Piotr Balcer
Piotr Balcer
I'm having a conceptual problem on how to combine this feature with arenas. I started implementing this where each segment had its own set of arenas, but this created a...
Rebased to 1.12
Merged in #5489 and backported to 1.12.
You are right, thanks for reporting this. I looked over that whole example, and I think it should be either rewritten or removed entirely.
In general, the defrag method was implemented so that it didn't move data unnecessarily. This means that if allocations are nicely laid-out sequentially in some blocks with high addresses, they...
PMDK doesn't ship with Rust bindings. But there are some PMem-related crates. If you want something high-level, I recommend looking into https://crates.io/crates/corundum. We are also thinking about creating pmem2 rust...
ideally, you'd remove the dependency on internal variables entirely since the alternative approach would be to include a compile-time version check.
`0 errors from 0 contexts` ;)
worksforme. ``` std::unique_lock lock((*proot)->pmutex); auto ret = (*proot)->cond.wait_for(lock, wait_time); ``` This is the code that vg is complaining about in your log. This scenario literally can't happen unless std::unique_lock doesn't...
@krzycz this is context of https://dl.acm.org/doi/abs/10.1145/3445814.3446744?sid=SCITRUS @dibang2008 has agreed to upstream those changes back. But yes, I agree with your points. But instead of squashing the commits, I would recommend...