e340baf6cc introduced a bug where fish would not exit from job_continue
when receiving a signal like SIGHUP. This means that it would not in turn
deliver SIGHUP to its children, who would therefore never exit. Those
children may attempt to write to stdout, in which case they would receive
EIO; this can cause other weird issues, like telnet using 100% CPU.
Fixes#1958
Prior to b0e09303a, simple jobs like `printf "%s\n" $line | read word _`
never hit the call to select() because they were reaped in the SIGCHLD
signal handler. With that commit, the signal handler no longer reaps
children, and a job like that would enter select() and hit the 10000μs
timeout before discovering that the job was already complete.
Fixes#1884.
Prior to this fix, a child process may be reaped in one of two ways:
1. By a call to waitpid() within job_continue
2. By a call to waitpid() within the SIGCHLD signal handler
Only the second call was with the WNOHANG option. Thus if the signal
handler fired first, and then the waitpid call fired, we could get a
deadlock because we'd end up waiting on a long-running process. I have
not been able to reproduce this on fish 1.x, though it seems like it
ought to reproduce there too.
This fix migrates the waitpid() call out of the signal handler; the
second class of calls moves to job_reap. This eliminates the possibility
of a race, because we check for job completion before calling waitpid,
and there is no longer the possibility of the job being marked as
complete asynchronously. It also results in a massive conceptual
simplification, since the signal handler is now very simple and easy to
reason about (no more walking jobs lists, etc).
This partially fixes a bug reported in #1273
With this change, jobs only track their block-level redirections. Process level redirections are correctly associated with the process, and at exec time we stitch them together (block, pipe, and process redirects).
This fixes the weird issues where redirects bleed across pipelines (like #877), and also allows us to play with the order in which redirections are applied, since the final list is constructed right before it's needed. This lets us put pipes after block level redirections but before process level redirections, so that a 2>&1-type redirection gets picked up after the pipe, i.e. it should fix https://github.com/fish-shell/fish-shell/issues/110
This is a significant change. The tests all pass. Cross your fingers.
On FreeBSD, compilation complains that "this file includes
<sys/termios.h> which is deprecated, use <termios.h> instead". On Linux
and FreeBSD, <sys/termios.h> literally just pulls in <termios.h>. On OS
X and Solaris, <termios.h> pulls in <sys/termios.h>.
<termio.h> doesn't exist on FreeBSD or Mac OS X, and on Linux is marked
as deprecated and just includes <termios.h>. It does exist on Solaris,
but no `struct termio` is ever actually used in the codebase.
