Mark Johnston
Mark Johnston
The TAP tests are still broken because the header is incorrect. `closefrom_test`: ``` 1..21 ok 1 - open ok 2 - highest_fd ok 3 - closefrom ok 4 - open...
Merged in 8268a31bcceb9ebe32d380cab792c89c5d897d15, thank you.
I dropped the procstat changes since those are coming from upstream and aren't critical to the rest of the PR.
kyua doesn't create epair interfaces. Tests do. They're responsible for cleaning up; if they don't, it's a test bug.
It's impossible to say without knowing which test you're talking about. Please let us know how to reproduce the problem you're seeing.
Unfortunately, this doesn't appear to help much for sort. I haven't yet tried using utrace() probes to see how often we're allocating a new buffer now vs. before, though.
> > Unfortunately, this doesn't appear to help much for sort. I haven't yet tried using utrace() probes to see how often we're allocating a new buffer now vs. before,...
> I wonder if there are a class of reallocs that stock jemalloc, etc. are able to handle in-place that CHERI cannot because we are only able to realloc if...
> > > I wonder if there are a class of reallocs that stock jemalloc, etc. are able to handle in-place that CHERI cannot because we are only able to...