U-Boot was hanging on hardware, but not Qemu. This is because on
the machines tested, namely the X200 and E6230 laptops supported
in Libreboot, the UART was disabled from coreboot.
This U-Boot patch from Simon Glass works around the issue by
silently disabling the UART when it isn't there. Instead,
output is sent to the display and U-Boot no longer hangs.
Signed-off-by: Leah Rowe <leah@libreboot.org>
It's really buggy on hardware. Disable for now.
I've contacted Simon Glass on IRC, asking about hardware.
Signed-off-by: Leah Rowe <leah@libreboot.org>
It's a new experimental payload in Libreboot, so we may aswell
start with the very latest release of U-Boot.
Signed-off-by: Leah Rowe <leah@libreboot.org>
it's important that we maintain realistic expectations.
x86 u-boot is not yet fully stable, so mark it as such.
Signed-off-by: Leah Rowe <leah@libreboot.org>
When building a coreboot image, if they enable the
x86 U-Boot payloads, sometimes what happens is you
have CROSS_COMPILE set, for i386-elf, but then it's
still set to that when later building 64-bit U-Boot,
which needs x86_64-elf.
We currently rely on hostcc to build U-Boot.
To mitigate this, unset CROSS_COMPILE in the main
loop of the trees script, for building project targets.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Currently seems to stall when booted from the GRUB
payload, but works when booted from the SeaBIOS menu.
I also tested it as a standalone payload and it seems
to boot. Will test on hardware next, and start adding
it to more mainboards.
Signed-off-by: Leah Rowe <leah@libreboot.org>
also bring the coreboot/next modules in line with
the recent merge that did away with coreboot/dell7
the submodules for coreboot/haswell were still there,
and have now been deleted; the haswell tree was used
for the NRI patches, which were moved to /default some
time ago
Signed-off-by: Leah Rowe <leah@libreboot.org>
coreboot/dell7 is now part of coreboot/next, which in turn
has been updated, to accomodate 3050 micro patchset 18:
https://review.coreboot.org/c/coreboot/+/82053/18
It incorporates my Verb/VBT patches, which are therefore
no longer included separately.
Mate has fixed the USB config; see diff for details.
The configuration of USB ports was wrong, before.
Signed-off-by: Leah Rowe <leah@libreboot.org>
NOTE: Support added for xarch target x86_64-elf,
but U-Boot failed to build with this error:
OBJCOPY lib/efi_loader/helloworld.efi
x86_64-elf-objcopy: lib/efi_loader/helloworld_efi.so: invalid bfd target
make[2]: *** [scripts/Makefile.lib:476: lib/efi_loader/helloworld.efi] Error 1
Since I'm building U-Boot for x86_64 *on* an x86-64
host, and since that is currently the recommended type
of machine to use for lbmk development, and since the
other x86 payloads currently don't cross compile anyway,
this is an acceptable compromise for now. This is because
at present, I'm not making U-Boot the primary payload on x86,
instead preferring to chain it from GRUB and SeaBIOS.
The target.cfg file for x86 u-boot shows xarch/xtree commented.
Uncomment these to compile on crossgcc instead of hostcc.
I mention 64-bit because I initially did this first, but decided
to do 32-bit first. I'll work on the 64-bit one next (SPL).
It's only enabled in QEMU for now.
Signed-off-by: Leah Rowe <leah@libreboot.org>
There were a lot of unnecessary patches, such as the VRAM
patches; as Nicholas Chin has explained to me, the drivers
for these machines will just allocate what RAM they want
anyway, so in a lot of cases the extra allocated Video RAM
simply reduces the total amount of memory for other uses.
In general, we have a lot of patches that have existed for
years. A much more aggressive sweep will be done in the next
major audit, especially when the revisions are updated again.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Thanks go to Nicholas Chin and Lorenzo Aloe for working on
and testing this code. Based on the 780 MT port.
Signed-off-by: Leah Rowe <leah@libreboot.org>
pin mod needed (soldering) but according to mate, you
can use some coffeelake CPUs on these machines, despite
them being intel 7th gen. this includes 8-core chips.
this patch enables the software configuration in coreboot.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This is for blanking the ME region on release builds.
This is required for lbmk when doing Libreboot releases,
on images that use an Intel ME region.
Signed-off-by: Leah Rowe <leah@libreboot.org>
I reset it temporarily back to 1.16.3 when testing the
SeaBIOS hanging bug on 3050 micro, but the revision had
no effect; the bug was caused by a bad coreboot config
Signed-off-by: Leah Rowe <leah@libreboot.org>
Remove what is now unnecessary bloat, for ensuring that
GRUB is the primary payload; SeaGRUB is the only preference,
as per lbmk design.
The SeaBIOS hanging issue was fixed, so SeaGRUB is OK now.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Again, I'm adapting the config to be as close to the
coreboot one as possible. I compiled directly from coreboot
earlier, and got SeaBIOS to work on my 3050.
I'm matching the setup as closely as possible. Once it works,
I can use that in a Libreboot release but then debug why the
old config wasn't working.
Signed-off-by: Leah Rowe <leah@libreboot.org>
I'm eliminating as many differences as possible between lbmk's
setup, and the setup that is default when simply building from
the gerrit patch, directly in coreboot, by just picking the
mainboard; in this way, coreboot picks SeaBIOS as payload. I
already changed the SeaBIOS configs, in the previous patch.
Upon testing, this seems to have fixed the SeaBIOS hanging. I
need to have both of these options selected, or SeaBIOS hangs
just after it says "Press ESC" for the boot menu.
With this config change, SeaBIOS does not hang; instead, it shows
the list of devices as normal, and boots your machine.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This diff matches the setup currently used in coreboot.
I'm eliminating as many differences as possible, while
I test the SeaBIOS hanging issue on Dell Optiplex 3050 Micro.
The actual SeaBIOS configs have also been modified, to match
the coreboot config.
Signed-off-by: Leah Rowe <leah@libreboot.org>
- Update the MEC5035 S3 patches to the versions that were sent upstream
to prevent conflicts with subsequent patches for that EC.
- Update the patch that enables the S3 SMI handler in mainboard code so
that all Latitudes use the handler.
- Add a new patch that tells the EC to route power button events to the
host so that the OS can decide what to do. Without it, the EC powers
off the system without letting the OS cleanly shut down.
Signed-off-by: Nicholas Chin <nic.c3.14@gmail.com>
Specifically, use the same revision that Mate used in patchset 15.
This will ensure that any issues are *not* caused by the coreboot
revision; this is being done, because the old coreboot revision was
from July, but patchset 15 from Mate is based on a September revision
of coreboot.
I've been eliminating as many variables as possible, trying to fix
SeaBIOS payload on this machine, because it hangs in Libreboot, but
not when building from gerrit directly, which means the coreboot
revision may be a factor (since I'm using his patches on an older
revision so upstream might have made some changes since then that
the port relies on).
For this, a new coreboot tree is used, called "dell7", referring to
the fact that Kabylake is Intel's 7th generation.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Use patchset 15 instead of 14:
config/coreboot/default/patches/0061-WIP-OptiPlex-3050-Micro-port.patch
Rebase the verb patch; patchset 15 modified the Makefile:
config/coreboot/default/patches/0064-dell-optiplex_3050-add-hda_verb.c.patch
We were using patchset 14 for the 3050 micro:
https://review.coreboot.org/c/coreboot/+/82053/14
Now we use patchset 15:
https://review.coreboot.org/c/coreboot/+/82053/15
Without this patch, the fans are always on a low setting, on
the Dell OptiPlex 3050 Micro, even under stress conditions. With
this patch, the fans change speed according to CPU temperature.
I had to rebase my verb patch, because Mate modified the Makefile
to add his sch5555 handler, on the same line where I add hda_verb.
Mate tells me he will merge my verb and vbt patches into a further
patchset later on. For now, I've simply rebased these patches on
top of Mate's newer work; I've told him he can use them in his port.
I'm probably going to now issue a new revision ROM image for
Libreboot 20241008, so that users can get this fix sooner.
Signed-off-by: Leah Rowe <leah@libreboot.org>
the .git directory never exists anyway, when doing a release,
so the purpose this is intended is defeated by lbmk's design.
individual headers say "pcsx-redux team" as copyright anyway,
and the code for generating that COPYING file, with MIT license
and correct years (matching the entire source code for the
open bios) remains correct.
a mitigation instead of this patch might be to maintain a hardcoded
list of authors, and manually update it over time, but this is not
required. however, it may be good practise for upstream to maintain
such a file. perhaps i should contact them?
Signed-off-by: Leah Rowe <leah@libreboot.org>
Due to quirks in how caching works in lbmk, this may be
error-prone. I'll properly address it in the next audit.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Riku used this for debugging, when adding the MXM support
to the HP EliteBook 8560w port. It will be useful for other
work that I have planned, so I'm archiving this too!
Riku has a lot of useful code, that I meant to import ages ago.
Once I'm done importing these in lbmk, I'll add backup repos.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Based on hell's code, but parses inteltool logs.
This will be useful for ports that I have planned, so
I'd like this to be included with Libreboot releases.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Used to dump MXM config for a given mainboard. We used this
for the HP EliteBook 8560w.
I meant to import this via config/git/ ages ago.
Signed-off-by: Leah Rowe <leah@libreboot.org>
This brings in the following important fix:
commit d128a0ae87086b37c0e5d7a8d934bcdee173402f
Author: Nicholas Chin <nic.c3.14@gmail.com>
Date: Fri Sep 27 22:57:22 2024 -0600
flashchips: Remove unsupported erase blocks for Winbond W25X{16,32,64}
This family of chips does not support the 0x52 (32 KiB block erase) and
0x60 (chip erase) opcodes according to their datasheet.
The full list of changes this brings in is as follows:
* d128a0a flashchips: Remove unsupported erase blocks for Winbond W25X{16,32,64}
* c6a924a Don't mention writing when erasing only (-E)
* dac4239 ch347_spi: Add 'spimode' parameter
* 56d236b chipset_enable: Add some newer AMD code names
* 3b9f152 chipset_enable: Probe AMD SPIBAR first and bail on ff
* 522160f meson: Add ft4222_spi
Nicholas Chin's patch fixes a bug on GM45 ThinkPads, where WX25
ICs (Winbond) could be read, but writes would fail in certain
cases because flashchips.c provided incorrect block erase commands.
This is unrelated to the --workaround-mx patch, for Macronix ICs.
Signed-off-by: Leah Rowe <leah@libreboot.org>
The deguard utility is executed within a subshell, and
the subshell does not handle error status. This patch
fixes that, so that the main shell also exits non-zero.
Signed-off-by: Leah Rowe <leah@libreboot.org>
I was build-testing gru_bob on an arm64 host, and got a
build error when compiling U-Boot.
Python.h missing - installing python3-devel fixes it.
Signed-off-by: Leah Rowe <leah@libreboot.org>
I added support earlier, on rom.sh, but the main build script
specifically defines which projects are to be compiled. I've
modified it so that pcsx-redux (just the BIOS part) will also
be compiled.
Signed-off-by: Leah Rowe <leah@libreboot.org>
I also checked the copyright declarations in the
directory src/mips/openbios where the PCSX-Redux BIOS
is, gleaning all the copyright years: 2019-2024 at this
time.
The years will be updated as and when PCSX-Redux is
updated in lbmk. Their BIOS is under MIT so I made lbmk
generate an appropriate COPYING file alongside the binary,
containing:
Copyright (c) 2019-2024 PCSX-Redux authors
Along with the actual text of the MIT license. With all
of this, the PCSX-Redux BIOS can now be included in
Libreboot releases.
No actual tarball is created. The release script in lbmk
simply copies the bin/ directory to ../roms
I'm leaving the PCSX-Redux BIOS release uncompressed,
because, and this will sound patronising because that is
my precise intention: Windows users don't know how to do
anything. If I provide a tarball to Windows users, they
won't know what to do. Libreboot releases always go on rsync
mirrors, which also have HTTP servers with indexing enabled,
for browsing release files.
I mention Windows users, because most people who use the PCSX
Redux BIOS will probably use it on a PlayStation emulator, and
most emulator users are on Windows. I can't really be bothered
to provide it as a .zip archive, and it's only 512kb, so just
provide it uncompressed in Libreboot releases!
Releases were already possible under this scheme, so this
patch really just adds the COPYING file. It's simply a courtesy
to the PCSX-Redux developers, providing proper credit to them.
Signed-off-by: Leah Rowe <leah@libreboot.org>
on 3050micro, we disable seabios as a primary payload,
making grub a pribary payload instead.
the way it worked, the roms were still named seagrub
and the seabios rom would be compiled, but with the wrong
path, so seabios wouldn't be executed; seabios would hang
anyway, on this board.
instead, engineer it in such a way as to disable seabios_
images on this board. also, rename seagrub_ to grub_.
i normally only permit seagrub, and not grub, but i make an
exception for 3050micro because we know grub works, but seabios
currently hangs on this board (which means no bsd).
Signed-off-by: Leah Rowe <leah@libreboot.org>
SeaBIOS is known to hang on this board. It is being investigated.
Add two variable options for target.cfg files:
* seabiosname
* grubname
This string defines where it would be located in CBFS.
Signed-off-by: Leah Rowe <leah@libreboot.org>
in some cases, on a fresh clone, the cached repo already
exists but lbmk tries to download it again. work around
this by checking that the directory exists; it's in the
main if statement, so that the "else" still applies. as
a result, the fallback to a live repo would un-fall back
to doing git-pull if the cached directory exists exists.
if it doesn't seem to make sense, it's because it doesn't.
this whole function needs to be rewritten better.
Signed-off-by: Leah Rowe <leah@libreboot.org>
Leah Rowe