead of harcoding it to 1.
TGL uses MSR IA32_CR_SF_QOS_MASK_1(0x1891) and IA32_CR_SF_QOS_MASK_2(0x1892)
as COS mask selection register and hence needs to be progarmmed accordingly.
Also JSL and TGL platforms the COS mask selection is mapped to bit 32:33
of MSR IA32_PQR_ASSOC(0xC8F) and need to be updated in edx(maps 63:32)
before MSR write instead of eax(maps 31:0). This implementation corrects
that as well.
BUG=b:149273819
TEST= Boot waddledoo(JSL), hatch(CML), Volteer(TGL)with NEM enhanced
CAR configuration.
Signed-off-by: Aamir Bohra <aamir.bohra@intel.com>
Change-Id: I54e047161853bfc70516c1d607aa479e68836d04
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43494
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Shreesh Chhabbi <shreesh.chhabbi@intel.corp-partner.google.com>
Reviewed-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
The IA32_SMRR_PHYS_MASK MSR contains a 'Lock' bit, which will cause the
core to generate a #GP if the SMRR_BASE or SMRR_MASK registers are
written to after the Lock bit is set; this is helpful with securing SMM.
BUG=b:164489598
Signed-off-by: Tim Wawrzynczak <twawrzynczak@chromium.org>
Change-Id: I784d1d1abec0a0fe0ee267118d084ac594a51647
Reviewed-on: https://review.coreboot.org/c/coreboot/+/44991
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Caveh Jalali <caveh@chromium.org>
Reviewed-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Enable long mode in SMM handler.
x86_32 isn't affected by this change.
* Enter long mode
* Add 64bit entry to GDT
* Use x86_64 SysV ABI calling conventions for C code entry
* Change smm_module_params' cpu to size_t as 'push' is native integer
* Drop to protected mode after c handler
NOTE: This commit does NOT introduce a new security model. It uses the
same page tables as the remaining firmware does.
This can be a security risk if someone is able to manipulate the
page tables stored in ROM at runtime. USE FOR TESTING ONLY!
Tested on Lenovo T410 with additional x86_64 patches.
Change-Id: I26300492e4be62ddd5d80525022c758a019d63a1
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/37392
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-by: Eugene Myers <cedarhouse1@comcast.net>
INVD is called below so if postcar is running in a cached environment
it needs to happen.
NOTE: postcar cannot execute in a cached environment if clflush is not
supported!
Change-Id: I37681ee1f1d2ae5f9dd824b5baf7b23b2883b1dc
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/37212
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
Xeon-SP Skylake Scalable Processor can have 36 CPU threads (18 cores).
Current coreboot SMM is unable to handle more than ~32 CPU threads.
This patch introduces a version 2 of the SMM module loader which
addresses this problem. Having two versions of the SMM module loader
prevents any issues to current projects. Future Xeon-SP products will
be using this version of the SMM loader. Subsequent patches will
enable board specific functionality for Xeon-SP.
The reason for moving to version 2 is the state save area begins to
encroach upon the SMI handling code when more than 32 CPU threads are
in the system. This can cause system hangs, reboots, etc. The second
change is related to staggered entry points with simple near jumps. In
the current loader, near jumps will not work because the CPU is jumping
within the same code segment. In version 2, "far" address jumps are
necessary therefore protected mode must be enabled first. The SMM
layout and how the CPUs are staggered are documented in the code.
By making the modifications above, this allows the smm module loader to
expand easily as more CPU threads are added.
TEST=build for Tiogapass platform under OCP mainboard. Enable the
following in Kconfig.
select CPU_INTEL_COMMON_SMM
select SOC_INTEL_COMMON_BLOCK_SMM
select SMM_TSEG
select HAVE_SMI_HANDLER
select ACPI_INTEL_HARDWARE_SLEEP_VALUES
Debug console will show all 36 cores relocated. Further tested by
generating SMI's to port 0xb2 using XDP/ITP HW debugger and ensured all
cores entering and exiting SMM properly. In addition, booted to Linux
5.4 kernel and observed no issues during mp init.
Change-Id: I00a23a5f2a46110536c344254868390dbb71854c
Signed-off-by: Rocky Phagura <rphagura@fb.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43684
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Angel Pons <th3fanbus@gmail.com>
The SMM_LOCK bit isn't in SMM_MASK_MSR, but in HWCR_MSR, so move it
there. The soc/amd/* code itself uses the bit definition when accessing
HWCR_MSR, so SMM_LOCK was just below the wrong MSR definition.
Also remove SMM_LOCK from comment about masking bits in SMM_MASK_MSR,
since that bit isn't in that MSR.
TEST=Checked the code and the corresponding BKDG/PPR.
Change-Id: I2df446f5a9e11e1e7c8d10256f3c2803b18f9088
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/43309
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
