Clean up codes for QEMU < 7.0 #3491

AkihiroSuda · 2025-05-02T07:06:56Z

QEMU 7.0 is documented as the minimum requirement since 9eb1396 (Dec 2022)
Also clean up the hint for QEMU 8.2.0, as QEMU 8.2.0 users should have already updated QEMU to 8.2.1+
QEMU 4.0 is still exceptionally allowed on linux/amd64 for the compatibility reason, but not tested
QEMU 6.2 is added to the CI

jandubois · 2025-05-02T16:20:24Z

I agree with bumping up the minimum version of QEMU, but if we remove warnings for older macOS versions, then we should have a top-level check that explains that Lima is no longer supported on that OS version instead of letting users run into problems.

However, I would disagree with breaking Lima on macOS 12, as it is often the latest version older Intel Macs can be upgraded to (e.g. I personally don't have any local Intel Mac that can run macOS 13 or later, but these are quad-core 4 GHz i7 machines with 16-64 GB of RAM and work otherwise totally fine using macOS 12 Monterey).

Yes, Apple considers this hardware (which is barely 10 years old) obsolete, and Homebrew doesn't support it anymore either. So I don't think we should spend any effort actively supporting it. But we shouldn't break it either unless it gets in the way of implementing new functionality.

AkihiroSuda · 2025-05-03T13:16:03Z

Restored the codes for macOS 12

afbjorklund · 2025-05-04T07:35:45Z

I think originally the request was to be able to use Lima with the system version of QEMU,
especially since one use case was to run the newer kernel in an Lima instance if needed...

Ubuntu	QEMU
~~20.04~~	4.2
22.04	6.2
24.04	8.2

RHEL	QEMU
7	1.5
8	6.2
9	9.0

So it would be nice to keep the minimum version at 6.2, unless 7.0 is somehow required?
Otherwise one has to build a newer version locally, which makes Lima harder to install.

docs: find out the oldest QEMU version that is compatible with Lima #320

But since we have already bumped it for Mac, can bump the minimum version for Linux too.

Then the users can choose, if they are able to upgrade their OS or if they want to install QEMU.

AkihiroSuda · 2025-05-04T08:26:07Z

Maybe we should allow QEMU 6.2 only on linux/amd64 host?
We should also have a ubuntu-22.04 CI?

AkihiroSuda · 2025-05-07T05:22:17Z

Added QEMU 6.2 CI (ubuntu-22.04)

jandubois · 2025-05-07T23:42:59Z

pkg/qemu/qemu.go

@@ -38,6 +38,10 @@ import (
 	"github.com/sirupsen/logrus"
 )

+// RecommendedQemuVersion is the recommemded QEMU version.
+// See commit 8c48cb3eef178914f1b8fffee324aab0c445fec5 .


Is this reference still needed? We use host: max now instead of host: cortex-a76.

So maybe the comment should be updated that we use -cpu max by default, which requires QEMU 7.0.0?

jandubois · 2025-05-07T23:44:07Z

pkg/qemu/qemu_linux_amd64.go

+// MinimumQemuVersion is the minimum supported QEMU version.
+// This is set to a lower version on linux/amd64 hosts due to a compatibility reason.
+// See qemu_others.go for the "real" minimum supported QEMU version.
+const MinimumQemuVersion = "4.0.0"


Is this actually going to work with the defaults? Won't we have to change -cpu max to -cpu host?

idk. Which version introduced -cpu max?

Maybe we should just set MinimumQemuVersion = "6.2.0" universally?

I believe -cpu max was added in 7.0.0, but I don't know if for all architectures, or if some came later.

ChatGPT says -cpu: max was introduced gradually:

x86_64: version 4

arm64: version 6

ppc64: version 7.1 or 7.2

riscv64: version 7

s390x: not supported at all (is this correct?)

s390x: not supported at all (is this correct?)

Nope 😅

s390x/cpumodel: Add "-cpu max" support (v3.1.0)
qemu/qemu@c61177881cb

x86_64: version 4

ChatGPT is just hallucinating 🤖
Actually it seems added in v2.9 ("i386" seems to cover x86_64 too)
qemu/qemu@c62f263

pkg/qemu/qemu.go

AkihiroSuda · 2025-05-08T05:51:14Z

Updated to set Minimum = 7.0.0, RecommendedOldest = 8.2.1 on macOS

jandubois

Thanks, LGTM

Personally not a big fan of having 5 different files to set a combination of 2 variables. This could be single function with runtime logic instead, that would be easier to understand. But fine if you prefer to do it this way.

jandubois

Thanks, LGTM

pkg/qemu/qemu.go

- QEMU 7.0 is documented as the minimum requirement since 9eb1396 (Dec 2022) - Also clean up the hint for QEMU 8.2.0, as QEMU 8.2.0 users should have already updated QEMU to 8.2.1+ - QEMU 4.0 is still exceptionally allowed on linux/amd64 for the compatibility reason, but not tested - QEMU 6.2 is added to the CI Signed-off-by: Akihiro Suda <akihiro.suda.cz@hco.ntt.co.jp>

jandubois

Thanks, LGTM

This MR contains the following updates: | Package | Update | Change | |---|---|---| | [lima-vm/lima](https://github.com/lima-vm/lima) | minor | `v1.0.7` -> `v1.1.1` | MR created with the help of [el-capitano/tools/renovate-bot](https://gitlab.com/el-capitano/tools/renovate-bot). **Proposed changes to behavior should be submitted there as MRs.** --- ### Release Notes <details> <summary>lima-vm/lima (lima-vm/lima)</summary> ### [`v1.1.1`](https://github.com/lima-vm/lima/releases/tag/v1.1.1) [Compare Source](lima-vm/lima@v1.1.0...v1.1.1) #### Changes - Fixed the guestagent path issues with Homebrew ([#3566](lima-vm/lima#3566), thanks to [@jandubois](https://github.com/jandubois)) - Documentation for disk management https://lima-vm.io/docs/config/disk/ ([#3539](lima-vm/lima#3539), thanks to [@Sonichigo](https://github.com/Sonichigo)) #### Usage ```console [macOS]$ limactl create [macOS]$ limactl start ... INFO[0029] READY. Run `lima` to open the shell. [macOS]$ lima uname Linux ``` *** The binaries were built automatically on GitHub Actions. The build log is available for 90 days: https://github.com/lima-vm/lima/actions/runs/15178234655 The sha256sum of the SHA256SUMS file itself is `0d2d3fb073c8e26df163937dd605e9f2b9f227814a697411cc2b8879347bdd7f` . *** Release manager: [@AkihiroSuda](https://github.com/AkihiroSuda) ### [`v1.1.0`](https://github.com/lima-vm/lima/releases/tag/v1.1.0) [Compare Source](lima-vm/lima@v1.0.7...v1.1.0) > \[!IMPORTANT] > Package maintainers should refer to "Hint for package maintainers" below Highlights: - Simplification of YAMLs - New port forwarder implementation by default - DragonflyBSD hosts - S390X and PPC64LE guests - Smaller binary packages #### Changes - Build system: - Split `lima-additional-guestagents-*.tar.gz` from `lima-*.tar.gz` ([#3503](lima-vm/lima#3503)) - Set `CONFIG_GUESTAGENT_COMPRESS=y` by default ([#3529](lima-vm/lima#3529)) - YAML: - Support inheritance ([#3072](lima-vm/lima#3072), thanks to [@jandubois](https://github.com/jandubois)) - Deprecate provision mode `ansible` ([#3451](lima-vm/lima#3451), thanks to [@afbjorklund](https://github.com/afbjorklund)) - Add new provision mode `data` ([#3302](lima-vm/lima#3302), thanks to [@jandubois](https://github.com/jandubois)) - QEMU: - Support DragonflyBSD hosts ([#3356](lima-vm/lima#3356), thanks to [@tuxillo](https://github.com/tuxillo)) - Support S390X guests ([#3319](lima-vm/lima#3319)) - Support PPC64LE guests ([#3488](lima-vm/lima#3488)) - TCG: change the default CPU from `qemu64` (x86-64 v1) to `max` ([#3487](lima-vm/lima#3487)) - Bump up the minimum QEMU version to v8.2.1 on ARM Mac. On Linux/x86\_64, QEMU v6.2.0 is still supported. ([#3491](lima-vm/lima#3491)) - WSL2: - Lots of improvements (several MRs, thanks to [@arixmkii](https://github.com/arixmkii)) - Port forwarding: - Enable faster gRPC implementation by default ([#3046](lima-vm/lima#3046)) - `limactl` CLI: - Add `--yes` flag as an alias of `--tty=false` ([#3342](lima-vm/lima#3342), thanks to [@suryaaprakassh](https://github.com/suryaaprakassh)) - Support resizing disk in `limactl edit` ([#3437](lima-vm/lima#3437), [#3533](lima-vm/lima#3533), thanks to [@songponssw](https://github.com/songponssw)) - Add `limactl disk import` command ([#3439](lima-vm/lima#3439), thanks to [@songponssw](https://github.com/songponssw)) - Rootless Containers: - Allow UID >= 524288 ([#3435](lima-vm/lima#3435)) - nerdctl: - Update from v2.0.4 to [v2.1.2](https://github.com/containerd/nerdctl/releases/tag/v2.1.2). ([#3483](lima-vm/lima#3483), [#3534](lima-vm/lima#3534), [#3560](lima-vm/lima#3560)) See also the release notes of [v2.0.5](https://github.com/containerd/nerdctl/releases/tag/v2.0.5), [v2.1.0](https://github.com/containerd/nerdctl/releases/tag/v2.1.0), [v2.1.1](https://github.com/containerd/nerdctl/releases/tag/v2.1.1). - nerdctl v2.1.1 included in Lima v1.1.0-rc.0 had a vulnerability of containerd CVE-2025-47290 (GHSA-cm76-qm8v-3j95). Other versions of Lima are not affected. - Templates: - `ubuntu-25.04`: New template ([#3445](lima-vm/lima#3445)). The `default` template still refers to Ubuntu 24.10, as `ubuntu-25.04` needs the very recent release of macOS on Intel Mac by default (see the note below). - `fedora-42`: New template ([#3434](lima-vm/lima#3434)) - `linuxbrew`: New template ([#3454](lima-vm/lima#3454)) - `almalinux-kitten-10`: New template ([#3084](lima-vm/lima#3084), thanks to [@afbjorklund](https://github.com/afbjorklund)) - `oraclelinux-8`: Fix virtiofs support ([#3441](lima-vm/lima#3441)) > \[!NOTE] > On Intel Mac with `--vm-type=vz`, macOS 15.5 or later is needed to boot `ubuntu-25.04` and `fedora-42` Full changes: https://github.com/lima-vm/lima/milestone/26?closed=1 Thanks to [@KGB33](https://github.com/KGB33) [@Nino-K](https://github.com/Nino-K) [@afbjorklund](https://github.com/afbjorklund) [@alexandear](https://github.com/alexandear) [@apachex692](https://github.com/apachex692) [@arixmkii](https://github.com/arixmkii) [@cakemanny](https://github.com/cakemanny) [@jandubois](https://github.com/jandubois) [@jonas-peter](https://github.com/jonas-peter) [@kairveeehh](https://github.com/kairveeehh) [@liangyuanpeng](https://github.com/liangyuanpeng) [@nirs](https://github.com/nirs) [@shenki](https://github.com/shenki) [@songponssw](https://github.com/songponssw) [@unsuman](https://github.com/unsuman) [@ycdzj](https://github.com/ycdzj) #### Hint for package maintainers Starting with v1.1, the official binary packages of Lima are split to two files for the each of the host OS and the architectures to save the disk space: 1. `lima-<VERSION>-Darwin-arm64.tar.gz`: (Built with `make native`) - The core components (`bin/lima`, `bin/limactl`, `share/lima/templates`, ...) - The guest agent for the native architecture (`share/lima/lima-guestagent.Linux-aarch64.gz`) 2. `lima-additional-guestagents-<VERSION>-Darwin-arm64.tar.gz`: (Built with `make additional-guestagents`) - The guest agents for emulating non-native architectures (`share/lima/lima-guestagent.Linux-{armv7l,ppc64le,riscv64,s390x,x86_64}.gz)`) For compatibility reason, `make` still builds the guest agents for all the architectures by default. Package maintainers are suggested to split their `lima` package to `lima` (`make native`) and `lima-additional-guestagents` (`make additional-guestagents`). #### Usage ```console [macOS]$ limactl create [macOS]$ limactl start ... INFO[0029] READY. Run `lima` to open the shell. [macOS]$ lima uname Linux ``` *** The binaries were built automatically on GitHub Actions. The build log is available for 90 days: https://github.com/lima-vm/lima/actions/runs/15154830653 The sha256sum of the SHA256SUMS file itself is `4391505a7f833c8245497b29fec65743abb1561a275037f6d268026284883c8a` . *** Release manager: [@AkihiroSuda](https://github.com/AkihiroSuda) </details> --- ### Configuration 📅 **Schedule**: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever MR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this MR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this MR, check this box --- This MR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate).

AkihiroSuda added the kind/refactoring Refactoring label May 2, 2025

AkihiroSuda added this to the v1.1.0 milestone May 2, 2025

AkihiroSuda force-pushed the qemu-cleanup branch from c981875 to fa0c9f3 Compare May 2, 2025 07:09

AkihiroSuda added the component/qemu QEMU label May 2, 2025

AkihiroSuda force-pushed the qemu-cleanup branch from fa0c9f3 to ff39ebc Compare May 2, 2025 07:10

AkihiroSuda changed the title ~~Clean up codes for QEMU < 7.0 and macOS < 13~~ Clean up codes for QEMU < 7.0 and macOS < 13.0 May 2, 2025

AkihiroSuda force-pushed the qemu-cleanup branch from ff39ebc to 8242bae Compare May 3, 2025 13:14

AkihiroSuda changed the title ~~Clean up codes for QEMU < 7.0 and macOS < 13.0~~ Clean up codes for QEMU < 7.0 May 3, 2025

AkihiroSuda requested a review from jandubois May 3, 2025 13:15

AkihiroSuda force-pushed the qemu-cleanup branch from 8242bae to 4760df7 Compare May 3, 2025 13:17

AkihiroSuda force-pushed the qemu-cleanup branch from 4760df7 to 4c4a898 Compare May 7, 2025 05:21

jandubois reviewed May 7, 2025

View reviewed changes

AkihiroSuda force-pushed the qemu-cleanup branch from 4c4a898 to 6381bed Compare May 8, 2025 05:50

AkihiroSuda force-pushed the qemu-cleanup branch from 6381bed to 00c3ed3 Compare May 8, 2025 21:56

jandubois previously approved these changes May 8, 2025

View reviewed changes

AkihiroSuda dismissed jandubois’s stale review via 223b5e2 May 9, 2025 01:03

AkihiroSuda force-pushed the qemu-cleanup branch from 00c3ed3 to 223b5e2 Compare May 9, 2025 01:03

jandubois previously approved these changes May 9, 2025

View reviewed changes

pkg/qemu/qemu.go Outdated Show resolved Hide resolved

AkihiroSuda dismissed jandubois’s stale review via 9bf7ae1 May 9, 2025 01:14

AkihiroSuda force-pushed the qemu-cleanup branch from 223b5e2 to 9bf7ae1 Compare May 9, 2025 01:14

AkihiroSuda force-pushed the qemu-cleanup branch from 9bf7ae1 to 7046952 Compare May 9, 2025 01:15

jandubois approved these changes May 9, 2025

View reviewed changes

jandubois merged commit 7530756 into lima-vm:master May 9, 2025
34 checks passed

Clean up codes for QEMU < 7.0 #3491

Clean up codes for QEMU < 7.0 #3491

Conversation

AkihiroSuda commented May 2, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jandubois commented May 2, 2025

Uh oh!

AkihiroSuda commented May 3, 2025

Uh oh!

afbjorklund commented May 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

AkihiroSuda commented May 4, 2025

Uh oh!

AkihiroSuda commented May 7, 2025

Uh oh!

jandubois May 7, 2025

Choose a reason for hiding this comment

Uh oh!

jandubois May 7, 2025

Choose a reason for hiding this comment

Uh oh!

AkihiroSuda May 7, 2025

Choose a reason for hiding this comment

Uh oh!

AkihiroSuda May 8, 2025

Choose a reason for hiding this comment

Uh oh!

jandubois May 8, 2025

Choose a reason for hiding this comment

Uh oh!

jandubois May 8, 2025

Choose a reason for hiding this comment

Uh oh!

AkihiroSuda May 8, 2025

Choose a reason for hiding this comment

Uh oh!

AkihiroSuda May 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

AkihiroSuda commented May 8, 2025

Uh oh!

jandubois left a comment

Choose a reason for hiding this comment

Uh oh!

jandubois left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jandubois left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

AkihiroSuda commented May 2, 2025 •

edited

Loading

afbjorklund commented May 4, 2025 •

edited

Loading

AkihiroSuda May 8, 2025 •

edited

Loading