Comment 9 for bug 1929926

Revision history for this message
Christian Ehrhardt  (paelzer) wrote :

I have prepared MPs
F: https://code.launchpad.net/~paelzer/ubuntu/+source/qemu/+git/qemu/+merge/410032
H: https://code.launchpad.net/~paelzer/ubuntu/+source/qemu/+git/qemu/+merge/410033
and a PPA: https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/4682

The hint that it is emulation helped and I now have working test steps and added an SRU Template for the bug. But it also made me wonder about the importance of having this at all.

I mean ... really ... "uretprobes"?
While nice those are not the most common use case for Ubuntu users IMHO.
And then s390x TCG emulation, that is rare as well and so far in the past was not even always really supported/wanted by IBM.
Combining the two I'm tempted to say we'd waste the time of the SRU team, waster internet BW to push a new qemu everywhere and waste testing time.

Then there also is plenty of fuzz applying it back to Focal (Hirsute is fine).
Applying patch ubuntu/lp-1929926-target-s390x-Fix-translation-exception-on-illegal-in.patch
patching file target/s390x/translate.c
Hunk #1 succeeded at 6317 with fuzz 2 (offset -95 lines).
Hunk #2 succeeded at 6334 (offset -95 lines).
Hunk #3 succeeded at 6362 (offset -95 lines).
Hunk #4 succeeded at 6371 with fuzz 1 (offset -95 lines).
Hunk #5 succeeded at 6385 with fuzz 2 (offset -103 lines).
Hunk #6 succeeded at 6445 (offset -104 lines).
Nothing evil - I reviewed and think it is fine, but still it adds to my point of "maybe not worth for what it fixes".

I wonder (no offense) if this has a real use case that IBM considers to be important for Ubuntu users or if this is just a "we have found it so let us fix it" case. If it is the latter then I'm tempted to say Won't Fix here. Or at least have it wait for another qemu SRU to come by that we can add it on top of.

Therefore my question to IBM, are there any use-cases or other strong reasons to push for this fix? If you do care about it I'm happy to help to carry this to a conclusion, but I'm asking for:

1. Please provide a use-case or reasoning why this actually deserves to be SRU-fixed and the few that want/need it can't just upgrade to e.g. 21.10? This isn't about me, we also need this to convince the SRU team

2. Please Test the linked PPA for Hirsute and Focal and report if both fulfill your needs

3. Please have a look at the Focal MP and confirm that you think the backport fuzz did not mess up anything

P.S. on potentially adding it on top of some other SRU:
- No other strong SRU in the queue right now (but there will be I'm sure).
- I'm also asking the security Team is something is inbound ...
- I'd still want 1-3 above be answered even if we piggy-back it on something else