As far as fusefs goes, we don't know the true blocksize. The FUSE protocol gives us no way of knowing. So using f_iosize is probably the best that we can do, and that's what this patch will do.

• Reduce duplication as suggested by @karels

Instead of doing anything fancy in the Makefile, how about you just create two separate test files named unix_passfd_dgram_test.c and unix_passfd_stream_test.c with contents like this:

#define TEST_PROTO SOCK_STREAM
#include <unix_passfd_test.c>

Here's another backwards-compatibility nit. In kevent, AIO events are level-triggered by default. That is, if kevent reports that an event is ready, but the caller for some reason calls kevent again without calling aio_complete, then kevent will immediately return ready again. But with AIO_KEVENT_FLAG_REAP, it won't be able to, because the kernel will no longer be aware of that event. I don't know if we need to care about users who depend on that level-triggering, but if we do an easy fix would be to only enable the new behavior if EV_ONESHOT is also set. It probably doesn't matter if we're requiring the user to explicitly set AIO_KEVENT_FLAG_REAP anyway. But it could matter if we use the new behavior by default.

• Also, reap similar dead code in aio_kqueue_test

Oh, sorry about that. I'll commit the fix. I didn't even realize that it was possible to leak memory in sh.

In D35336#801073, @markj wrote:

One other thing that might be useful is a script to enable some or all feature flags on a pool. The WIP makefs -t zfs implementation leaves all feature flags off.

This is going to be very useful. Will you create a matching change that will run zpool online -e on firstboot, or just assume that the VM images will be built with autoexpand=true?

In D35331#800920, @dteske wrote:

Do you need me to commit this or do you have a venue/access?

• Restore local variable declaration

This patch no longer applies, due to ports change f3c7a7686768351238894670b203c28526d247c1 .

In D35309#800308, @imp wrote:

So long as we don't have two disks with that label, we'll be fine...
So this is better than what we had, but not quite to where we need to get to....
Though the functionality we need to get there isn't in bsdinstall...

Thanks for fixing this.

In D35271#799385, @karels wrote:

I haven't used atf. I can guess at much of the layout, but don't know how to run it. Is there documentation somewhere?

Good catch. Do you think you could add a testcase to sbin/ping/tests/ping_test.sh ?

• Add 4s tests and make some magic numbers less magical.

I agree with you that two million is the new limit. As for other places that need changing, this is what I found:

• linprocfs: no overflow
• linux_sysinfo: overflows at a load average of 32? I don't know how to test it though. I'll ask trasz; he knows a lot about the linuxulator.
• snake_saver: no overflow
• schedcpu: overflows at 1 million, I think. Or maybe sooner but I'm having trouble figuring out the maximum value of ts_estcpu. But that should be plenty.
• tty_info: overflows at 10,000. I guess I'll go ahead and fix this one.
• sysctl_vm_loadavg: no overflow

Closing this review, which was superseded by D21377 .

Ping @imp . Do you think you could finish reviewing the tests sometime?

LGTM, and it works out of the box with at least one file system in ports.

Does anything already set fsname? libfuse, perhaps?

This looks awesome! It bothers me too when I have 100 different "/dev/fuse" file systems mounted, and I can't simply umount /dev/fuse. Speaking of which, does umount http://192.168.1.10 work with your patch? Could you please add a test for that in tests/sys/fs/fusefs/destroy.cc and another general test for the new feature in tests/sys/fs/fusefs/mount.cc? I can help with that, if you show me example usage with nmount. Also, I notice the "suse.com" in your username. Should we record this work as "Sponsored by: SuSE" when we commit it? Finally, are any other operating systems compatible with this usage?