Certain platforms need paging enabled during cache-as-ram because
dirty lines are being evicted by a heavy speculative frontend. Paging
needs to be enabled in order to utilize the NX (no execute) bit for
the regions that are strictly data (such as the stack). This utility
creates 32-bit PAE page tables using a static address space, and
the resulting tables have entries for all the PDPTEs such that it makes
it easy to enable 2MiB naturally aligned DRAM mappings once memory is
trained. Either binary files can be generated or C files. The pages that
are linked use a default base address of 0xaa000000 that can be changed at
runtime to reflect where the page tables are actually loaded. Or
specify a physical address on the command line that is known a priori.
iomap.txt:
0xd0000000, 0x100000000, UC, NX # All of MMIO
0xff000000, 0x100000000, WP, # memory-mapped SPI
0xffff8000, 0x100000000, WP, # XIP bootblock
0xfef00000, 0xfefc0000, WB, NX # CAR
0xfef40000, 0xfefc0000, WB, # verstage
0xfef20000, 0xfefc0000, WB, # romstage
0xfef40000, 0xfefc0000, WB, # fsp-m
$ go run util/x86/x86_page_tables.go --iomap_file=iomap.txt
Merged address space:
00000000d0000000 -- 00000000fef00000 UC NX : 375 big 256 small
00000000fef00000 -- 00000000fef20000 WB NX : 0 big 32 small
00000000fef20000 -- 00000000fefc0000 WB : 0 big 160 small
00000000fefc0000 -- 00000000ff000000 UC NX : 0 big 64 small
00000000ff000000 -- 0000000100000000 WP : 8 big 0 small
Total Pages of page tables: 5
Pages linked using base address of 0xaa000000.
BUG=b:72728953
Change-Id: I47625a24979b196011e2293712a8cdbdbb880d79
Signed-off-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-on: https://review.coreboot.org/24919
Reviewed-by: Furquan Shaikh <furquan@google.com>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Funtion fstat will return -1 if there's any error, 0 if successful.
Check that fstat return is equal to 0, print error message and exit if
not 0.
This fixes CIDs 1353018 and 1353020
BUG=b:72062481
TEST=Build no errors
Change-Id: I83284d9125c75a29471f213f88b9181d5edba2e6
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/23827
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
In case of console dump for only the last boot, cbmem utility checks
for a list of regex in provided order. When pre-cbmem console
overflows, "Pre-CBMEM <stage> console overflowed.. " message is added
before "... <stage> starting" message. This change fixes the order of
regex in cbmem utility to match this.
Test=Verified on soraka that "cbmem -1" correctly dumps the data
starting from Pre-CBMEM romstage overflowed.
Change-Id: I9c5667bbd02ed3e93ac77a4f42e87d754a062919
Signed-off-by: Furquan Shaikh <furquan@chromium.org>
Reviewed-on: https://review.coreboot.org/23800
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
board_status.sh was originally written for use cases where the DUT
is remote, i.e. accessed via serial port or SSH. This lead to some
issues when attempting to run the script on the DUT itself.
This patch attempts to handle the local use case more gracefully.
sudo is used when running the cbmem command, and the '-c' option
can be used to set cbmem path in case it's not in the default path
used by sudo.
Change-Id: I62957678ccae65fc46fd6ddf5ae92983d36cffad
Signed-off-by: David Hendricks <david.hendricks@gmail.com>
Reviewed-on: https://review.coreboot.org/21566
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
When running 'make distclean' on coreboot, the build cleans the tools
as well. Since secimage didn't have a distclean target, it gave an error
that the distclean target didn't exist. This didn't actually affect
anything more than the secimage clean, but it was impossible to tell
that from the warning:
% make distclean
make[1]: *** No rule to make target 'distclean'. Stop.
Change-Id: I4b4bcc1ab48e767218d31e455d23527acedf4953
Signed-off-by: Martin Roth <martinroth@google.com>
Reviewed-on: https://review.coreboot.org/23666
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
The registers were taken from the wrong addess since the spibar offset
was not added to it.
This also fixes the endianness.
Change-Id: I8bb91517770359599fe5f579c4686434da8d1c27
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/23478
Reviewed-by: Nico Huber <nico.h@gmx.de>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
ff4025c5f "sb/intel/bd82x6x: Reduce function-disable mess"
Removed most of the writes to RCBA(FD) and renamed the function to
mainboard_rcba_config.
Writes to FD are properly handled in ramstage, so no need to do it in
romstage.
Change-Id: I4edb75569ceec2d2f1308755a66d286202ca0ae6
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/23486
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Nico Huber <nico.h@gmx.de>
The `lint-extended-015-final-newlines` script skips over executable
files and thus leaves script files unchecked.
Use `file` to find scripts and include them in the `final newlines`
checks. Whitelisting is used including bash, perl, python and sh
scripts.
Change-Id: I8649b261b7e2cbbac7f9b90a9ace3f1c7b0eedeb
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23325
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Martin Roth <martinroth@google.com>
Add support for a mmio resource in the devicetree to allow
memory-mapped IO addresses to be assigned to given values.
AMD platforms perform a significant amount of configuration through
these MMIO addresses, including I2C bus configuration.
BUG=b:72121803
Change-Id: I5608721c22c1b229f527815b5f17fff3a080c3c8
Signed-off-by: Justin TerAvest <teravest@chromium.org>
Reviewed-on: https://review.coreboot.org/23319
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Reviewed-by: Furquan Shaikh <furquan@google.com>
The function read() returns the number of bytes actually read. Program is
assuming it actually read the required number of bytes without checking.
This is wrong.
This fixes CIDs 1353019 and 1353021
BUG=b:72062481
TEST=Build no errors
Change-Id: I22d41b3de4eac5369f512f78b1b31cc1a250f787
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/23304
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
File open function <open()> will return -1 if there's any error. Check that
the return is greater or equal to 0 before using fstat(). Print error message
and exit if there's an error.
This fixes CIDs 1353018, 1353020, 1353027 and 1353028
BUG=b:72062481
TEST=Build no errors
Change-Id: I77d6973d1ad1eadb93922866e618038045be5937
Signed-off-by: Richard Spiegel <richard.spiegel@silverbackltd.com>
Reviewed-on: https://review.coreboot.org/23303
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Aaron Durbin <adurbin@chromium.org>
Linters try to determine whether they are running in a git worktree so
that `git grep` can be used instead of `grep`. These checks are done in
different not truly correct ways and thus the linters don't use `git
grep` when running from a worktree subdirectory, e.g. in a git subtree
environment.
Unify checks using `git rev-parse --is-inside-work-tree`.
Change-Id: I3f54afc99ad0f0e3052cffdd32bdd9649cf3d720
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23297
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
`genbuild_h.sh` checks whether it is running from a coreboot's own git
worktree to decide whether to use git as the time source. This check
fails when `${top}/.git` is a gitfile, e.g. when coreboot is a
submodule.
Add a proper `git rev-parse` call to check the condition, remove `$top`
which is not used anymore.
Change-Id: I8bb13d607a01f4f28fa8b165769e0a1f702da362
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23256
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Some non-static declaration remains. If they were made
static, the compiler would output some warnings:
bincfg.y:30:1: warning: useless storage class specifier in empty declaration
};
^
bincfg.y:47:1: warning: useless storage class specifier in empty declaration
};
^
bincfg.y:22:12: warning: ‘yylex’ used but never defined
static int yylex (void);
^~~~~
bincfg.y:456:13: warning: ‘set_input_string’ used but never defined
static void set_input_string(char* in);
^~~~~~~~~~~~~~~~
Change-Id: I753e99c4a8290f9edd9abcda9af8e33b6ccfe406
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>
Reviewed-on: https://review.coreboot.org/23243
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
As the code was moved from the Makefile.inc to a separate file in
commit 9ab8ae6a (util/gitconfig: Make gitconfig a bash script),`$(MAKE)`
was replaced by `remake`, introducing dependency on this tool which is
basically a `make` with debugging capabilities. Many developers don't
have `remake` installed, leading to pre-commit hooks being not executed
properly. Apparently this was an unintentional change.
Furthermore, special treatment of `make` tool via the `%MAKE%`
substitution performed during hooks' deployment is still desired. Use
case is calling `remake gitconfig` to set `remake` as the `make` tool in
the hooks. To accomplish this, add a parameter that is passed from the
Makefile.inc to gitconfig.sh.
Change-Id: Ia78e06567b904b342dc9b7778569201fe02e6897
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23096
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Martin Roth <martinroth@google.com>
The `git sup-destroy` alias uses a subshell in order to make `git
submodule deinit` deinitialize all submodules. This isn't necessary as
the `--all` switch does the same.
Furthermore, `git submodule init && git submodule update` equals to `git
submodule --init`.
Change-Id: Ib690d66795da4049bb0bb350a0609cf2e6b5c4c4
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23249
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
This automatically generates an FMAP region for the MRC_CACHE driver
which is easier to handle than a cbfsfile.
Adds some spaces and more comments to Makefile.inc to improve
readability.
Tested on Thinkpad x200 with some proof of concept patches.
Change-Id: Iaaca36b1123b094ec1bbe5df4fb25660919173ca
Signed-off-by: Arthur Heymans <arthur@aheymans.xyz>
Reviewed-on: https://review.coreboot.org/23150
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The `gitconfig.sh` script installs hooks to the according directories
(for coreboot and its submodules). It has the `hooks` directory
hard-coded to be `.git/hooks`, which makes the installation fail when
coreboot itself is a submodule because then `.git` becomes a gitfile.
Replace hard-coded path handling using the according `git rev-parse`
calls.
Change-Id: I778e20be24bb27d0081c9e1c12883117d6d50347
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23253
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The name blobtool is confusing as 'blob' is also used to
describe nonfree software in binary form.
Since this utility deals with binary configurations it
makes more sense to call it bincfg.
Change-Id: I3339274f1c42df4bb4a6b30b9538d91c3c03d7d0
Signed-off-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>
Reviewed-on: https://review.coreboot.org/23239
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Kocialkowski <contact@paulk.fr>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The bash script `genrelnotes` checks for `.git` to be present to
determine whether the current directory is the top directory of a git
worktree. This check is rather weak and doesn't handle many edge cases
like that of a broken gitfile.
Add a proper `git rev-parse` call to check the condition.
Change-Id: I32b06ca982d55fd8e88e55651b6bc53014905823
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23252
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The Primary to Sideband Bridge (P2SB) is the interface to Private Con-
figuration Registers (PCR) including GPIO configuration. Of course,
access is restricted to Intel partners and criminals, so the PCI device
is hidden from the OS. Probably we only need to fetch the SBREG_BAR
address and can hide the PCI device again after that.
Change-Id: Ic121a09f021708aab82ae4b9d76d6c3c6fb884fa
Signed-off-by: Nico Huber <nico.huber@secunet.com>
Reviewed-on: https://review.coreboot.org/19588
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
GCC includes `sched.h` after poisoning calloc(). This results in a
build failure with Musl libc. We work around the issue by including
`sched.h` earlier and throw around some void pointers so we only
have to do it in one place.
Change-Id: I1d5462eb9a448147a95dd4ec50361b3f5a28910c
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/22786
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
Reviewed-by: Philipp Deppenwiese <zaolin.daisuki@gmail.com>
Looks like we were unnecessarily dragging this around for some time now.
GCC's installation manual doesn't mention libelf as a requirement and a
build of crossgcc-i386 doesn't show any sign of it being used.
This also fixes a lot issues on non-GNU distributions that were intro-
duced by switching to the elfutils version of libelf.
Change-Id: Iff308a9bed9ae3842557d251b75d1faadfafe0da
Signed-off-by: Nico Huber <nico.h@gmx.de>
Reviewed-on: https://review.coreboot.org/22773
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-by: Martin Roth <martinroth@google.com>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
In the buildgcc script, there is a check that the tools required are
installed. When a tool is missing, a message is output suggesting an
installation method, e.g. `sudo apt-get install foo` on debian-based
systems.
When run on a true, vanilla debian system, the error message provides
only a generic hint because the `please_install()` function fails to
detect the OS kind. Detection is based on definition of `ID_LIKE` in
`/etc/os-release` yet such systems only define `ID` to `debian`.
This commit closes the detection gap. Tested on debian 9 (stretch).
Change-Id: I3c867837e9157bee13010bd0a005028c369ce55f
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23231
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Patrick Georgi <pgeorgi@google.com>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
They both have a device id of 0x3c. The former is part of the PCI chip set
accessible via port 0x3f0 while the latter is a standalone LPC chip accessible
via 0x2e/0x4e depending on strapping.
They're not register compatible: the VT82C686 only provides a FDC, LPT and part
of UARTs.
The VT82C686 documentation suggests it has revision 0x00 while the VT1211
datasheet indicates 0x01. Nevertheless, the VT1211 I happen to have hs a
revision of 0x02. Thus the revision is probably not good enough to tell one
from the another.
Change-Id: Ic7529c84724c8d6b9eb75b863f1bceef5e4b52b5
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Reviewed-on: https://review.coreboot.org/22254
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
The `secimage` utility uses OpenSSL to calculate HMAC, which it does in
a rather unorthodox way, using deprecated `HMAC_CTX_init` API and
repeated calling of `HMAC_Init_ex` without a clear reason. The former
causes build errors with OpenSSL 1.1 while the rest of the
`HmacSha256Hash` function is confusing and overly complex.
Make `HmacSha256Hash` use a single OpenSSL API call. Test passed:
resulting signed binary remains identical.
Change-Id: Ib23c0ad96f9d8cc30ad357de8c0b0ba967c7d724
Signed-off-by: Alex Thiessen <alex.thiessen.de+coreboot@gmail.com>
Reviewed-on: https://review.coreboot.org/23069
Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Reviewed-by: Paul Menzel <paulepanter@users.sourceforge.net>
Reviewed-by: Stefan Reinauer <stefan.reinauer@coreboot.org>
