Binutils 2.38 added support for efi-app-aarch64
Still use binary mode if we have an older objcopy
Add check for incompatible gnu-efi crt0 containing the header section
which gets added by objcopy and if used results in duplicate header
and subsequently a broken binary
Signed-off-by: Callum Farmer <gmbr3@opensuse.org>
libefi.a just provided the c helper API that was slowly removed. As we
do not depend on anything provided by it anymore, it is safe to drop
now.
Since the ST/BS/RT pointers are very convenient and needed everywhere,
they are retained and initialized by us.
This stops using global GUID variables provided by gnu-efi. The globals
presumably exist to make EFI calls easier. But these GUIDs are just
128bits, so it's cheap to just put them on the stack when needed. C99
compound literals makes this even easier.
This also adds a convenience macro to create GUID pointers and uses it
everywhere. Not forcing a GUID definition to be a compound literal makes
them easier to use when assigning them to variables:
EFI_GUID before = *SOME_CONST_GUID;
EFI_GUID after = SOME_GUID;
And MAKE_GUID_PTR() makes it more explicit what is happening. It was
easy to confuse the old code as a cast:
before(&(EFI_GUID) SOME_GUID);
after(MAKE_GUID_PTR(SOME));
grub (at least on fedora) nowadays implements the boot loader spec, but
not the boot loader interface. Hence let's split out the commands
specific to each in two groups in the --help text. This way the first
group just covers boot lodaer spec stuff (i.e. stuff on disk in the
ESP/XBOOTLDR partition). And the other covers talking to the boot loader
via EFI var.
Removing debugging related sections is apparently not enough to truely
strip a (PE) binary.
Better late than never. It gets rid of this pesky warning from sbsign:
warning: data remaining[174080 vs 193173]: gaps between PE/COFF sections?
The unlink command removes an entry from the ESP including
referenced files that are not referenced in other entries. That is
useful eg to have multiple entries that use the same kernel with
different options.
The cleanup command removes all files that are not referenced by any
entry.
The custom print helpers have been replaced with explicit checks at the
call site to keep this in line with the way it is done in userspace. Any
calls where the check has been ommited should not need them as the value
is expected to alawys be around.
This fixes 3e87a057a7, which passed the
path to the wrong Print() call. Miraculously, this was printing the
correct path during testing and was therefore missed.
Let's introduce a common implementation of a function that checks
whether we are booted on a kernel with systemd-stub that has TPM PCR
measurements enabled. Do our own userspace measurements only if we
detect that.
PCRs are scarce and most likely there are projects which already make
use of them in other ways. Hence, instead of blindly stepping into their
territory let's conditionalize things so that people have to explicitly
buy into our PCR assignments before we start measuring things into them.
Specifically bind everything to an UKI that reported measurements.
This was previously already implemented in systemd-pcrphase, but with
this change we expand this to all tools that process PCR measurement
settings.
The env var to override the check is renamed to SYSTEMD_FORCE_MEASURE,
to make it more generic (since we'll use it at multiple places now).
This is not a compat break, since the original env var for that was not
included in any stable release yet.
When measuring data into a PCR we are supposed to hash the data on the
CPU and then pass the hash value over the wire to the TPM2. That's all
good as long as the data we intend to measure is not sensitive.
Let's be extra careful though if we want to measure sensitive data, for
example the root file system volume key. Instead of just hashing that
and passing it over the wire to the TPM2, let's do a HMAC signature
instead. It's also a hash operation, but should protect our secret
reasonably well and not leak direct information about it to wiretappers.
Now that we have ukify and mkosi has been updated to use it, we
have a solution in place to make sure that PE sections don't overlap
in a UKI so let's drop the workaround to avoid overlapping PE sections.
We converted to not using #ifdef for most of our defines because the syntax is
nicer and we are protected against typos and can set -Werror=undef. Let's do
the same for SD_BOOT. The define is nicely hidden in build.h for normal builds,
and for EFI builds we were already setting SD_BOOT on the commandline.
This erase_obj() machinery looks like voodoo and creates an awful lot of
noise as soon as we get back to building with -O0. We can do this in a
more simple way by introducing a struct that holds the information we
need on cleanup. When building with optimization enabled, all this gets
inlined and the eraser vanishes.
Yu Watanabe
Gerd Hoffmann
Luca Boccassi
Callum Farmer
Jan Janssen
Lennart Poettering
Ludwig Nussel
Daan De Meyer
Zbigniew Jędrzejewski-Szmek