The value stored in `gen` is only ever `1` or `0`. Storing `1` causes
Clang to warn, since the only valid values for a 1-bit int are -1 and 0:
```
amdfwtool.c:1487:27: error: implicit truncation from 'int' to a one-bit
wide bit-field changes value from 1 to -1
[-Werror,-Wsingle-bit-bitfield-constant-conversion]
1487 | amd_romsig->efs_gen.gen = EFS_BEFORE_SECOND_GEN;
```
TEST=Rebuilt coreboot; no warning was emitted.
Change-Id: Ibd83be8302e8a717db7e7dc86a403b5648976586
Signed-off-by: George Burgess IV <gbiv@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/83412
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Reviewed-by: Matt DeVillier <matt.devillier@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The purpose of integration function is to pack the FWs into table. We
need to remove other process. Create a dedicate function to link all
the tables together. And this linking function is only called when
both the level 1 and level 2 directory are created. This simplifies
the main function and logic.
TEST=Identical test on all AMD SOC platform
Change-Id: Ieaf97208e943c79d7b76ea62eea9355138c220b9
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/81252
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
The entry in the table has two categaries, file and pointer. For the
pointer, it does not take table space. The ISH, PSP level 2, BIOS
table are all the pointer type. So integration function only packs FWs
located in folder amd_blobs. And only FWs increase the table size.
So the table size is only set once. Later calls only update the count
and fletcher. The table has a header at least, so the size can not be
0.
The fill_dir_header can take the parameter count as 0, such PSP level
1 only with ISH-A and ISH-B. It doesn't have any file type entries.
This actually reverts
https://review.coreboot.org/c/coreboot/+/78274
and adds other changes.
TEST=Identical test on all AMD SOC platform
Change-Id: I5dfbbb55912c8e37243c351427a8df89c12e5da8
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/81255
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Cleanup the messy code. The code left in main is all about filling
tables.
To help to do this,
1. Some local variables are put into global struct.
2. Add some functions. Set some functions to global.
TEST=Identical test on all AMD platforms
Change-Id: Ia25c3fd5de7ae48054359f0f6551d91d7a4f6828
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/78311
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
The space defined by size of the L1 table can not overlap with ISH
header. For other cases, the size defines the directory and its
content.
The PSP spec does not say it quite clearly. This change is partly
based on guess and can make extraction tool work so far.
Change-Id: Id4fbc6d57d7ea070a9478649a96af92be9441289
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/78274
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
When updating the Makefiles, to keep from having to update two files at
the same time, import Makefile.mk into the external Makefile. This
allows the bulk of the settings to be in a single location.
While I'm here, I adjusted the print statements to match the rest of
coreboot.
Change-Id: Id5b869f49b34b22e6a02fc086e7b42975141a87e
Signed-off-by: Martin Roth <gaumless@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80715
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
The .inc suffix is confusing to various tools as it's not specific to
Makefiles. This means that editors don't recognize the files, and don't
open them with highlighting and any other specific editor functionality.
This issue is also seen in the release notes generation script where
Makefiles get renamed before running cloc.
Signed-off-by: Martin Roth <gaumless@gmail.com>
Change-Id: I434940ebb46853980596f7ad55d27a62c90280fa
Reviewed-on: https://review.coreboot.org/c/coreboot/+/80123
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Felix Singer <service+coreboot-gerrit@felixsinger.de>
Reviewed-by: Felix Held <felix-coreboot@felixheld.de>
Check for pkg-config presence and fail out with actionable message.
BUG=b:302521446
TEST=Build successfully with working pkg-config and failed build with no
pkg-config
Change-Id: I5d604145c919e7f71680d1e095dc68cb21868319
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/78191
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
The mapped windows is up to 16M. Even if the flash size is 32MB, it is
not mapped at 0xFE000000.
So using "0xFFFFFFFF - rom_size + 1" to get the "rom_base_address" can
only explain well when rom_size is less or equal to 16MB. For larger
size, it is not physically correct (Even though it can get expected
result).
If the flash size is larger than 16M, we assume the given addresses
are already relative ones. So we don't need the physical base address
any more.
This commit is part of a series of patches to support 32/64M flash.
BUG=b:255374782
Change-Id: I9eea45f0be45a959c4150030e7e213923510ad68
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/72959
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Besides fw.cfg, each combo entry needs dedicated APCB files. If no new
APCB is provided, the main APCB is used for all entries.
The combo is fully supported after this.
Change-Id: I21c2bf7d98ded43848ae8a8bb61d1ded1a277f88
Signed-off-by: Zheng Bao <fishbaozi@gmail.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/58620
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Some stages in bootflow prefer to use 16 bytes UUID instead of
traditional 2 bytes FWID to identify the firmware components they
verify/validate. Hence add version 2 of hash table which identifies
firmware components using UUID. Other than UUID and a reserved field for
alignment reasons, the format of the hash table is very similar to hash
table v1.
BUG=b:277292697
TEST=Build and boot to OS in Myst with PSP Verstage enabled. Ensure that
the hash table v2 is built and installed into BIOS image for the
components that are configured in amdfw.cfg file. Ensure that the
validation by PSP is successful for all the relevant components during
the boot flow.
Change-Id: I2899154086cf8e90c3327178157b07ead034b16e
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/76586
Reviewed-by: Matt DeVillier <matt.devillier@amd.corp-partner.google.com>
Reviewed-by: Tim Van Patten <timvp@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Currently this tool generates a hash table to verify signed binaries,
with a 2 byte FWID as the only kind of identifier. Going forward some
binaries are going to adopt 16 byte UUID identifiers and more binaries
will follow in the future SoCs. Hence add support for handling multiple
firmware identifier types. While at this remove the unused fwid from the
PSP FW table.
BUG=b:277292697
TEST=Build BIOS image and boot to OS in Myst & Skyrim.
Change-Id: I5180dc0fe812b174b1d40fea9f00a85d6ef00f2f
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/76585
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin L Roth <gaumless@gmail.com>
Hash table containing hashes of all signed PSP binaries is compiled at
build time and installed into the concerned CBFS. During boot, PSP
verstage reads the hash table binary and passes it to PSP bootloader.
PSP bootloader in turn uses the hash table to verify the signed PSP
binaries. Currently the hashes for all the signed PSP binaries are
compiled into one hash table. On upcoming platforms with more number of
signed PSP binaries, PSP bootloader does not have resources to handle
one monolithic hash table. Instead PSP bootloader recommends splitting
them into smaller hash tables (currently limited to 3 hash tables).
Update amdfwtool tool to support splitting hash tables. This is done by
adding an optional hash table id to the entries in the amdfw.cfg file.
By default, one hash table binary is always compiled and it's name is of
the format ${signed_rom}.hash. If an entry has a hash table id defined,
then this utility will compile a separate hash table binary whose name
is of the format ${signed_rom}.${N}.hash where N is the hash table id.
BUG=b:277292697
TEST=Build Skyrim BIOS image and boot to OS. Ensure that the hash table
is identical with and without this change. Perform suspend/resume
cycles, warm/cold reset cycles for 50 iterations each.
TEST=Artificially inject hash table id against some entries in
amdfw.cfg and ensure that the concerned hash table binaries are getting
compiled.
Change-Id: I7ef338d67695a34c33b5c166924832939f381191
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/75188
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Passing this option tells amdfwtool to create a text file, containing
the versions of the blobs below:
- PSP bootloader (type 0x01),
- SMU firmware (type 0x08),
- AGESA bootloader 0 (type 0x30),
- PSP bootloader AB (type 0x73).
Created file can be embedded into CBFS which allows to read the version
of blobs at runtime. This way version of blobs used to build the
coreboot image can be verified at runtime and also from the binary file.
Format of manifest file is following:
$ cat build/amdfw_manifest
type: 0x01 ver:00.35.00.13
type: 0x08 ver:00.5a.23.a6
type: 0x30 ver:2a.14.b0.10
type: 0x73 ver:00.35.00.13
BUG=b:224780134
TEST=Tested on Skyrim device
Change-Id: Idaa3a02ace524f44cfa656e34308bd896016dff6
Signed-off-by: Grzegorz Bernacki <bernacki@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/74266
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martin.roth@amd.corp-partner.google.com>
There are 2 regex patterns defined to process the lines from *fw.cfg:
1) for lines with mandatory entries
2) for lines with mandatory + optional entries
Consolidate the regex pattern. Add enums for matching regex caller
groups so that the human readable group IDs can be used instead of magic
numbers.
BUG=None
TEST=Build Skyrim BIOS which only have mandatory entries. Build Guybrush
BIOS image which have both mandatory and optional entries. Confirm that
the amdfw.rom built before and after this change have matching SHA in
both Skyrim and Guybrush images. This ensures that the optional level
entries in Guybrush are handled as expected. Boot to OS in Skyrim.
Change-Id: I7289ddbbec4d5daefe64f59b687ba3a4af46d052
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/74950
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Tim Van Patten <timvp@google.com>
Reviewed-by: Raul Rangel <rrangel@chromium.org>
Since commit 2f6b7d557d ("amdfwtool: Move the filling of table headers
into functions"), the combo_psp_directory union element in the
embedded_firmware is unused and the new_psp_directory element is used in
all places, so replace the union of new_psp_directory and
combo_psp_directory with just the new_psp_directory struct element.
Signed-off-by: Felix Held <felix-coreboot@felixheld.de>
Change-Id: I35d339b3084ec8f93210095c233f5e68296d0013
Reviewed-on: https://review.coreboot.org/c/coreboot/+/73834
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
If Recovery/Backup APCB is not passed, then AMD_BIOS_APCB_BK entry is
not populated. But PSP expects that bios directory entry to be
populated. Also on mainboards where both APCB and recovery APCB are same
(eg. Skyrim), 2 copies of the same APCB are added to amdfw*.rom. Update
amdfwtool to support not passing recovery/backup APCB. If the recovery
APCB is not passed, then populate AMD_BIOS_APCB_BK entry and make it
point to the same offset as AMD_BIOS_APCB entry.
BUG=b:240696002
TEST=Build and boot to OS in Skyrim. Ensure that the device can enter
recovery mode. Perform multiple suspend/resume cycles.
Change-Id: I031ba817573cd35160f5e219b1b373ddce69aa6b
Signed-off-by: Karthikeyan Ramasubramanian <kramasub@google.com>
Reviewed-on: https://review.coreboot.org/c/coreboot/+/73661
Reviewed-by: Fred Reitberger <reitbergerfred@gmail.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
