Ubuntu on IBM z Systems

luajit - is (or could) s390x be supported?

Bug #1838743 reported by Christian Ehrhardt 
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ubuntu on IBM z Systems
Triaged
Medium
bugproxy
luajit (Ubuntu)
Triaged
Undecided
Unassigned

Bug Description

Hi,
currently this isn't built on s390x which makes a long list of things not existing in s390x (all universe but anyway).

I was wondering if luajit should actually be working on s390x, to then enable it in Debian/Ubuntu packaging which would on a rebuild bring all of the below packages to s390x.

But IMHO it needs a check by an s390x authoritative person to build and check if it is ok there.
Otherwise it might build not not work well and hence suddenly have created a bunch of bad-behaving packages which isn't what anyone wants.

$ reverse-depends src:luajit -r eoan
Reverse-Depends
===============
* aegisub [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* cantor-backend-lua [amd64 i386] (for libluajit-5.1-2)
* ettercap-common [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* fastnetmon [amd64 arm64 i386 ppc64el] (for libluajit-5.1-2)
* knot-resolver [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libedje-bin [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libedje1 [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libefl-all-dev [amd64 armhf i386 ppc64el] (for libluajit-5.1-dev)
* libelua-bin [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libelua1 [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libevas1 [amd64 armhf i386 ppc64el] (for libluajit-5.1-2)
* libnginx-mod-http-lua [amd64 armhf i386] (for libluajit-5.1-2)
* libobs0 [amd64] (for libluajit-5.1-2)
* love [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* lua-ljsyscall (for luajit)
* lua-torch-cwrap (for luajit)
* lua-torch-dok (for luajit)
* lua-torch-graph (for luajit)
* lua-torch-nn (for luajit)
* lua-torch-nngraph (for luajit)
* lua-torch-optim (for luajit)
* lua-torch-paths (for luajit)
* lua-torch-sundown [amd64 arm64 armhf i386 ppc64el] (for luajit)
* lua-torch-sys (for luajit)
* lua-torch-torch7 [amd64 armhf i386 ppc64el] (for luajit)
* lua-torch-trepl [amd64 armhf i386 ppc64el] (for luajit)
* lua-torch-xlua (for luajit)
* luakit [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* minetest [amd64 armhf i386] (for libluajit-5.1-2)
* minetest-server [amd64 armhf i386] (for libluajit-5.1-2)
* nageru [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* neovim [amd64 armhf i386] (for libluajit-5.1-2)
* obs-studio [amd64] (for libluajit-5.1-2)
* python3-lupa [amd64 arm64 armhf i386] (for libluajit-5.1-2)
* rspamd [amd64 armhf i386] (for libluajit-5.1-2)
* suricata [amd64 armhf i386] (for libluajit-5.1-2)
* suricata-hyperscan [amd64 i386] (for libluajit-5.1-2)
* sysbench [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* sysdig [amd64 armhf i386] (for libluajit-5.1-2)
* torch-core-free [amd64 armhf ppc64el] (for luajit)
* torch-trepl (for luajit)
* trafficserver [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)
* uwsgi-plugin-luajit [amd64 arm64 armhf i386 ppc64el] (for libluajit-5.1-2)

Frank Heimes (fheimes)
tags: added: s390x
tags: added: universe
Changed in ubuntu-z-systems:
assignee: nobody → bugproxy (bugproxy)
tags: added: reverse-proxy-bugzilla
summary: - is (or could) s390x be supported?
+ luajit - is (or could) s390x be supported?
bugproxy (bugproxy)
tags: added: architecture-s39064 bugnameltc-179787 severity-high targetmilestone-inin1804
Revision history for this message
Christian Ehrhardt  (paelzer) wrote :

I just got feedback from xnox that in the initial enablement of s390x it surely was not supported (2015). But things might have changed, so hearing IBMs statement for 2019 - especially in case they worked on it - would be awesome.

Frank Heimes (fheimes)
Changed in ubuntu-z-systems:
importance: Undecided → Medium
status: New → Triaged
Revision history for this message
bugproxy (bugproxy) wrote : Comment bridged from LTC Bugzilla

------- Comment From <email address hidden> 2019-09-25 02:42 EDT-------
(In reply to comment #4)
Do you know which of the listed packages are performance critical and might be relevant for mainframe workloads?

Stefan SF. had a brief look into LuaJIT and he couldn't find any important user:

- The torch project is not under active development anymore. The much more popular pytorch framework does not depend on luajit.
- Also luajit itself does not appear to be actively developed right now. However, there appear to be some forks which got more traction over the last months:
https://www.freelists.org/post/luajit/latest-news,3

There also is a request to support LuaJIT for the microservices management platform istio (https://github.com/istio/istio). But it isn't clear whether LuaJIT support is a hard requirement here.

Revision history for this message
bugproxy (bugproxy) wrote :

------- Comment From <email address hidden> 2019-09-25 02:58 EDT-------
From the link Christian posted internally: https://github.com/LuaJIT/LuaJIT/pull/395
it can be seen that the S/390 LuaJIT port from Mike Munday has been added to the fork from siddesh:

https://github.com/siddhesh/LuaJIT

From there it went into Fedora 31:

https://koji.fedoraproject.org/koji/buildinfo?buildID=1254116

Siddesh talked about integrating this into Debian as well. But it did not seem to have made its way into the Debian package so far. Perhaps it is just a matter of making sure that this will happen :)

Revision history for this message
Frank Heimes (fheimes) wrote :

Ok, so if Siddesh or IBM is able to complete the integration into Debian, it would be possible by Canonical to sync it to Ubuntu for the next upcoming release. And with that probably almost all of the packages mentioned in the bug description would be addressed as well.

Frank Heimes (fheimes)
Changed in ubuntu-z-systems:
status: Triaged → Incomplete
Revision history for this message
Dimitri John Ledkov (xnox) wrote :

The patch looks nice.

I do see that luajit is alive again and merged things. I also see that moonjit is active too.

Needs investigation if we can switch to a better / more maintained / more featureful luajit implementation.

Changed in ubuntu-z-systems:
status: Incomplete → Confirmed
status: Confirmed → Triaged
Changed in luajit (Ubuntu):
status: New → Triaged
Revision history for this message
bugproxy (bugproxy) wrote :

------- Comment From <email address hidden> 2020-11-12 09:39 EDT-------
Some update info:
Here is the PR: https://github.com/LuaJIT/LuaJIT/pull/631
No reaction from the maintainer though.

Revision history for this message
bugproxy (bugproxy) wrote :

------- Comment From <email address hidden> 2021-02-11 03:38 EDT-------
@Xnox, how do we want to proceed here. Your opinion. Many thx.

bugproxy (bugproxy)
tags: added: targetmilestone-inin2110
removed: targetmilestone-inin1804
Matthieu Clemenceau (mclemenceau)
tags: added: fr-1396
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.