Update to allocate udev_stack using (MAX(USB_HUB_MAX_DEPTH, USB_SS_HUB_DEPTH_MAX) + 1) instead of a hardcoded 7.

I'm sorry I forgot about this for so long. It seems like a good idea to me.

In D19940#428666, @jbeich wrote:

In D19940#428596, @mikael.urankar_gmail.com wrote:

In D19940#428579, @jbeich wrote:

• FreeBSD >= 12 should try elf_aux_info before KERN_PROC_AUXV, similar to how getauxval is tried before /proc/self/auxv on Linux

man elf_aux_info
No manual entry for elf_aux_info
-> not implemented

See multimedia/libvpx/files/patch-vpx__ports_{arm,ppc}__cpudetect.c for NEON and VSX examples. As the 2nd argument to elf_aux_info is a pointer one has to be a bit more careful with types i.e., libc sanity checks aren't foolproof to avoid a crash.

CC @mmel per rS324815 and @ian per planned manpage.

Make the final check (grep) work with the fact that 'set -e' is in effect, by directly testing the result of grep rather than using $? after the fact.

In D19871#426582, @andrew wrote:

Do we need something similar to intr_pic_init_secondary to also disable the timers on non-boot CPUs?

I haven't reviewed this change, but I did run-test it (on armv7 only). I updated to r345693 then applied the patch from this review, and ran the test code from https://github.com/emaste/test-tls-initial-exec. The test failed before patching and worked afterwards.

I can't think of any situation where /kernel is right, or where providing a wrong answer is better than providing the modern default (which could also be wrong, but no more wrong than /kernel has been for 20+ years now).

Looks good and works great.

Add the required change to sys/conf/files, missed in the first diff.