-
tsoome
am, do we have problems with code.illumos.org? gerrit seems to be gone and git pull is also waiting....
-
tsoome
ping appears to respond: code.illumos.org is alive
-
tsoome
gerrit is showing 502 Bad Gateway
-
danmcd
DUMB question. In mdb... can you somehow specify a demangled C++ symbol (e.g. `base::Allocate`) to get its address? Or is that way too complicated due to C++-isms (overloading, etc.) and you just have to cope with what you can discover from, say, a stack trace?
-
jbk
you can't, you have to use the mangled name
-
jbk
I've not thought about it if it's possible to go from the mangled name backwards or not (at least without additional context)
-
sommerfeld
danmcd: I've seen you comment about disabling E-cores on Alder Lake. what's the rationale for that? lack of uniform performance, or something actually breaking? (is it just the bootleg AVX-512 only on P-cores, or something else?)
-
MarcelT
Is gerrit down?
-
tsoome
since morning.
-
tsoome
(my morning:)
-
MarcelT
any ETA?
-
tsoome
no idea
-
sommerfeld
jclulow is the person who usually puts gerrit back together when it falls apart.
-
richlowe
technically bdha at least also can, but what tends to go wrong these days is "AI"
-
richlowe
so the old days of one of us just kicking it as we walked past are probably gone
-
richlowe
sommerfeld, danmcd: I'm interested in E v. P too
-
richlowe
(not that I have any ideas, but arm64 has the problem, or will eventually)
-
richlowe
though at least I'm pretty sure all arm cores do the same thing, at least, dammit intel.
-
sommerfeld
(where "AI" is fundamentally "web crawlers that don't respect robots.txt")
-
sommerfeld
richlowe: there are multiple dimensions to it, certainly. If the only difference between the cpus is performance it's a less serious issue than if they support subtly different ISAs
-
sommerfeld
I think that the AVX-512-on-P-cores-only thing was "fixed" by intel disabling AVX-512 using progressively bigger hammers.
-
sommerfeld
with interludes of bios vendors working around the initial disable
-
sommerfeld
sillycross.github.io/2022/06/11/2022-06-11 is an interesting read - P cores seem to be rather unpredictable.
-
richlowe
oh, that _is_ interesting.
-
jclulow
I have restarted gerrit. Let me know if you're still having trouble.
-
gitomat
[illumos-gate] 17355 libc: possible memory leak -- Toomas Soome <tsoome⊙mc>
-
tsoome
thanks!
-
MarcelT
jclulow: yes, gerrit works, thank you!
-
sommerfeld
jclulow: thank you!