Conversation
Natanael Copa
ncopa@fosstodon.org
#alpinelinux riscv64 package builds are progressing: ~690 packages left to build.
I think the real problem is the cases where the testsuite hangs. You always wonder if something deadlocked or if it is just slow. should you wait another day or just cancel it?
2
0
0
Jarkko Sakkinen
jarkko
@ncopa Most issues I've had with RISC-V have been way or another related to CPU caches. The ISA spec defines these properties with "lowest common denominator" approach, which is a design decision, as the spec's goal is not to define CPU architecture but more like seed for bunch of architectures. That said, I don't believe tho that this could be the root cause for such a wide spread :-)
E.g. when you read or write memory you don't necessarily always have any silver bullet to know how the hardware caches an address range, and often synchronization can require some extra dance with memory fences depending on which RISC-V CPU you run it.
E.g. when you read or write memory you don't necessarily always have any silver bullet to know how the hardware caches an address range, and often synchronization can require some extra dance with memory fences depending on which RISC-V CPU you run it.
0
0
2
@ncopa can't runners use timeout(1)?
1
0
0
Natanael Copa
ncopa@fosstodon.org@whynothugo sure they can, but what do you set it to for things that actually takes a day to build?
1
0
0
Jarkko Sakkinen
jarkko
1
0
1
@ncopa @whynothugo note: spent zero seconds on feasibility just wrote a remark :-)
1
0
0
@ncopa @whynothugo often when you get situations like that you are NOT prepared beforehand. Then you try to add kprobes or even inject log messages to scripts/code which can be super time-consuming effort. And sometimes the incident happens occasionally with unpredictable pattern...
Getting to the point where you can reason can be a long road...
Getting to the point where you can reason can be a long road...
0
0
0