When large SPDs are used, we face two problems:
- too many CPU cycles are spent during the linear searches in the SPD
for each packet
- too much contention on multi socket systems, since we use a single
shared lock.
Main changes:
- added the sysctl tree 'net.key.spdcache' to control the SPD cache
(disabled by default).
- cache the sp indexes that are used to perform SP lookups.
- use a range of dedicated mutexes to protect the cache lines.
Submitted by: Emeric Poupon <emeric.poupon@stormshield.eu>
Reviewed by: ae
Sponsored by: Stormshield
Differential Revision: https://reviews.freebsd.org/D15050
If a thread waiting on sx dropped Giant it would not be properly
reacquired on exit from the routine, later resulting in panics
indicating Giant is not held (when it should be).
The bug was not present in the original patch sent to pho, I wittingly
added it just prior to the commit and only smoke-tested it.
Reported by: pho
A constant stream of readers could completely starve writers and this is not
a hypothetical scenario.
The 'poll2_threads' test from the will-it-scale suite reliably starves writers
even with concurrency < 10 threads.
The problem was run into and diagnosed by dillon@backplane.com
There was next to no change in lock contention profile during -j 128 pkg build,
despite an sx lock being at the top.
Tested by: pho
Writers waiting on readers to finish can set the RW_LOCK_WRITE_SPINNER
bit. This prevents most new readers from coming on. However, the last
reader to unlock also clears the bit which means new readers can sneak
in and the cycle starts over.
Change the code to keep the bit after last unlock.
Note that starvation potential is still there: no matter how many write
spinners are there, there is one bit. After the writer unlocks, the lock
is free to get raided by readers again. It is good enough for the time
being.
The real fix would include counting writers.
This runs into a caveat: the writer which set the bit may now be preempted.
In order to get rid of the problem all attempts to set the bit are preceeded
with critical_enter.
The bit gets cleared when the thread which set it goes to sleep. This way
an invariant holds that if the bit is set, someone is actively spinning and
will grab the lock soon. In particular this means that readers which find
the lock in this transient state can safely spin until the lock finds itself
an owner (i.e. they don't need to block nor speculate how long to spin
speculatively).
Tested by: pho
I have a system that is very unstable after resuming from suspend-to-RAM
but only if HPET is used as the event timer. The theory is that SMM
code / firmware could be enabling HPET for its own uses and unexpected
interrupts cause a trouble for it. Originally I wanted to solve the
problem in hpet_suspend() method, but that was insufficient as the event
timer could get reprogrammed again.
So, it's better, for my case and in general, to stop the event timer(s)
before entering the hardware suspend.
MFC after: 4 weeks
Differential Revision: https://reviews.freebsd.org/D15413
They're only useful when multiple threads may share an epoch record,
and that can't happen with non-preemptible sections.
Reviewed by: mmacy
Differential Revision: https://reviews.freebsd.org/D15507
The interlock pointer is non-NULL by definition and the compiler see through
that and eliminates the NULL checks. Just remove them from the code as they
play no role.
No difference in generated assembly.
two copyrights. The line originated with the Berkeely Regents, who
we have not approached about removing it (it's honestly too trivial
to be worth that fight). Restore it to rwatson's line as well. He
can decide if he wants it or not on his own. Matt clearly doesn't
want it, per project preference and his own statements on IRC.
Noticed by: rgrimes@
Archs using in-tree gcc were broken with `warning: redundant
redeclaration of 'cap_no_rights' [-Wredundant-decls]`.
Sponsored by: The FreeBSD Foundation
