version reports "armhf" on arm64
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
juju-core |
Fix Released
|
Critical
|
Ian Booth | ||
1.18 |
Fix Released
|
Critical
|
Ian Booth |
Bug Description
On arm64, juju --version reports being on armhf:
ubuntu@dannf1:~$ uname -m
aarch64
ubuntu@dannf1:~$ juju --version
1.18.0-trusty-armhf
The issue is that arch.go uses archREs to footprint the architecture based on "runtime.GOARCH". However, as per the comment var above archREs, it is a map of `uname -m` output to juju arch. In the case of arm64, runtime.GOARCH != `uname -m`. runtime.GOARCH is "arm64" and `uname -m` is "aarch64".
This may just be a cosmetic problem, but I do see version.Current used in various places (e.g. to look for upgrades), so just in case there be dragons here....
I did some checking and confirmed that the selection of runtime.GOARCH for this platform is intentional (I first guessed a bug in gccgo).
Related branches
- Juju Engineering: Pending requested
-
Diff: 24 lines (+2/-1)2 files modifiedjuju/arch/arch.go (+1/-1)
juju/arch/arch_test.go (+1/-0)
- Juju Engineering: Pending requested
-
Diff: 24 lines (+2/-1)2 files modifiedjuju/arch/arch.go (+1/-1)
juju/arch/arch_test.go (+1/-0)
description: | updated |
Changed in juju-core: | |
assignee: | nobody → Ian Booth (wallyworld) |
status: | Triaged → In Progress |
Changed in juju-core: | |
status: | In Progress → Fix Committed |
Changed in juju-core: | |
status: | Fix Committed → Fix Released |
It should report arm64, but it certainly shouldn't be giving armhf (which
would be the 32 bit version).
We need it to say arm64 internally because when it goes to look for tools,
it will look for something matching that string.
On Wed, Apr 9, 2014 at 4:34 AM, dann frazier <email address hidden>wrote:
> Public bug reported: /bugs.launchpad .net/bugs/ 1304742 /bugs.launchpad .net/juju- core/+bug/ 1304742/ +subscriptions
>
> On arm64, juju --version reports being on armhf:
>
> ubuntu@dannf1:~$ uname -m
> aarch64
> ubuntu@dannf1:~$ juju --version
> 1.18.0-trusty-armhf
>
> The issue is that arch.go uses archREs to footprint the architecture
> based on "runtime.GOARCH". However, as per the comment var above
> archREs, it is a map of `uname -m` output to juju arch. In the case of
> arm64, runtime.GOARCH != `uname -m`. runtime.GOARCH is "arm64" and
> `uname -m` is "aarch64".
>
> This may just be a cosmetic problem, but I do see version.Current used
> in various places (e.g. to look for upgrades), so just in case there be
> dragons here....
>
> I did some checking and confirmed that the selection of runtime.GOARCH
> for this platform is intentional (I first guessed a bug in gccgo).
>
> ** Affects: juju-core
> Importance: Undecided
> Status: New
>
>
> ** Tags: arm64 hs-arm64
>
> ** Description changed:
>
> On arm64, juju --version reports being on armhf:
>
> ubuntu@dannf1:~$ uname -m
> aarch64
> ubuntu@dannf1:~$ juju --version
> 1.18.0-trusty-armhf
>
> The issue is that arch.go uses archREs to footprint the architecture
> based on "runtime.GOARCH". However, as per the comment var above
> archREs, it is a map of `uname -m` output to juju arch. In the case of
> arm64, runtime.GOARCH != `uname -m`. runtime.GOARCH is "arm64" and
> `uname -m` is "aarch64".
>
> This may just be a cosmetic problem, but I do see version.Current used
> in various places (e.g. to look for upgrades), so just in case there be
> dragons here....
> +
> + I did some checking and confirmed that the selection of runtime.GOARCH
> + for this platform is intentional (I first guessed a bug in gccgo).
>
> --
> You received this bug notification because you are subscribed to juju-
> core.
> https:/
>
> Title:
> version reports "armhf" on arm64
>
> To manage notifications about this bug go to:
> https:/
>