Dario Faggioli
ae98fa881a
- qtests test are not realiable when run inside OBS builders, so let's disable that part of the testsuite for now. There is work ongoing to run it somewhere else (on dedicated hosts) to avoid loosing coverage. (bsc#1204566) - Improve dependency handling (e.g., what's recommended vs. what's required. - Add a subpackage (qemu-headless) that brings in all the packages that are needed for creating VMs with tools like virt-install or VirtManager, run either locally or from a remote host. (bsc#1202166) OBS-URL: https://build.opensuse.org/request/show/1031414 OBS-URL: https://build.opensuse.org/package/show/Virtualization/qemu?expand=0&rev=740 |
||
---|---|---|
_constraints | ||
_multibuild | ||
.gitattributes | ||
.gitignore | ||
50-seabios-256k.json | ||
60-seabios-128k.json | ||
80-kvm.rules | ||
80-qemu-ga.rules | ||
ath5k-Add-missing-AR5K_EEPROM_READ-in-at.patch | ||
block-io_uring-revert-Use-io_uring_regis.patch | ||
bridge.conf | ||
build-be-explicit-about-mx86-used-note-n.patch | ||
bundles.tar.xz | ||
config.sh | ||
Disable-some-tests-that-have-problems-in.patch | ||
DSDT.pcie | ||
enable-cross-compilation-on-ARM.patch | ||
help-compiler-out-by-initializing-array.patch | ||
hw-scsi-megasas-check-for-NULL-frame-in-.patch | ||
hw-smbios-handle-both-file-formats-regar.patch | ||
Ignore-spurious-GCC-12-warning.patch | ||
increase-x86_64-physical-bits-to-42.patch | ||
ipxe-Makefile-fix-issues-of-build-reprod.patch | ||
ksm.service | ||
kvm.conf | ||
linux-user-add-more-compat-ioctl-definit.patch | ||
linux-user-drop-conditionals-for-obsolet.patch | ||
linux-user-Fake-proc-cpuinfo.patch | ||
linux-user-lseek-explicitly-cast-non-set.patch | ||
linux-user-remove-conditionals-for-many-.patch | ||
linux-user-use-max-as-default-CPU-model-.patch | ||
linux-user-use-target_ulong.patch | ||
Make-char-muxer-more-robust-wrt-small-FI.patch | ||
Make-installed-scripts-explicitly-python.patch | ||
Makefile-define-endianess-for-cross-buil.patch | ||
Makefile-fix-build-with-binutils-2.38.patch | ||
meson-enforce-a-minimum-Linux-kernel-hea.patch | ||
meson-install-ivshmem-client-and-ivshmem.patch | ||
meson-remove-pkgversion-from-CONFIG_STAM.patch | ||
net-tulip-Restrict-DMA-engine-to-memorie.patch | ||
PPC-KVM-Disable-mmu-notifier-check.patch | ||
qboot-add-cross.ini-file-to-handle-aarch.patch | ||
qemu-7.1.0.tar.xz | ||
qemu-7.1.0.tar.xz.sig | ||
qemu-binfmt-conf-Modify-default-path.patch | ||
qemu-bridge-helper-reduce-security-profi.patch | ||
qemu-guest-agent.service | ||
qemu-ifup | ||
qemu-kvm.1.gz | ||
qemu-rpmlintrc | ||
qemu-supportconfig | ||
qemu.changes | ||
qemu.keyring | ||
qemu.spec | ||
qemu.spec.in | ||
Raise-soft-address-space-limit-to-hard-l.patch | ||
README.PACKAGING | ||
Revert-roms-efirom-tests-uefi-test-tools.patch | ||
Revert-tests-qtest-enable-more-vhost-use.patch | ||
roms-change-cross-compiler-naming-to-be-.patch | ||
roms-Makefile-add-cross-file-to-qboot-me.patch | ||
roms-Makefile-pass-a-packaging-timestamp.patch | ||
roms-sgabios-Fix-csum8-to-be-built-by-ho.patch | ||
scsi-generic-check-for-additional-SG_IO-.patch | ||
scsi-generic-replace-logical-block-count.patch | ||
seabios-switch-to-python3-as-needed.patch | ||
sgabios-Makefile-fix-issues-of-build-rep.patch | ||
Silence-GCC-12-spurious-warnings.patch | ||
stub-out-the-SAN-req-s-in-int13.patch | ||
supported.arm.txt | ||
supported.ppc.txt | ||
supported.s390.txt | ||
supported.x86.txt | ||
tests-change-error-message-in-test-162.patch | ||
tests-qemu-iotests-Triple-timeout-of-i-o.patch | ||
update_git.sh | ||
xen_disk-Add-suse-specific-flush-disable.patch | ||
xen-add-block-resize-support-for-xen-dis.patch | ||
xen-ignore-live-parameter-from-xen-save-.patch |
The qemu package includes a special maintenance workflow in order to support git based patching, including of submodules. Please use it in order to have changes you make be acceptable to the package maintainers. A bash script (update_git.sh) implements this workflow. A local clone of the upstream repo(s) is required for the packaging workflow. If none is found, the qemu superproject will be cloned as a shallow repo for the workflow script's use. If you intend to do more than superficial work in the qemu package, unshallow it. The submodule local repos aren't needed unless you intend to do patch work there. See config.sh for the local repo paths. The qemu.spec file is generated from a template file (qemu.spec.in), so to make spec file changes outside of the patch file references (handled by the script), you need to edit the template file, and NOT the qemu.spec file. The spec file's version and patch references are added when update_git.sh is passed certain commands, as described below. If you are not modifying any patches or their order, but just need to update the spec file from a changed template, run 'bash ./update_git.sh refresh'. If the set of patches is being modified, including their order, you will want to first run 'bash ./update_git.sh pkg2git', which makes the current package patch set available in a local git branch named 'frombundle' (see config.sh for the locations). This incarnation of the package patches comes from a "bundle of git bundles", included as a package source file named bundles.tar.xz, which the script extracts to the corresponding local frombundle branch. To add, remove or modify the package patches, you will then do that work in a local branch whose name corresponds to that qemu package's release version as follows: The qemu major and minor release numbers are part of the name, prefixed by 'opensuse-', so, for example, for the v5.2.0 based SLE-15-SP3 qemu, the branch would be named 'opensuse-5.2'. You want to start your work based on the current patchqueue as contained in 'frombundle', so one approach is to do (depending on your local repo's current state), 'git checkout -f --recurse-submodules -B opensuse-5.2 frombundle', and then cherry-pick upstream patches into it, or interactive rebase it to modify or delete patches, etc. Be careful to keep the branch based on the upstream tag which represents the package tarball however! Once you have the patch queue ready to go, simply run 'bash ./update_git.sh git2pkg' (in your local obs branch directory) to refresh the bundles.tar.xz file, as well as the package spec and patch files. The package qemu.changes file is modified to list added or removed patches as a starting point for documenting the change. When you are ready to check in the package, using 'bash update_git.sh ci' is provided as a convenience (and to help preserve correct spec file formatting, since a 'normal osc check-in' messes up the spec file a bit). Additional Notes: The maintainer and automation use another workflow mode dealing with packaging the latest upstream qemu. See 'LATEST' references in the scripts for details, as this is an 'expert mode' and isn't documented here. Patches which are from an upstream git repo should have the commit id recorded just below the Subject line (after a blank line) as follows: Git-commit: <40-char-sha-id> If a patch is anticipated to be shortly included in upstream repo, mark that fact by doing the above with 40 0's, which will flag it as needing to be updated in the near future. We try to maintain, if possible, patch ordering as follows: (tarball as base), patches which come from upstream commits in commit order, patches which will soon be upstream (as mentioned above), followed by our private patches. Bug or feature tracking identifiers should also be added to the patch similarly, using the abbreviations identified here: http://en.opensuse.org/openSUSE:Packaging_Patches_guidelines#Current_set_of_abbreviations using the "Reference:" tag, with multiple entries comma separated. The ability to provide a conditional inclusion of a patch (eg based on architecture, is provided by this workflow by using the "Include-If:" tag, as shown here: Include-If: %ifarch aarch64 This will cause the patch application in the spec file to be done as follows: %ifarch aarch64 %patch0013 -p1 %endif A trick worth noting is, if a given git tracked patch is to be applied in a way that can't be done in the normal patching section of the spec file, you can still include the patch, and use it by name with the patch program elsewhere in the spec file by doing something such as: Include-If: %if 0%{?patch-possibly-applied-elsewhere} (this variable will remain undefined in the spec file) And then elsewhere in the spec file, the actual patch (eg specially-handled-change.patch) is referenced as eg: patch -p1 < %_sourcedir/specially-handled-change.patch