kevans [Sun, 10 Jun 2018 02:36:38 +0000 (02:36 +0000)]
lualoader: Support variable substitution in env var settings
We support both of the following cases of substitution:
bar="y"
foo="${bar}"
foo="$bar"
The latter substitution syntax is, of course, not recommended- all
punctuation must be considered potential variable names, and we do not go
through the effort of searching the different combinations of, for instance,
"$x.y.z" to determine if the variable is $x, $x.y, or $x.y.z.
This is not officially documented as supported, but it has worked in
forthloader for what is most likely a long time as `evaluate` is used to
process the right hand side of the assignment.
imp [Sun, 10 Jun 2018 02:30:09 +0000 (02:30 +0000)]
Remove old, dead compat code.
We no longer need to od these things conditionally, and the fallbacks
are to 4.2BSD era defaults, which nobody uses anymore. Vixie cron has
diverged from upstream anyway in our tree, and it's not clear there's
actually a viable upstream anymore. Plus, we don't follow the
vendor-supplied code pattern here.
I'm doing this to reduce false positives from grep.
kevans [Sun, 10 Jun 2018 01:38:52 +0000 (01:38 +0000)]
lualoader: Process loader_conf_files properly
loader.conf(5) documents loader_conf_files to mean "additional configuration
files to be processed right after the present file." However, lualoader
ignored loader_conf_files after processing /boot/defaults/loader.conf.
Rewrite these bits to process loader_conf_files after each loaded file.
eadler [Sun, 10 Jun 2018 00:02:56 +0000 (00:02 +0000)]
top(1): permit sub-second delay times
This removes the getuid check for delay==0. It didn't prevent users from
writing similar programs in the general case. In theory, if top(1) is
among one of the few restricted programs you're allowed to run, it may
have helped a little, but there are better ways of handling that case.
rmacklem [Sat, 9 Jun 2018 23:38:22 +0000 (23:38 +0000)]
Delete some macros that are unused.
These macros were added because they were used by the pNFS server last
year. However, they are no longer used by the pNFS server code and
might as well be deleted.
This is a partial reversion of r326735.
rmacklem [Sat, 9 Jun 2018 23:14:59 +0000 (23:14 +0000)]
Delete an unused macro and clean up a comment about it.
NFSDEV_MIRRORSTR was defined for the pNFS server, but has not been used,
so this patch deletes it. It also cleans up the comment and hopefully
makes it more readable.
bapt [Sat, 9 Jun 2018 21:09:24 +0000 (21:09 +0000)]
diff: Import fixes from OpenBSD
original commit log by miller@OpenBSD r1.46:
Fix exit value when diffing directories with missing files and the -N
or -P options are not used. From Ibrahim Khalifa
kevans [Sat, 9 Jun 2018 18:11:46 +0000 (18:11 +0000)]
bsdgrep(1): Some more int -> bool conversions and name changes
Again motivated by upcoming work to rewrite a bunch of this- single-letter
variable names and slightly misleading variable names ("lastmatches" to
indicate that the last matched) are not helpful.
markj [Sat, 9 Jun 2018 15:10:49 +0000 (15:10 +0000)]
Don't process DWARF generated from non-C/C++ code.
ctfconvert(1) is not designed to handle DWARF generated from such code,
and will generally fail in non-obvious ways. Use an explicit check to
help catch such potential failures.
Reported by: Johannes Lundberg <johalun0@gmail.com>
MFC after: 2 weeks
kevans [Sat, 9 Jun 2018 15:10:39 +0000 (15:10 +0000)]
stand: Consolidate checking for boot flags driven by environment vars
e.g. boot_mute, boot_single, boot_verbose, and friends; we checked for these
in multiple places, consolidate into common/ and allow a setting of "NO" for
any of these to turn them off. This allows systems with multiple
loader.conf(5) or loader.conf(5) overlay systems to easily turn off
variables in later processed files by setting it to NO.
Reported by: Nick Wolff @ iXsystems
Reviewed by: imp
andrew [Sat, 9 Jun 2018 14:47:49 +0000 (14:47 +0000)]
In the ThunderX BGX network driver we were skipping the NULL terminator
when parsing the phy type, however this is included in the length returned
by OF_getprop. To fix this stop ignoring the terminator.
If a locally generated packet is routed (with route-to/reply-to/dup-to) out of
a different interface it's passed through the firewall again. This meant we
lost the inp pointer and if we required the pointer (e.g. for user ID matching)
we'd deadlock trying to acquire an inp lock we've already got.
Pass the inp pointer along with pf_route()/pf_route6().
Explicitly change the link state when we assingn an address.
Since we are setting IFF_UP flag on SIOCSIFADDR, it is possible, that
after this link state information still not initialized properly.
This leads to problems with routing, since now interface has
IFCAP_LINKSTATE capability and a route is considered as working only
when interface's link state is in LINK_STATE_UP (see RT_LINK_IS_UP()
macro).
eadler [Sat, 9 Jun 2018 03:08:04 +0000 (03:08 +0000)]
LOCKS: update current locks
- use a wild card for releng instead of listing them out. This ignores a
technicality that re@ owns the branches during release cycles, but this
has not been document in this file in a long time (if ever).
eadler [Sat, 9 Jun 2018 02:14:33 +0000 (02:14 +0000)]
top(1): use a different command to toggle tid vs pid
- By popular demand, implement a different switch ("T") for toggling
between thread id and process id.
- Add an assert that the size of command chars is as expected.
- Also clean up some messiness I found when implementing this.
- Further document the new flag.
shurd [Fri, 8 Jun 2018 21:49:19 +0000 (21:49 +0000)]
Remove tx task spinning added in r333686
This caused issues with PASTE. Just remove the reschedule since the DELAY()
should be enough for use cases such as pkt-gen which were failing before the
change.
mjg [Fri, 8 Jun 2018 21:40:03 +0000 (21:40 +0000)]
uma: implement provisional api for per-cpu zones
Per-cpu zone allocations are very rarely done compared to regular zones.
The intent is to avoid pessimizing the latter case with per-cpu specific
code.
In particular contrary to the claim in r334824, M_ZERO is sometimes being
used for such zones. But the zeroing method is completely different and
braching on it in the fast path for regular zones is a waste of time.
kib [Fri, 8 Jun 2018 18:32:26 +0000 (18:32 +0000)]
Restore release semantic for the old thread unlock on arm64.
With the introduction of pmap_switch(), the DSB instruction on the
address map switch is not necessary executed, which is fixed by
changing the unlock store to release. Also remove comment which
documented pre-pmap_switch() code.
Reviewed by: andrew
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
adrian [Fri, 8 Jun 2018 18:21:57 +0000 (18:21 +0000)]
[ath_hal] Return failure if noise floor calibration fails.
If we fail noise floor calibration then we may end up with a deaf NIC
which we can't recover without a full chip reset.
Earlier chips seem to get less stuck in this condition versus AR9280/later
and AR9300/later, but whilst here just fix up the AR5212 era chips to also
return NF calibration failures.
This HAL routine would only return failure if the channel was not configured.
This is a no-op until the driver side code for doing resets and the HAL
code for being told about the reset type (and then handling it!) is
implemented.
Tested:
* AR9280, STA mode
* AR2425, STA mode
* AR9380, STA mode
During scans (scrubs or resilvers), it sorts the blocks in each transaction
group by block offset; the result can be a significant improvement. (On my
test system just now, which I put some effort to introduce fragmentation into
the pool since I set it up yesterday, a scrub went from 1h2m to 33.5m with the
changes.) I've seen similar rations on production systems.
Approved by: Alexander Motin
Obtained from: ZFS On Linux
Relnotes: Yes (improved scrub performance, with tunables)
Differential Revision: https://reviews.freebsd.org/D15562
mjg [Fri, 8 Jun 2018 05:40:36 +0000 (05:40 +0000)]
uma: fix up r334824
Turns out there is code which ends up passing M_ZERO to counters.
Since counters zero unconditionally on their own, just ignore drop the
flag in that place.
mmacy [Fri, 8 Jun 2018 04:58:03 +0000 (04:58 +0000)]
hwpmc: simplify calling convention for hwpmc interrupt handling
pmc_process_interrupt takes 5 arguments when only 3 are needed.
cpu is always available in curcpu and inuserspace can always be
derived from the passed trapframe.
While facially a reasonable cleanup this change was motivated
by the need to workaround a compiler bug.
In the process of optimizing the tail call the tf pointer was getting
clobbered:
(kgdb) up
at /storage/mmacy/devel/freebsd/sys/dev/hwpmc/hwpmc_mod.c:4709
4709 pmc_save_kernel_callchain(ps->ps_pc,
(kgdb) up
1205 error = pmc_process_interrupt(cpu, PMC_HR, pm, tf,
resulting in a crash in pmc_save_kernel_callchain.
mjg [Fri, 8 Jun 2018 03:16:16 +0000 (03:16 +0000)]
uma: remove M_ZERO support for pcpu zones
Nothing in the tree uses it and pcpu zones have a fundamentally different use
case than the regular zones - they are not supposed to be allocated and freed
all the time.
This reduces pollution in the allocation fast path.
kevans [Fri, 8 Jun 2018 01:25:07 +0000 (01:25 +0000)]
bsdgrep(1): Slooowly peel away the chunky onion
(or peel off the band-aid, whatever floats your boat)
This addresses two separate issues:
1.) Nothing within bsdgrep actually knew whether it cared about line numbers
or not.
2.) The file layer knew nothing about the context in which it was being
called.
#1 is only important when we're *not* processing line-by-line. #2 is
debatably a good idea; the parsing context is only handy because that's
where we store current offset information and, as of this commit, whether or
not it needs to be line-aware.
mjg [Fri, 8 Jun 2018 00:47:24 +0000 (00:47 +0000)]
amd64: fix a retarded bug in memset
memset fills the target buffer from a byte-sized value passed in as the
second argument.
The fully-sized (8 bytes) register containing it is named %rsi. Lower 4 bytes
can be referred to as %esi and finally the lowest byte is %sil.
Vast majority of all the callers just zero the target buffer and set it up by
doing xor %esi,%esi which has a side-effect of zeroing the upper parts of
the register as well. Some others do a word-sized move to %esi which has the
same result.
However, there are callers which only fill %sil. This does *not* clear up
the rest of the register.
The value of %rsi is multiplied by $0x0101010101010101 to create a 8-byte sized
pattern for 8-byte stores.
Prior to the patch, the func just blindly took %rsi assuming the unwanted bytes
are zeroed out. Since this is not the case for the callers which only play with
%sil (the rest of the register can have absolutely anything), the resulting
pattern can be garbage.
This has potential for funny bugs. One side effect (which was not amusing)
after enabling it instead of bzero was that the kernel was hanging on boot
as a xen domU.
Reported by: Trond Endrestøl <Trond.Endrestol fagskolen.gjovik.no>
Pointy hat: me
glebius [Fri, 8 Jun 2018 00:15:08 +0000 (00:15 +0000)]
UMA memory debugging enabled with INVARIANTS consists of two things:
trashing freed memory and checking that allocated memory is properly
trashed, and also of keeping a bitset of freed items. Trashing/checking
creates a lot of CPU cache poisoning, while keeping debugging bitsets
consistent creates a lot of contention on UMA zone lock(s). The performance
difference between INVARIANTS kernel and normal one is mostly attributed
to UMA debugging, rather than to all KASSERT checks in the kernel.
Add loader tunable vm.debug.divisor that allows either to turn off UMA
debugging completely, or turn it on only for a fraction of allocations,
while still running all KASSERTs in kernel. That allows to run INVARIANTS
kernels in production environments without reducing load by orders of
magnitude, but still doing useful extra checks.
Default value is 1, meaning debug every allocation. Value of 0 would
disable UMA debugging completely. Values above 1 enable debugging only
for every N-th item. It isn't possible to strictly follow the number,
but still amount of debugging is reduced roughly by (N-1)/N percent.
glebius [Thu, 7 Jun 2018 22:38:40 +0000 (22:38 +0000)]
Add new functionality and syntax to cron(1) to allow to run jobs at a
given interval, which is counted in seconds since exit of the previous
invocation of the job. Example user crontab entry:
@25 sleep 10
The example will launch 'sleep 10' every 35 seconds. This is a rather
useless example above, but clearly explains the functionality.
The practical goal here is to avoid overlap of previous job invocation
to a new one, or to avoid too short interval(s) for jobs that last long
and doesn't have any point of immediate launch soon after previous run.
Another useful effect of interval jobs can be noticed when a cluster of
machines periodically communicates with a single node. Running the task
time based creates too much load on the node. Running interval based
spreads invocations across machines in cluster. Note that -j/-J won't
help in this case.
rrs [Thu, 7 Jun 2018 19:57:55 +0000 (19:57 +0000)]
Fix build issue with const and volatile and the
myriad ways that the various compliers treat this. The
only safe prefetch appears to be for AMD. The other
compilers either are not volatile or are not const :(
benno [Thu, 7 Jun 2018 18:59:32 +0000 (18:59 +0000)]
Break recursion involving getnewvnode and zfs_rmnode.
When we're at our vnode limit, getnewvnode will call into the vnode LRU
cache to free up vnodes. If the vnode we try to recycle is a ZFS vnode we
end up, eventually, in zfs_rmnode. If the ZFS vnode we're recycling
represents something with extended attributes, zfs_rmnode will call
zfs_zget which will attempt to allocate another vnode. If the next vnode we
try to recycle is also a ZFS vnode representing something with extended
attributes we can recurse further. This ends up being unbounded and can end
up overflowing the stack.
In order to avoid this, restructure zfs_rmnode to simply add the extended
attribute directory's object ID to the unlinked set, thus not requiring the
allocation of a vnode. We then schedule a task that calls zfs_unlinked_drain
which will do the work of properly marking the vnodes for unlinking.
zfs_unlinked_drain is also called on mount so these will be cleaned up
there.
kevans [Thu, 7 Jun 2018 18:38:48 +0000 (18:38 +0000)]
bsdgrep(1): Don't initialize fts_flags twice
Admittedly, this is a clang-scan complaint... but it wasn't wrong. fts_flags
is initialized by all cases in the switch(), which should be fairly obvious.
Annotate this anyways.
kevans [Thu, 7 Jun 2018 18:27:58 +0000 (18:27 +0000)]
bsdgrep(1): Do some less dirty things with return types
Neither procfile nor grep_tree return anything meaningful to their callers.
None of the callers actually care about how many lines were matched in all
of the files they processed; it's all about "did anything match?"
This is generally just a light refactoring to remind me of what actually
matters as I'm rewriting these bits to care less about 'stuff'.
marius [Thu, 7 Jun 2018 18:24:25 +0000 (18:24 +0000)]
- Once we have shifted arguments up to thrice, base-bits-dir is $1 rather
than $4. Introduce $BASEBITSDIR for clarity and to avoid repeating this
mistake in the future. Fixing this ensures that we pick up newly built
boot bits native to the target rather for/from the host.
- Apply some of the argument quoting fixes done in r287635 but missing in
later revisions.
rrs [Thu, 7 Jun 2018 18:18:13 +0000 (18:18 +0000)]
This commit brings in a new refactored TCP stack called Rack.
Rack includes the following features:
- A different SACK processing scheme (the old sack structures are not used).
- RACK (Recent acknowledgment) where counting dup-acks is no longer done
instead time is used to knwo when to retransmit. (see the I-D)
- TLP (Tail Loss Probe) where we will probe for tail-losses to attempt
to try not to take a retransmit time-out. (see the I-D)
- Burst mitigation using TCPHTPS
- PRR (partial rate reduction) see the RFC.
Once built into your kernel, you can select this stack by either
socket option with the name of the stack is "rack" or by setting
the global sysctl so the default is rack.
Note that any connection that does not support SACK will be kicked
back to the "default" base FreeBSD stack (currently known as "default").
To build this into your kernel you will need to enable in your
kernel:
makeoptions WITH_EXTRA_TCP_STACKS=1
options TCPHPTS
Sponsored by: Netflix Inc.
Differential Revision: https://reviews.freebsd.org/D15525
kib [Thu, 7 Jun 2018 17:04:34 +0000 (17:04 +0000)]
Account for dmap limit when selecting the pages for the bootstrap
pagetables.
physmap[] can be inconsistent with the physical memory limit due to
buggy bios, or to the hw.physmem tunable. Since bootstrap pagetables
are initialized by accesses through the DMAP, we must ensure that DMAP
really cover the selected pages. This is only relevant when machine
has less than 4G RAM and buggy BIOS, which is the combination on Acer
Chromebook 720.
The call to mp_bootaddress() is moved later to have Maxmem initialized.
An alternative could be to always cover 4G for DMAP, but this change
seems to be simpler.
Reported and tested by: grembo
Reviewed by: royger
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
Differential revision: https://reviews.freebsd.org/D15675
bdrewery [Thu, 7 Jun 2018 16:16:22 +0000 (16:16 +0000)]
Stop using head(1) which is not available in installworld.
installworld should not be executing this anyhow but there is some
obscure case doing it still. The head(1) binary is not part of
ITOOLS and there's no need to add it.
avg [Thu, 7 Jun 2018 14:46:52 +0000 (14:46 +0000)]
x86: reorganize code that deals with unexpected NMI-s
Expected NMI-s are those than are either generated by the software (such
as a CPU sending NMI to other CPU) or generated by the hardware after
the software configured it to do so (such as NMI-s on PMC events).
Some unexpected NMI-s can be caused by hardware failures and it is
possible to inquire the hardware about them (somewhat like MCA but much
more primitive) using an EISA mechanism. In some cases the origin of
the NMI can remain truly unknown.
This commit should not change any functionality. It just reorganizes
the code, so that it is easier to extend with new checks for the origin
of the NMI. Also, it frees the code that has nothing to do with ISA
from DEV_ISA.