It is already possible to disable swapping out kernel stacks via sysctl vm.swap_enabled (which has a wrong description) and configure it via sysctls vm.swap_idle_threshold1 and vm.swap_idle_threshold2. What is the problem you are trying to solve?

For a long time, "swapping out" a process has only made its U area (later, only the kernel stack) pageable. This is a few pages per thread. Therefore, comparing vmspace_resident_count does not seem to make much sense.

As of rS303524 (July 2016), _WITH_GETLINE is no longer checked by stdio.h. Therefore, there are two possibilities:

Looks useful.

OK for bin/sh tests and man pages.

It looks like daemon()'s chdir and close actions are performed just before returning successfully; therefore, capability mode code could also call daemon(1, 1) and perform the chdir and/or close actions after it, either open-coded or calling new function(s).

Sorry for the delay.

This will leave the calls to /usr/bin/fortune in share/skel/dot.login and share/skel/dot.profile referring to something that does not and will not exist (the port will install as /usr/local/bin/fortune).

Accepted with a minor change to the man page.

Hmm, looking again, the idea of continuing is good, but printing diagnostics without affecting the exit status is usually not POSIX-compliant. To fix this, the global variable exitstatus can be set to 1. Note that there are existing places in function.c that do this wrong (which need not be fixed for this review).

An alternative implementation would write to a temporary file, then rename onto the destination once stdin closes. This is how ksh93's >;pathname redirection works (this does not mean I think this definitely should be done like that, or that the shell is the best place to implement this feature).

Note that all my comments on the cpuset_parselist() implementation also apply to the original version in cpuset.c; therefore, they could be fixed in a separate commit.

In D10363#260440, @theraven wrote:

This probably means that we no longer need to expose _DefaultRuneLocale in the header.

Looks good except for the missing DESTDIR error check.

As a side effect, this change brings in recent updates to the man page (this is good).

In D12006#249341, @sevan wrote:

@jilles best to just revert the commit & leave it as-is?

This seems a useful hint.

Looks reasonable as far as I can see.

This is internally consistent, but does not match what POSIX says. POSIX says only <sys/shm.h> is required and the permissions should be coded using octal constants (yuck). However, if a programmer wants to use S_IRUSR and the like, that is fine. POSIX defines the exact values these constants should have. I don't think POSIX needs to be followed exactly in this.

Note the paragraph below starting with "The above declaration of struct sigaction is not literal.". The man page is inconsistent now.

The code change by itself looks good (including all the reasoning about which extensions POSIX permits) but I would prefer not to encourage more use of expr(1) :(

As a side effect, this change allows options and operands to be mixed in GNU style. Either this should be disabled by prepending "+" to the optstring or this should be mentioned in the commit message.

I looked at what NetBSD and GNU expr do, and they vary in what they accept.

This looks similar to what I did manually to make a -DNO_CLEAN build work.

The test plan should also include running the Kyua/ATF tests.

If at all possible, I think it is best to avoid placing the traced process in a new process group, since this breaks things like terminal access control (the traced process will be in a background process group as far as the tty is concerned). Given that the new process group is removed, forwarding signals may make sense (but will duplicate signals if they were originally sent to the process group).