Gene Cooperman

> My earlier comment was about respecting errno for dmtcp-internal calls -- those that have nothing to do with wrappers such as calls originating from coordinator, dmtcp_restart, etc. First, I...

@karya0, I could think about this more, and then decide what I think. There is a nice discussion of this business of whether a successful syscall (or library call?) is...

@ZxYuan , the CRAC repo that you mentioned should still be available. I tested the URL that you mention, and it still works. Our current plans are to merge that...

And @karya0 had noted: @gc00: Would it help if we created a new dmtcp-specific mutex type with corresponding dmtcp_mutex_lock, dmtcp_mutex_unlock, etc.? This has been on my radar for a while...

I added `DMTCP_PARALLEL_AUTOTEST` environment variable to distinguish the two levels (top-level autotest and parallel autotest on indidvidual test). This also fixes the inlilne request by @karya0. And the latest commit...

@rohgarg and @karya0, We'll have to wait until @karya0 approves the changes to unblock this. However, in testing on our 16-core computer, I'm seeing that 'make check-parallel' fails often on:...

@rohgarg, I tested on our 16-core computer _without_ `make check-parallel`. The ordinary test is also failing on things like `posiz-mq1`. Had you said that you were seeing this test fail...

I diagnosed the bug. I place the following three tests first in `autotest.py`: `posix-mq1` and `posix-mq1` (again). After that, even the first posix-mq1 fails. So, there appears to be some...

So, I found one difference between the 2.5 branch and the 3.0/master branch. The 2.5 branch configure sets `TEST_POSIX_MQ="no"`. So, we're not doing this problematic test there. Separately, I renamed...

I just did `posix-mq1` manually, and even without DMTCP, I can see: > % test/posix-mq1 mq_open() failed: No such file or directory mq_open() failed: Too many open files So, the...