bug: Fix lockup when RLIMIT_NOFILE is large

[markovejnovic]

Fixes #540

This PR implements a happy path for closing large arrays of file descriptors on Linux. This could also be implemented on FreeBSD, but I do not have a FreeBSD machine I can test this on.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[markovejnovic]

I have confirmed that this implementation solves my problem. I don't know how helpful this strace -f log is for you, but it seems that we use close_range successfully.

[pid 280940] dup2(14, 100)              = 100
[pid 280940] dup2(13, 101)              = 101
[pid 280940] dup2(8, 102)               = 102
[pid 280940] dup2(7, 103)               = 103
[pid 280940] prlimit64(0, RLIMIT_CORE, {rlim_cur=0, rlim_max=0}, NULL) = 0
[pid 280940] rt_sigprocmask(SIG_SETMASK, [], NULL, 8) = 0
[pid 280940] close(14)                  = 0
[pid 280940] close(13)                  = 0
[pid 280940] openat(AT_FDCWD, "/dev/null", O_RDWR) = 13
[pid 280940] dup2(13, 0)                = 0
[pid 280940] dup2(9, 1)                 = 1
[pid 280940] dup2(10, 2)                = 2
[pid 280940] close(13)                  = 0
[pid 280940] uname({sysname="Linux", nodename="fedora", ...}) = 0
[pid 280940] prlimit64(0, RLIMIT_NOFILE, NULL, {rlim_cur=1024, rlim_max=512*1024}) = 0
[pid 280940] close_range(4, 99, 0)      = 0
[pid 280940] close_range(104, 1023, 0)  = 0

Attaching full strace log (notice I aborted the program a couple seconds in, since the issue arises early).

strace.debug.log

Also tested that my docker container successfully runs fuzzilli under root with the absurdly large RLIMIT_NOFILE.