At the time opt-in was introduced adding yourself as a writer was esrializing
across the mount point. Nowadays it is fully per-cpu, the only impact being
a small single-threaded hit on top of what's there right now.
Vast majority of the overhead stems from the call to VOP_GETWRITEMOUNT which
has is done regardless.
Should someone want to microoptimize this single-threaded they can coalesce
looking the mount up with adding a write to it.
bintime()/binuptime().
The algorithm to read the consistent snapshot of current timehand is
repeated in each accessor, including the details proper rollup
detection and synchronization with the writer. In fact there are only
two different kind of readers: one for bintime()/binuptime() which has
to do the in-place calculation, and another kind which fetches some
member from struct timehand.
Extract the logic into type-checked macros, GETTHBINTIME() for bintime
calculation, and GETTHMEMBER() for safe read of a structure' member.
This way, the synchronization is only written in bintime_off() and
getthmember().
In bintime_off(), use overflow-safe calculation of th_scale *
delta(timecounter). In tc_windup, pre-calculate the min delta value
which overflows and require slow algorithm, into the new timehands
th_large_delta member.
This part with overflow fix was written by Bruce Evans.
Reported by: Mark Millard <marklmi@yahoo.com> (the overflow issue)
Tested by: pho
Discussed with: emaste
Sponsored by: The FreeBSD Foundation (kib)
MFC after: 3 weeks
Returned value has type based on the argument, meaning consumers no longer
have to cast in the commmon case.
This commit keeps the kernel compilable without patching the rest.
It appears that FREEBSD-vendor is an idea that never really took off
and we should probably just remove it, but until then we might as well
record the correct version.
We need to add user-facing deprecation notices for TCP wrappers; start
with a note in the upgrade process docmentation.
Sponsored by: The FreeBSD Foundation
This includes a small battery of /memreserve/ fixes to make sure dtc is
properly writing these regions into the output file and reading them back
out.
As of this update, dtc will now also assume common defaults for -I/-O if
only one is specified; namely, dts for one implies dtb for the other and
vice versa (Requested by: jhibbits, preserves GPL dtc behavior too).
MFC after: 1 week
Assert on not specifying any of the (soon to be) required flags as well
as specifying both of them.
Pointed out by: cem, hselasky
Reviewed by: hselasky, kib
Approved by: kib (mentor)
Differential Revision: https://reviews.freebsd.org/D23678
this replaces the following near the syscall exit:
cmp $0x39,%rax
ja 0xffffffff8108f82c
movabs $0x200001800060005,%rcx
bt %rax,%rcx
jae 0xffffffff8108f82c
with:
test %edi,%edi
jne 0xffffffff8091a49c
This in particular significantly shortens amd64_syscall, which otherwise
keeps jumping forward over 2KB of code in total.
Note some of these branches should be either eliminated altogether or
coalesced.
The hack in question is intended to workaround seemingly bogus #line markers
in cpp output. As far as I can tell, llvm cpp doesn't do this by default, so
there's no reason to add -P.
In our /bin/sh, the main incantation should be placed in a sub-shell in
order to properly pipe the output to fgrep.
The main motivation for this change is admittedly to stop emitting the noise
about clang not being gcc in make -s buildworld
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D22952
The latter is a typedef of the former; the typedef exists and these bits are
representing vmprot values, so use the correct type.
Submitted by: sigsys@gmail.com
MFC after: 3 days
