Add fsp_write_line function which may be called by FSP to output debug
serial data to the console.
TEST=Build and run on Galileo Gen2
Change-Id: Ib01aef448798e47ac613b38eb20bf25537b9221f
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/16128
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
Commit 0d9cd92e (chromeos: Clean up elog handling) removed the
individual elog_init() calls from mainboards that did them and automated
adding certain events through the boot state machine. Unfortunately,
the new code would sometimes not log any specific event at all, and
thereby also never call elog_init() (through elog_add_event()) which
adds the "System boot" event.
We can assume that any board that configures the eventlog at all
actually wants to use it, so let's just add another call to elog_init()
to the boot state machine so we can ensure it gets called at least once.
BRANCH=None
BUG=chrome-os-partner:56001
TEST=Booted Kevin, confirmed that eventlog code runs again.
Change-Id: Ibe7bfc94b3e3d11ba881399a39f9915991c89d8c
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://review.coreboot.org/16118
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
Instead of relying on global state to determine if an error
occurred provide the ability to know if an add or shrink
operation is successful. Now the call chains report the
error back up the stack and out to the callers.
BUG=chrome-os-partner:55932
Change-Id: Id4ed4d93e331f1bf16e038df69ef067446d00102
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16104
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Don't conditionally compile parts of the code. The unused pieces
get culled by the linker, and the #if's just clutter things up.
BUG=chrome-os-partner:55932
Change-Id: Ic18b2deb0cfef7167c05f0a641eae2f4cdc848ee
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16102
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Furquan Shaikh <furquan@google.com>
There were checks against global variables trying to determine
failing cases of elog_find_flash(). Instead move the checks
into elog_find_flash() and return value indicating failure.
A minimum 4KiB check was added to ensure the eventlog is at
least that size which makes the heuristic checks cleaner.
BUG=chrome-os-partner:55932
Change-Id: I4d9d13148555e05d4f217a10f995831a0e437fc3
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16101
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
There were 3 variables indicating the state of the event log
region. However, there's no need to keep track of those
individually. The only thing required is to know is if
elog_scan_flash() failed. There's no other tracking required
beyond that.
BUG=chrome-os-partner:55932
Change-Id: I88ad32091d3c37966a2ac6272f8ad95bcc8c4270
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16100
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
There were multiple paths where writes and erases of the flash
were being done. Instead provide a single place for synchronizing
the non-volatile storage from the mirrored event log. This
synchronization point resides as the very last thing done when
adding an event to the log. The shrinking check happens before
committing the event to non-volatile storage so there's no need
to attempt a shrink in elog_init() because any previous events
committed already honored the full threshold.
BUG=chrome-os-partner:55932
Change-Id: Iaec9480eb3116fdc2f823c25d028a4cfb65a6eaf
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16099
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Instead of treating offsets relative to after the header make
the offsets relative to the in-memory mirror buffer. This
simplifies the logic in that all offsets are treated the same.
It also allows one to remove a global variable.
BUG=chrome-os-partner:55932
Change-Id: I42491e05755d414562b02b6f9ae47f5c357d2f8a
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16098
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
A region_device can be used to represent the in-memory mirror
of the event log. The region_device infrastructure has builtin
bounds checking so there's no need to duplicate that. In addition,
it allows for removing much of the math juggling for the buffer
size, etc.
BUG=chrome-os-partner:55932
Change-Id: Ic7fe9466019640b449257c5905ed919ac522bb58
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16097
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
There's only 2 users of checking if the event buffer is cleared
to the EOL value. Each were passing pointers of the in-memory
mirror while also doing calculations for the size to check. Since
the in-memory mirror is one big buffer the only thing required
to know is the offset to start checking from. The check is always
done through the end of the buffer.
BUG=chrome-os-partner:55932
Change-Id: Icd4a7edc74407d6578fc93e9eb533abd3aa17277
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16096
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
Instead of taking pointers and back-calculating the
proper offset perform writes in terms of the offsets
within the elog region in flash.
BUG=chrome-os-partner:55932
Change-Id: I5fd65423f5a6e03825c788bc36417f509b58f64d
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16095
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins)
The elog_flash_erase() was only called to erase the entire
elog region in flash. Therefore, drop the parameters and
perform the full erase.
BUG=chrome-os-partner:55932
Change-Id: I6590347ae60d407bc0df141e9196eb70532f8585
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16094
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
There was a check against the next event offset against
the shrink size in elog_shrink(). However, all calls
to elog_shrink() were conditionalized on the next
event offset exceeding the full threshold. The shrink
size is set to the minimum of the full threshold and
a percentage of the elog region size. Therefore, it's
impossible for the next event offset to be less than
the shrink size because full threshold is always greater
than or equal to the shrink size.
BUG=chrome-os-partner:55932
Change-Id: Ie6ff106f1c53c15aa36a82223a235a7ac97fd8c7
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16093
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
For the elog shrink case we log the number of bytes shrunk
from the event log. However, when clearing the log the
size recorded was the entire region size including the header
as well as the event region space. To be more consistent
mark the clearing event with the number of bytes actually
cleared out (excluding the header size).
BUG=chrome-os-partner:55932
Change-Id: I7c33da97bd29a90bfe975b1c6f148f181016f13f
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/16092
Tested-by: build bot (Jenkins)
Reviewed-by: Furquan Shaikh <furquan@google.com>
The -b FSP_LOC argument to cbfstool is only valid for the COREBOOT
CBFS. Don't pass that value for all other CBFS regions.
Change-Id: Ib5321e7a7dbee8d26eb558933c8ce3fea50b11fe
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/14641
Tested-by: build bot (Jenkins)
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
If EC_GOOGLE_CHROMEEC is enabled, ensure that the EC is in correct mode
before running memory init. This saves additional memory training
required in recovery path because of reboot later in ramstage.
BUG=chrome-os-partner:54245
Change-Id: Ic71c054afdcd0001cea95563fe513783b56f3e60
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/16034
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Although we have already support for the flash chip N25Q128 there is a
similar type available which has the same geometry and opcodes but
unfortunately a slightly different device type ID. While the already
supported N25Q128 has the ID 0xbb18 this one has the ID 0xba18.
To make both types available in the flash support table, use N25Q128A as
the flash name. This name can be found in the datasheet which can be
found here:
https://www.micron.com/~/media/documents/products/data-sheet/nor-flash/serial-nor/n25q/n25q_128mb_3v_65nm.pdf
TEST=Booted and verified that MRC cache could be written
Change-Id: I02a47692efb23a9a06a289c367488abd256b8e0c
Signed-off-by: Werner Zeh <werner.zeh@siemens.com>
Reviewed-on: https://review.coreboot.org/16061
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Add the Kconfig value to point to the checklist data files.
TEST=Build and run on Galileo Gen2
Change-Id: I3737b46162214fad139382193de944ec5d175645
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/16039
Tested-by: build bot (Jenkins)
Reviewed-by: Martin Roth <martinroth@google.com>
The TPM version string has become much longer recently, and the
TPM_FW_VER register available on VID 1ae0 devices supports reading in
arbitrary size quantities.
Let's read 50 bytes at a time to reduce the SPI register read wrapper
overhead, and increase the length limit to 300 bytes to accommodate
longer version strings.
TEST=verified on the Kevin device:
localhost ~ # grep cr50 /sys/firmware/log
Firmware version: RO_A: 0.0.1/84e2dde7 RO_B:* 0.0.2/13eda43f RW_A:* cr50_v1.1.5005-444ddb7 RW_B: cr50_v1.1.5005-5aac83c
cr50_v1.1.5005-444ddb7 private-cr51:v0.0.66-bd9a0fe tpm2:v0.0.259-8f3d735 cryptoc:v0.0.4-5319e83 2016-07-31 10:58:05 vbendeb@kvasha
Change-Id: Ifaf28c1a9a3990372a9cec108c098edbe50d3243
Signed-off-by: Vadim Bendebury <vbendeb@chromium.org>
Reviewed-on: https://review.coreboot.org/16000
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Choose appropriate debug levels for the various messages in the FSP
driver. Change:
* BIOS_DEBUG --> BIOS_SPEW: Normal FSP driver output level, allows
builder to disable FSP driver output by selecting
CONFIG_DEFAULT_CONSOLE_LOGLEVEL_7
* BIOS_ERROR --> BIOS_CRIT: These errors will prevent coreboot and the
payload from successfully booting
TEST=Build and run on Galileo Gen2
Change-Id: Ic3352de2022e16482bf47fc953aedeef8f0c2880
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/16003
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add a Kconfig value to enable display of FSP header. Move the display
code into a separate module to remove it entirely from the final image.
TEST=Build and run on Galileo Gen2
Change-Id: I7047a9e58e6a6481c8453dbfebfbfe69dc8823d8
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/16002
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Other SOC platforms need to handle the FspNotify calls in the same way
as Apollo Lake. Migrate the FspNotify calls into the FSP 2.0 driver.
Provide a platform callback to handle anything else that needs to be
done after the FspNotify call.
Display the MTRRs before the first call to fsp_notify.
TEST=Build and run on Galileo Gen2
Change-Id: I1ff327d77516d4ea212740c16c2514c2908758a2
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15855
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Move all FSP error handling into the FSP 2.0 driver. This removes the
need to implement error handling within the SOC code.
TEST=Build and run on Galileo Gen2
Change-Id: I4d548b4c90d369d3857c24f50f93e7db7e9d3028
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15853
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add support to display the HOBs returned by FSP:
* Add Kconfig value to enable HOB display
* Move hob_header, hob_resource and uuid_name structures into util.h
* Move hob_type enum into util.h
* Remove static from the debug utility functions
* Add fsp_ prefix to the debug utility functions
* Declare the debug utility functions in debug.h
* Add HOB type name table
* Add more GUID values
* Add new GUID name table for additional GUIDs
* Add routine to convert EDK-II GUID into a name
* Add SOC specific routine to handle unknown GUID types
* Add routine to convert HOB type into a name
* Add SOC specific routine to handle unknown HOB types
* Add routine to display the hobs
TEST=Build and run on Galileo Gen2
Change-Id: I10606d752859fff0f4f08a5ac03ab129b2c96d1f
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15851
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Add UPD display support:
* Add a Kconfig value to enable UPD value display
* Add a routine to display a UPD value
* Add a call before MemoryInit to display the UPD parameters
* Add a routine to display the architectural parameters for MemoryInit
* Add a weak routine to display the other UPD parameters for MemoryInit
* Add a call before SiliconInit to display the UPD parameters
* Add a weak routine to display the UPD parameters for SiliconInit
TEST=Build and run on Galileo Gen2.
Change-Id: I35fb8410c0bccf217b32af4b8bbe5ad6671f81f6
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15847
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Display the MTRR values in the following locations:
* Before the call to FspMemoryInit to document coreboot settings
* After the call to FspMemoryInit
* Before the call to FspSiliconInit
* After the call to FspSiliconInit
* After the call to FspNotify
* Before the call to FspNotify added in patch 15855
TEST=Build and run on Galileo Gen2
Change-Id: I8942ef4ca4677501a5c38abaff1c3489eebea53c
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15849
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Disable the chatty FSP behavior for normal builds. Use a Kconfig value
to enable the display of the FSP call entry points, the call parameters
and the returned status for MemoryInit, SiliconInit and FspNotify. The
debug code is placed into drivers/intel/fsp2_0/debug.c.
TEST=Build and run on Galileo Gen2
Change-Id: Iacae66f72bc5b4ba1469f53fcce4669726234441
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15989
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
(val & 4) == 1 is always false. Since val & 4 is either zero or
non-zero, just drop the second test (for "== 1").
Validated against the data sheet that this is really the right register,
bit and value.
Change-Id: I627df9a9b4fddfff486689e405f52a3b54135eef
Signed-off-by: Patrick Georgi <pgeorgi@chromium.org>
Found-by: Coverity Scan #1241864
Reviewed-on: https://review.coreboot.org/16009
Tested-by: build bot (Jenkins)
Reviewed-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Keep this enabled by default as most x86 platforms could have PCI-e
slots equipped with one of these Intel WiFi adapters.
The Kconfig entries under google boards had no function previously,
the variable was never referenced.
Change-Id: I728ce3fd83d51d4e5e32b848a2079c5fcee29349
Signed-off-by: Kyösti Mälkki <kyosti.malkki@gmail.com>
Reviewed-on: https://review.coreboot.org/15931
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
These files are required by storm and gale boards for enabling elog
support in ramstage.
BUG=chrome-os-partner:55639
Change-Id: I2bbfee2acf2bfe2f896a8619b1276dcea1b87f16
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/15893
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
With VBOOT_VERIFY_FIRMWARE separated from CHROMEOS, move recovery and
developer mode check functions to vboot. Thus, get rid of the
BOOTMODE_STRAPS option which controlled these functions under src/lib.
BUG=chrome-os-partner:55639
Change-Id: Ia2571026ce8976856add01095cc6be415d2be22e
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/15868
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
If the system is in recovery, store the newly generated MRC data using a
dummy version which is not legit. This ensures that on next normal boot,
new MRC data will be generated and stored.
BUG=chrome-os-partner:55699
Change-Id: Ib13e8c978dc1b4fc8817fab16d0e606f210f2586
Signed-off-by: Furquan Shaikh <furquan@google.com>
Reviewed-on: https://review.coreboot.org/15828
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Quark does not support the rdmsr and wrmsr instructions. In this case
use a SOC specific routine to support the setting of the MTRRs. Migrate
the code from FSP 1.1 to be x86 CPU common.
Since all rdmsr/wrmsr accesses are being converted, fix the build
failure for quark in lib/reg_script.c. Move the soc_msr_x routines and
their depencies from romstage/mtrr.c to reg_access.c.
TEST=Build and run on Galileo Gen2
Change-Id: Ibc68e696d8066fbe2322f446d8c983d3f86052ea
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15839
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
The removal of ELOG_FLASH_BASE and ELOG_FLASH_SIZE resulted
in the FMAP region for the eventlog to be honored. However,
certain systems seem to have a large eventlog region that
wasn't being used in practice. Because of the malloc() in the
eventlog init sequence a large allocation was now being requested
that can exhaust the heap. Put back the 4KiB capacity until
the resource usage is fixed.
BUG=chrome-os-partner:55593
Change-Id: Ib54b396b48e5be80f737fc3feb0d58348c0d2844
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/15835
Tested-by: build bot (Jenkins)
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Furquan Shaikh <furquan@google.com>
Separate NO_XIP_EARLY_STAGES from loading FSP-M into cache-as-RAM.
Quark executes romstage directly from the SPI flash part (in-place),
but loads FSP-M into ESRAM. This split occurs because ESRAM is too
small to hold everything while debugging.
Platforms executing FSP-M directly from the SPI flash need to select
FSP_M_XIP.
TEST=Build and run on Galileo Gen2.
Change-Id: Ib5313ae96dcec101510e82438b1889d315569696
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15848
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Build the UART drivers for the postcar stage.
TEST=Build and run on Galileo Gen2
Change-Id: I8bf51135ab7e62fa4bc3e8d45583f2feac56942f
Signed-off-by: Lee Leahy <leroy.p.leahy@intel.com>
Reviewed-on: https://review.coreboot.org/15843
Tested-by: build bot (Jenkins)
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Now that FMAP is a first class citizen in coreboot
there's no reason to have alternate locations for ELOG.
If one wants eventlog support they need to specify the
ELOG entry in the FMAP. The one side effect is that
the code was previously limiting the size to 4KiB
because the default ELOG_AREA_SIZE was 4KiB. However,
that's no longer the case as the FMAP region size is
honored.
Change-Id: I4ce5f15032387155d2f56f0de61f2d85271ba606
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/15814
Tested-by: build bot (Jenkins)
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>
FSP 2.0 spec only defines 2 reset request (COLD, WARM) exit codes. The
rest 6 codes are platform-specific and may vary. Modify helper function
so that only basic resets are handled and let SoC deal with the rest.
Change-Id: Ib2f446e0449301407b135933a2088bcffc3ac32a
Signed-off-by: Andrey Petrov <andrey.petrov@intel.com>
Reviewed-on: https://review.coreboot.org/15730
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Tested-by: build bot (Jenkins)
