The effect of pointer aliasing on writes is that any data on CPU
registers that has been resolved from (non-const and non-volatile)
memory objects has to be discarded and resolved. In other words, the
compiler assumes that a pointer that does not have an absolute value
at build-time, and is of type 'void *' or 'char *', may write over
any memory object.
Using a unique datatype for MMIO writes makes the pointer to _not_
qualify for pointer aliasing with any other objects in memory. This
avoid constantly resolving the PCI MMCONF address, which is a derived
value from a 'struct device *'.
Change-Id: Id112aa5e729ffd8015bb806786bdee38783b7ea9
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31752
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
This patch cleans up remaining uses of raw boolean Kconfig values I
could find by wrapping them with CONFIG(). The remaining naked config
value warnings in the code should all be false positives now (although
the process was semi-manual and involved some eyeballing so I may have
missed a few).
Change-Id: Ifa0573a535addc3354a74e944c0920befb0666be
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31813
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Previously, the size of memory made for vboot_working_data
through the macro VBOOT2_WORK was always specified in each
individual memlayout file. However, there is effectively no
reason to provide this customizability -- the workbuf size
required for verifying firmware has never been more than 12K.
(This could potentially increase in the future if key sizes
or algorithms are changed, but this could be applied globally
rather than for each individual platform.)
This CL binds the VBOOT2_WORK macro to directly use the
VB2_WORKBUF_RECOMMENDED_DATA_SIZE constant as defined by vboot
API. Since the constant needs to be used in a linker script, we
may not include the full vboot API, and must instead directly
include the vb2_constants.h header.
BUG=b:124141368, b:124192753
TEST=Build locally for eve
TEST=util/lint/checkpatch.pl -g origin/master..HEAD
TEST=util/abuild/abuild -B -e -y -c 50 -p none -x
TEST=make clean && make test-abuild
BRANCH=none
CQ-DEPEND=CL:1504490
Change-Id: Id71a8ab2401efcc0194d48c8af9017fc90513cb8
Signed-off-by: Joel Kitching <kitching@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31474
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
This patch will fix these checkpatch errors in src/arch/mips/.
- src/arch/mips/ashldi3.c:22: WARNING: Prefer 'unsigned int' to bare use of 'unsigned'
- src/arch/mips/bootblock_simple.c:35: WARNING: braces {} are not necessary for any
arm of this statement
Change-Id: Ic859913b93dc8ed6ff64b551c8a6baf72d28c75a
Signed-off-by: Asami Doi <d0iasm.pub@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31938
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
In the current state of the tree we do not utilise the
mechanism of having per-device overrides for PCI bus
ops.
This change effectively inlines all PCI config accessors
for ramstage as well.
Change-Id: I11c37cadfcbef8fb5657dec6d620e6bccab311a4
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31753
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
In case PCI_IO_CFG_EXT=n parameter 'reg' was not
properly truncated to 8 bits and it would overflow
to dev.fn part of the register.
A similar thing could happen with 'dev' but that
value originates from PCI_DEV() macro unlike 'reg'.
Change-Id: Id2888e07fc0f2b182b4633a747c1786e5c560678
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31847
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
By design only 'reg' parameter can have the two least-
significant bits set. As 'reg' is often a constant,
'0xCFC + (reg & 3)' resolves to an immediate value
already at buildtime, unlike (addr & 3) which depends
of a constant (but non-immediate) value of 'dev' in
ramstage.
Change-Id: I6e729fe800c92b1ce4994ad2b4203072fa75a958
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31754
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Create the way of adding the discrete VGA OpROM at config UI (alternative to
./cbfstool ./cb.rom add -f vgabios_dgpu.bin -n pci1002,6663.rom -t optionrom )
DGPU options are accessible only if CONFIG_VGA_BIOS is enabled.
Signed-off-by: Mike Banon <mikebdp2@gmail.com>
Change-Id: I0a7bf0fe95c833cf3df0c7cb20fc27b6ab218c5a
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31449
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
This patch adds timestamp for "end of romstage" with postcar if platform
has selected postcar as dedicated stage.
If postcar stage doesn't exist then "end of romstage" timestamp will get
call while starting of ramstage as exist today.
TEST=It's been observed that "end of romstage" timestamp doesn't appear
in "cbmem -t" log when ramstage is not getting executed. As part of this fix
"end of romstage" timestamp is showing in "cbmem -t" log on Intel platform
where POSTCAR is a dedicated stage.
Change-Id: I17fd89296354b66a5538f85737c79145232593d3
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31608
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
This patch adds dedicated timestamp value for postcar stage.
TEST=Able to see "start of postcar" and "end of postcar" timestamp
while executing cbmem -t after booting to chrome console.
> cbmem -t
951:returning from FspMemoryInit 20,485,324 (20,103,067)
4:end of romstage 20,559,235 (73,910)
100:start of postcar 20,560,266 (1,031)
101:end of postcar 20,570,038 (9,772)
Change-Id: I084f66949667ad598f811d4233b4e639bc4c113e
Signed-off-by: Subrata Banik <subrata.banik@intel.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/31762
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Until now the TCPA log wasn't working correctly.
* Refactor TCPA log code.
* Add TCPA log dump fucntion.
* Make TCPA log available in bootblock.
* Fix TCPA log formatting.
* Add x86 and Cavium memory for early log.
Change-Id: Ic93133531b84318f48940d34bded48cbae739c44
Signed-off-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/29563
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Rudolph <siro@das-labor.org>
Since ACPI v2.c, this field is access_size.
Currently, coreboot is using ACPI v3,so we can drop '.resv' field.
Change-Id: I7b3b930861669bb05cdc8e81f6502476a0568fe0
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/31701
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Nico Huber <nico.h@gmx.de>
INT_MODEL defined in ACPI 1.0 and renamed to reserved since V 2.0.
The value for this field is zero but 1 is allowed to maintain
compatibility with ACPI 1.0.
So set this value to zero as we are using greater version than ACPI 1.0.
Change-Id: I910ead4e5618c958a7989f4c309a3a4bb938e31a
Signed-off-by: Elyes HAOUAS <ehaouas@noos.fr>
Reviewed-on: https://review.coreboot.org/c/29986
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: David Guckian
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
coreboot performs MP-Init in a parallel way. That leads to the fact
that the order, in which the CPUs are woken up, can vary from boot to
boot. The creation of the MADT table just parses the devicetree and
takes the CPUs reported there as it is for creating the single local
APIC entries. Therefore, the OS will see different order of CPUs.
There are CPUs out there (like Apollo Lake for example) which have
shared caches on core-level and if the order is random this can end up
in assigning cores to different tasks or even OSes (in a virtual
environment) which uses the same cache. This in turn will produce
performance penalties across these distributed tasks/OSes.
Though there is a way to discover the core- and cache-topology it will
in the end be necessary to take the APIC-ID into account. To simplify
it, one can achieve the same output by sorting the APIC-IDs in an
ascending order. This will lead to the fact that CPUs that share a given
cache will be reported right next to each other in the MADT.
Change-Id: Ida74f9f00a4e2a03107a2124014403de60462735
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/c/31545
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Fix up the logic of when to include VBOOT2_WORK symbols on x86,
which are only needed when VBOOT_STARTS_IN_BOOTBLOCK is enabled.
Also correct the value of the __PRE_RAM__ macro in the case that
VBOOT_STARTS_IN_ROMSTAGE is selected. In this case, DRAM is
already up and verstage should not be considered pre-ram.
BUG=b:124141368, b:124192753
TEST=Build locally for eve
TEST=util/lint/checkpatch.pl -g origin/master..HEAD
TEST=util/abuild/abuild -B -e -y -c 50 -p none -x
TEST=make clean && make test-abuild
BRANCH=none
Change-Id: Ie51e8f93b99ab230f3caeede2a33ec8b443e3d7a
Signed-off-by: Joel Kitching <kitching@google.com>
Reviewed-on: https://review.coreboot.org/c/31541
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
When <symbols.h> was first introduced, it only declared a handful of
regions and we didn't expect that too many architectures and platforms
would need to add their own later. However, our amount of platforms has
greatly expanded since, and with them the need for more special memory
regions. The amount of code duplication is starting to get unsightly,
and platforms keep defining their own <soc/symbols.h> files that need
this as well.
This patch adds another macro to cut down the definition boilerplate.
Unfortunately, macros cannot define other macros when they're called, so
referring to region sizes as _name_size doesn't work anymore. This patch
replaces the scheme with REGION_SIZE(name).
Not touching the regions in the x86-specific <arch/symbols.h> yet since
they don't follow the standard _region/_eregion naming scheme. They can
be converted later if desired.
Change-Id: I44727d77d1de75882c72a94f29bd7e2c27741dd8
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/c/31539
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
