| Git Clone URL: | https://aur.archlinux.org/edk2-git.git (read-only, click to copy) |
|---|---|
| Package Base: | edk2-git |
| Description: | EDK2 UEFI Shell |
| Upstream URL: | https://github.com/tianocore/edk2 |
| Licenses: | BSD |
| Provides: | uefi-shell |
| Submitter: | markzz |
| Maintainer: | markzz |
| Last Packager: | markzz |
| Votes: | 31 |
| Popularity: | 0.000001 |
| First Submitted: | 2022-01-12 20:51 (UTC) |
| Last Updated: | 2022-01-12 20:51 (UTC) |
This no longer compiles. Same issue as the comment below by "AkechiShiro". I've tried editing the PKGBUILD file so we cd into "edk2" rather than "edk2-git" but no luck due to patches not being able to apply (I don't think the patch "edk2-git-202102-brotli-1.0.9.patch" is required anymore?).
I've had more success compiling by hand.
Hi,
At the moment, the build fails with this error :
==> Verifying source file signatures with gpg...
openssl-1.1.1l.tar.gz ... Passed
==> Removing existing $srcdir/ directory...
==> Extracting sources...
-> Creating working copy of edk2 git repo...
Cloning into 'edk2'...
done.
Updating files: 100% (8323/8323), done.
Switched to a new branch 'makepkg'
-> Extracting openssl-1.1.1l.tar.gz with bsdtar
-> Extracting brotli-1.0.9.tar.gz with bsdtar
==> Starting prepare()...
/home/xxxxx/.cache/paru/clone/edk2-git/PKGBUILD: line 75: cd: edk2-git: No such file or directory
==> ERROR: A failure occurred in prepare().
Aborting...
error: failed to build 'edk2-git-r29698.6062002bd5-1 (edk2-ovmf-git)':
error: packages failed to build: edk2-git-r29698.6062002bd5-1 (edk2-ovmf-git)
Maybe due to upstream changes ?
Great news, markzz! Looking forward to an updated build.
My rant below boils down to what the package provides. At time of writing, it only provided 2 files, OVMF_CODE.fd and OVMF_VARS.fd.
That isn't super-useful; what would be, is a single merged firmware file OVMF.fd ready to boot into QEMU with no fuzz.
Soon I'll be uploading an edk2-git set of packages that will replace this one. It should address some of the concerns below this comment.
realkstrawn93: This was due to edk2 being incompatible with Arch's version of brotli, which should be fixed in the new pkgbase.
Also, ulidtko, I never cared what every whatever thing on the internet says, you need to use your own brain. I know for a fact that this package worked back when it built properly. :)
Getting the following build failure:
brotli/c/dec/decode.c:2034:14: error: argument 4 of type ‘uint8_t ’ {aka ‘unsigned char ’} declared as a pointer [-Werror=vla-parameter]
My guess is that it's an upstream bug, but it's a showstopper nonetheless.
Can you please fix this so that it produces a single OVMF.fd file? Not (only) a pair of ovmf_{code,vars}_x64.bin like it does now.
Every damn readme, wiki, gist & blogpost says basically "just pass -bios /path/to/OVMF.fd & bob's your uncle, qemu runs uefi". Well. No such thing here.
I guess it's something extremely trivial, prolly just concatenation of two files (in the right order) -- but, all these .fdf & .dsc & obscure build workspaces and whatnot are extremely confusing. I don't need to learn all this to simply test if my usb stick will boot. A single ready-to-use file would help a lot.
For now, I've uninstalled this package and switched to extra/edk2-ovmf instead. It does provide both OVMF_CODE.fd and OVMF_VARS.fd -- but also a single OVMF.fd which Simply Works with Qemu.
Since this conflicts with extra/ovmf, it would be nice if it deposited the files in the same location so that I didn't have to modify my qemu config. A Simple symbolic link could do the trick, or installation to both locations?
package() {
install -D -m644 "${srcdir}"/edk2/Build/OvmfX64/RELEASE_${_toolchain_opt}/FV/OVMF_CODE.fd "${pkgdir}"/usr/share/ovmf/ovmf_code_x64.bin
install -D -m644 "${srcdir}"/edk2/Build/OvmfX64/RELEASE_${_toolchain_opt}/FV/OVMF_VARS.fd "${pkgdir}"/usr/share/ovmf/ovmf_vars_x64.bin
install -D -m644 "${srcdir}"/edk2/OvmfPkg/License.txt "${pkgdir}"/usr/share/licenses/ovmf/License.txt
# symlink
mkdir -p "${pkgdir}"/usr/share/ovmf/x64
ln -s /usr/share/ovmf/ovmf_code_x64.bin "${pkgdir}"/usr/share/ovmf/x64/OVMF_CODE.fd
ln -s /usr/share/ovmf/ovmf_vars_x64.bin "${pkgdir}"/usr/share/ovmf/x64/OVMF_VARS.fd
# or install to 2 spots
install -D -m644 "${srcdir}"/edk2/Build/OvmfX64/RELEASE_${_toolchain_opt}/FV/OVMF_CODE.fd "${pkgdir}"/usr/share/ovmf/x64/OVMF_CODE.fd
install -D -m644 "${srcdir}"/edk2/Build/OvmfX64/RELEASE_${_toolchain_opt}/FV/OVMF_VARS.fd "${pkgdir}"/usr/share/ovmf/x64/OVMF_VARS.fd
}
Hmm... The upstream URL points to something that says: "Whoops, we can't find that page"... I'm a bit confused about which of these packages to use for my virtual (Windows 10) machine:
aur/edk2-ovmf 20180815-1 (+9 0.01%) (Out-of-date 2019-05-23)
EDK2 ovmf firmware from Fedora 29 rpm
aur/ovmf-git 1:r25361.514c55c185-1 (+31 0.00%) (Installed: 1:r25460.83463154af-1)
Tianocore UEFI firmware for qemu.
extra/ovmf 1:r25737.89910a39dc-1 (1.3 MiB 2.0 MiB)
Tianocore UEFI firmware for qemu.
I previously used aur/edk2-ovmf, but now it says out-of-date since about a month ago, so I'm wondering if I should just use ovmf-git and if that's safe (I have the impression that extra/ovmf is too old)...? Anyway, I would have no doubts continuing to use edk2-ovmf, if it wasn't flagged as out-of-date, for the past month...
markzz: thank you very much :)
JuniorJPDJ: Updated to fix this. In the future, if there's an issue with the build like this, please flag the PKGBUILD as out-of-date so that it will be apparent when I log into the AUR interface.
Pinned Comments