5:27 a.m.
Are the aarch64 test systems running on real or emulated hardware? (Is
there a way to tell?)
I ask because I'm seeing bad numerical results from a test and would
like to know if I can eliminate qemu as a possible cause -- not that I
think that's likely.
7:34 a.m.
On Thu, Sep 19, 2019 at 12:28 PM Dave Love <loveshack(a)fedoraproject.org> wrote:
Are the aarch64 test systems running on real or emulated hardware? (Is
there a way to tell?)
I ask because I'm seeing bad numerical results from a test and would
like to know if I can eliminate qemu as a possible cause -- not that I
think that's likely.
Do you mean a system to test SRPM on aarch64? or a system to test a
open source code?
Do you mean CI testing system or an aarch64 server for adhoc test?
10:51 a.m.
Jun Aruga <jaruga(a)redhat.com> writes:
On Thu, Sep 19, 2019 at 12:28 PM Dave Love
<loveshack(a)fedoraproject.org> wrote:
>
> Are the aarch64 test systems running on real or emulated hardware? (Is
> there a way to tell?)
>
> I ask because I'm seeing bad numerical results from a test and would
> like to know if I can eliminate qemu as a possible cause -- not that I
> think that's likely.
Do you mean a system to test SRPM on aarch64? or a system to test a
open source code?
Do you mean CI testing system or an aarch64 server for adhoc test?
I don't know why it matters, but I'm testing dynamic micro-architecture
selection added to a library I've packaged.
11:21 a.m.
> Do you mean a system to test SRPM on aarch64? or a system to
test a
> open source code?
> Do you mean CI testing system or an aarch64 server for adhoc test?
I don't know why it matters, but I'm testing dynamic micro-architecture
selection added to a library I've packaged.
Because there are some proper choices depending on the situation.
* Test by mock command (aarch64 qemu environment). (SRPM, adhoc)
https://github.com/rpm-software-management/mock/wiki/Feature-forcearch
* Test by Koji (aarch64 native) (SRPM, a kind of CI)
* Test by Copr (aarch64 native or qemu. I am not sure)
* Test by Packit service (aarch64 native, maybe) (SRPM, CI)
* Test qemu-user-static RPM and aarch64 container
* on local (Source, adhoc)
* with CI services (Source, CI)
* Test with Travis CI with qemu (aarch64 qemu environment) (Source, CI)
* Test with native aarch64 supporting CI (Source CI)
* only CI: Shippable CI, Drone CI,
* Works on ARM (free aarch64 server for open source) (Source, adhoc or CI)
For 2nd half, below documents might be helpful for you.
* https://github.com/junaruga/fedora-workshop-multiarch
* https://github.com/junaruga/ci-multi-arch-test
Jun
12:52 p.m.
On Thu, 19 Sep 2019 at 06:28, Dave Love <loveshack(a)fedoraproject.org> wrote:
Are the aarch64 test systems running on real or emulated hardware? (Is
there a way to tell?)
I ask because I'm seeing bad numerical results from a test and would
like to know if I can eliminate qemu as a possible cause -- not that I
think that's likely.
It depends on what level of emulation you mean. Are you meaning "is it
an x86_64 acting like a aarch64?" then the answer is no. If you are
meaning "is it an aarch64 guest system on an aarch64" then probably
yes... we don't have enough hardware to not use virtualization which
does have some emulation at some level.
--
Stephen J Smoogen.
5:10 a.m.
Stephen John Smoogen <smooge(a)gmail.com> writes:
It depends on what level of emulation you mean. Are you meaning
"is it
an x86_64 acting like a aarch64?" then the answer is no.
Yes, that's what I meant, thanks.
If you are
meaning "is it an aarch64 guest system on an aarch64" then probably
yes... we don't have enough hardware to not use virtualization which
does have some emulation at some level.
Sure, they're documented as VMs (and virt-what shows so). I may
misunderstand the virtualization, but I thought that a straight KVM VM
runs the instructions directly on the ThunderX hardware it advertises,
which would pin the failure either on target specifics in the library or
gcc. I'm not sure I have the enthusiasm to debug it anyway...
5:28 a.m.
On Fri, 20 Sep 2019 11:10:41 +0100
Dave Love <loveshack(a)fedoraproject.org> wrote:
Stephen John Smoogen <smooge(a)gmail.com> writes:
> It depends on what level of emulation you mean. Are you meaning "is
> it an x86_64 acting like a aarch64?" then the answer is no.
Yes, that's what I meant, thanks.
> If you are
> meaning "is it an aarch64 guest system on an aarch64" then probably
> yes... we don't have enough hardware to not use virtualization which
> does have some emulation at some level.
Sure, they're documented as VMs (and virt-what shows so). I may
misunderstand the virtualization, but I thought that a straight KVM VM
runs the instructions directly on the ThunderX hardware it advertises,
which would pin the failure either on target specifics in the library
or gcc. I'm not sure I have the enthusiasm to debug it anyway...
generally user space runs without any emulation in KVM and only
privileged instructions (used in kernel) are emulated or have some
special treatment
Dan
1:13 p.m.
On 9/19/19 3:27 AM, Dave Love wrote:
Are the aarch64 test systems running on real or emulated hardware?
(Is
there a way to tell?)
You mean the ones listed at
https://fedoraproject.org/wiki/Test_Machine_Resources_For_Package_Maintai...
?
If so, they are vm's running on real hardware with kvm.
kevin
--
I ask because I'm seeing bad numerical results from a test and would
like to know if I can eliminate qemu as a possible cause -- not that I
think that's likely.
_______________________________________________
devel mailing list -- devel(a)lists.fedoraproject.org
To unsubscribe send an email to devel-leave(a)lists.fedoraproject.org
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives:
https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org
5:11 a.m.
Kevin Fenzi <kevin(a)scrye.com> writes:
On 9/19/19 3:27 AM, Dave Love wrote:
> Are the aarch64 test systems running on real or emulated hardware? (Is
> there a way to tell?)
You mean the ones listed at
https://fedoraproject.org/wiki/Test_Machine_Resources_For_Package_Maintai...
?
Yes, apologies for not being explicit. Thanks.
9:31 a.m.
On Thu, Sep 19, 2019 at 11:27:21AM +0100, Dave Love wrote:
(Is there a way to tell?)
virt-what !
Rich.
--
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming and virtualization blog: http://rwmj.wordpress.com
virt-p2v converts physical machines to virtual machines. Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v
1674
days inactive
1678
days old
10 comments
6 participants
participants (6)
-
Dan Horák -
Dave Love -
Jun Aruga -
Kevin Fenzi -
Richard W.M. Jones -
Stephen John Smoogen