grog [Sun, 1 Mar 2020 22:10:37 +0000 (22:10 +0000)]
Remove comment about Blackthorn winds, apparently imported from 4.4BSD
Lite. Nowadays it's trivial to find the explanation, such as at
https://www.deseret.com/2000/2/27/19493013/blackthorn-winds-make-bushes-bud.
It doesn't seem appropriate to replace it with an explanation.
mjg [Sun, 1 Mar 2020 21:53:46 +0000 (21:53 +0000)]
fd: move vnodes out of filedesc into a dedicated structure
The new structure is copy-on-write. With the assumption that path lookups are
significantly more frequent than chdirs and chrooting this is a win.
This provides stable root and jail root vnodes without the need to reference
them on lookup, which in turn means less work on globally shared structures.
Note this also happens to fix a bug where jail vnode was never referenced,
meaning subsequent access on lookup could run into use-after-free.
imp [Sun, 1 Mar 2020 18:18:01 +0000 (18:18 +0000)]
Remove support for FreeBSD 4-7 from the iscsi initiator.
Also, inline/remove now empty or trivial macros. CAM has evolved enough this
code couldn't work there anyway, and the API sweeep commits made since then were
made unconditional.
imp [Sun, 1 Mar 2020 17:27:30 +0000 (17:27 +0000)]
Remove all the compatibility hacks for systems that predate FreeBSD 8. Some of
these look to be cut and pasted from other drivers since this driver was
committed to FreeBSD 7-current and MFC'd to FreeBSD 6. The ones for FreeBSD 4
and 5 likely never were working...
imp [Sun, 1 Mar 2020 16:45:54 +0000 (16:45 +0000)]
Remove FreeBSD 7 conditional code... We've had a lot of other changes since then
and al_eth appears only in the Annapurna ARM designs which arrived in the tree
after that.
tijl [Sun, 1 Mar 2020 13:12:04 +0000 (13:12 +0000)]
linuxulator: Map scheduler priorities to Linux priorities.
On Linux the valid range of priorities for the SCHED_FIFO and SCHED_RR
scheduling policies is [1,99]. For SCHED_OTHER the single valid priority is
0. On FreeBSD it is [0,31] for all policies. Programs are supposed to
query the valid range using sched_get_priority_(min|max), but of course some
programs assume the Linux values are valid.
This commit adds a tunable compat.linux.map_sched_prio. When enabled
sched_get_priority_(min|max) return the Linux values and sched_setscheduler
and sched_(get|set)param translate between FreeBSD and Linux values.
Because there are more Linux levels than FreeBSD levels, multiple Linux
levels map to a single FreeBSD level, which means pre-emption might not
happen as it does on Linux, so the tunable allows to disable this behaviour.
It is enabled by default because I think it is unlikely that anyone runs
real-time software under Linux emulation on FreeBSD that critically relies
on correct pre-emption.
This fixes FMOD, a commercial sound library used by several games.
PR: 240043
Tested by: Alex S <iwtcex@gmail.com>
Reviewed by: dchagin
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D23790
cperciva [Sat, 29 Feb 2020 22:31:23 +0000 (22:31 +0000)]
Add -N option to powerd(8) to ignore "nice" time.
With powerd_flags="-N", this makes powerd(8) exclude "nice" time when
computing the CPU utilization. This makes it possible to prevent
CPU-intensive "background" processes from spinning up the CPU.
Note that only *userland* CPU usage belonging to "nice" processes is
excluded; we do not track whether time spent in the kernel is on behalf
of nice or non-nice processes, so kernel-intensive nice processes can
still result in the CPU being sped up.
csjp [Sat, 29 Feb 2020 19:17:24 +0000 (19:17 +0000)]
Currently kernel audit events for jail_set(2), jail_get(2), jail_attach(2),
jail_remove(2) and finally setloginclass(2) are not being converted and
committed into userspace. Add the cases for these syscalls and make sure
they are being converted properly.
emaste [Sat, 29 Feb 2020 17:10:54 +0000 (17:10 +0000)]
retire in-tree GPL dtc devicetree compiler
Now that we no longer have GCC 4.2.1 in the tree and can assume FreeBSD
is being built with a C++11 compiler available, we can use BSDL dtc
unconditionally and retire the GPL dtc.
GPL dtc now has FreeBSD CI support via Cirrus-CI to help ensure it
continues to build/work on FreeBSD and is available in the ports tree
if needed.
The copy of (copyfree licensed) libfdt that we actually use is in
sys/contrib/libfdt so the extra copy under contrib/dtc/libfdt can be
removed along with the rest of the GPL dtc.
Reviewed by: kevans, ian, imp, manu, theraven
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D23192
dim [Sat, 29 Feb 2020 14:11:27 +0000 (14:11 +0000)]
Fix the following -Werror warning from clang 10.0.0:
sys/arm64/arm64/identcpu.c:1170:5: error: misleading indentation; statement is not part of the previous 'if' [-Werror,-Wmisleading-indentation]
break;
^
sys/arm64/arm64/identcpu.c:1168:4: note: previous statement is here
if (fv[j].desc[0] != '\0')
^
The break should be after the if statement, indented one level less.
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D23871
emaste [Sat, 29 Feb 2020 13:15:01 +0000 (13:15 +0000)]
src.opts.mk: simplify Clang and lld bootstrap defaults
With the retirement of GCC 4.2.1 we can assume the host compiler supports
C++11, and can simplify the Clang and LLD defaults. Clang and lld are now
enabled by default everywhere, and are used as the bootstrap compiler and
linker for all targets except MIPS.
emaste [Sat, 29 Feb 2020 03:25:51 +0000 (03:25 +0000)]
remove GCC 4.2.1 build infrastructure
As described in Warner's email message[1] to the FreeBSD-arch mailing
list we have reached GCC 4.2.1's retirement date. At this time all
supported architectures either use in-tree Clang, or rely on external
toolchain (i.e., a contemporary GCC version from ports).
GCC 4.2.1 was released July 18, 2007 and was imported into FreeBSD later
that year, in r171825. GCC has served us well, but version 4.2.1 is
obsolete and not used by default on any architecture in FreeBSD. It
does not support modern C and does not support arm64 or RISC-V.
Thanks to everyone responsible for maintaining, updating, and testing
GCC in the FreeBSD base system over the years.
brooks [Fri, 28 Feb 2020 21:13:15 +0000 (21:13 +0000)]
Define SCTL_MASK32 when COMPAT_FREEBSD32 is defined.
Remove the list of architectures and depend on COMPAT_FREEBSD32 which is
defined (if relevent) in opt_global.h and thus defined everywhere in
the kernel.
This is a minor change in behavior in that 32-bit compat for sysctls now
depends on COMPAT_FREEBSD32 rather than on the potential for 32-bit
compat support. The prior arrangement may have been part of an attempt
to allow 32-bit compat to be loadable, but such attempts are doomed to
failure (due to the fact that ioctls have no meaning without the
associated file descriptor) without vastly more refactoring and some
sort of COMPAT_FREEBSD32_SUPPORT option.
pfg [Fri, 28 Feb 2020 20:43:35 +0000 (20:43 +0000)]
/etc/services: attempt to bring the database to this century 1/2.
This is the result of splitting r358153 in two, in order to avoid a build
system bug and being able to merge the change to previous releases..
Document better this file, updating the URL to the IANA registry and closely
match the official services.
For system ports (0 to 1023) we now try to follow the registry closely, noting
some historical differences where applicable.
As a side effect: drop references to unofficial Kerberos IV which was EOL'ed
on Oct 2006[1]. While it is conceivable some people may still use it in some
very old FreeBSD machines that can't be replaced easily, the use of it is
considered a security risk. Also drop the unofficial netatalk, which we
supported long ago in the kernel but was dropped long ago.
Leave for now smtps, even though it conflicts with IANA's submissions.
The change should have very little visibility, if any, but should be a
step closer to the current IANA database.
rlibby [Fri, 28 Feb 2020 18:32:36 +0000 (18:32 +0000)]
amd64 atomic.h: minor codegen optimization in flag access
Previously the pattern to extract status flags from inline assembly
blocks was to use setcc in the block to write the flag to a register.
This was suboptimal in a few ways:
- It would lead to code like: sete %cl; test %cl; jne, i.e. a flag
would just be loaded into a register and then reloaded to a flag.
- The setcc would force the block to use an additional register.
- If the client code didn't care for the flag value then the setcc
would be entirely pointless but could not be eliminated by the
optimizer.
A more modern inline asm construct (since gcc 6 and clang 9) allows for
"flag output operands", where a C variable can be written directly from
a flag. The optimizer can then use this to produce direct code where
the flag does not take a trip through a register.
In practice this makes each affected operation sequence shorter by five
bytes of instructions. It's unlikely this has a measurable performance
impact.
markj [Fri, 28 Feb 2020 16:05:18 +0000 (16:05 +0000)]
Add a blocking counter KPI.
refcount(9) was recently extended to support waiting on a refcount to
drop to zero, as this was needed for a lockless VM object
paging-in-progress counter. However, this adds overhead to all uses of
refcount(9) and doesn't really match traditional refcounting semantics:
once a counter has dropped to zero, the protected object may be freed at
any point and it is not safe to dereference the counter.
This change removes that extension and instead adds a new set of KPIs,
blockcount_*, for use by VM object PIP and busy.
Reviewed by: jeff, kib, mjg
Tested by: pho
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D23723
bz [Fri, 28 Feb 2020 11:16:41 +0000 (11:16 +0000)]
mld6: initialize oifp to avoid bogus results/panics in edge cases
In certain cases (probably not during normal operation but observed in
the lab during development) ip6_ouput() could return without error
and ifpp (&oifp) not updated.
Given oifp was never initialized we would take the later branch
as oifp was not NULL, and when calling icmp6_ifstat_inc() we would
panic dereferencing a garbage pointer.
For code stability initialize oifp to NULL before first use to always
have a deterministic value and not rely on a called function to behave
and always and for ever do the work for us as we hope for.
jkim [Thu, 27 Feb 2020 22:36:16 +0000 (22:36 +0000)]
Do not free p and g parameters after calling DH_set0_pqg(3).
It is specifically mentioned in the manual page. Note it has no functional
change in reality because DH_set0_pqg() cannot fail when both p and g are
not NULL.
dim [Thu, 27 Feb 2020 19:59:17 +0000 (19:59 +0000)]
Merge r358406 from the clang1000-import branch:
Fix the following -Werror warning from clang 10.0.0:
sys/arm/arm/identcpu-v6.c:227:5: error: misleading indentation; statement is not part of the previous 'if' [-Werror,-Wmisleading-indentation]
if (val & CPUV7_CT_CTYPE_RA)
^
sys/arm/arm/identcpu-v6.c:225:4: note: previous statement is here
if (val & CPUV7_CT_CTYPE_WB)
^
This was due to an accidentally inserted tab before the if statement.
hrs [Thu, 27 Feb 2020 19:49:59 +0000 (19:49 +0000)]
Fix poor performance of ftp(1) due to small SO_SNDBUF and SO_RCVBUF.
ftp(1) from vendor/tnftp always tried the following for
every TCP connection:
1. Get the current buffer length of SO_SNDBUF and SO_RCVBUF
by getsockopt(2).
2. Invoke setsockopt(2) to set them to the same values
after checking if they are in a range between 8 KiB to 8 MiB.
This behavior broke dynamic buffer sizing enabled by
default (net.inet.tcp.{recv,send}buf_auto sysctls) and
led to a very poor transfer rate. The fetch(1) utility
does not have this problem.
This change prevents SO_SNDBUF and SO_RCVBUF from configuring
when the buffer auto-sizing is enabled unless the buffer sizes are
explicitly specified.
hrs [Thu, 27 Feb 2020 19:40:29 +0000 (19:40 +0000)]
Fix broken STARTTLS when SharedMemoryKey is enabled.
OpenSSL 1.1 API patch for sendmail had a bug which
prevented sm_RSA_generate_key() function from working.
This function is used to generate a temporary RSA key
for a shared memory region used for TLS processing.
Note that 12.0 and 12.1-RELEASE include this bug.
This affects only if SM_CONF_SHM compile-time
option (enabled by default) and SharedMemoryKey
run-time option (not enabled by default) in a .cf file are
specified. The latter corresponds to confSHARED_MEMORY_KEY in
a .mc file.
PR: 242861
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D23734
jeff [Thu, 27 Feb 2020 19:05:26 +0000 (19:05 +0000)]
Simplify lazy advance with a 64bit atomic cmpset.
This provides the potential to force a lazy (tick based) SMR to advance
when there are blocking waiters by decoupling the wr_seq value from the
ticks value.
imp [Thu, 27 Feb 2020 15:34:30 +0000 (15:34 +0000)]
Better check for floating point type.
Use __riscv_flen instead of __riscv_float_abi_soft. While the latter works for
userland (and one could argue it's more correct), it fails for the kernel. We
compile the kernel with -mabi=lp64 (eg soft float abi) to avoid floating point
instructions in the kernel. We also compile the kernel -march=rv64imafdc for
hard float kernels (eg those with options FPE), but with -march=rv64imac for
softfloat kernels (eg those with FPE). Since we do this, in the kernel (as in
userland) __riscv_flen will be defined for 'riscv64' and not for 'riscv64sf'.
This also removes the -DMACHINE_ARCH hack now that it's no longer needed.
Longer term, we should return the ABI from the sysctl hw.machine_arch like on
amd64 for i386 binaries.
avg [Thu, 27 Feb 2020 14:12:43 +0000 (14:12 +0000)]
dsl_dataset_promote_sync: populate 'oldname' before using it
It's very unlikely that zfsvfs_update_fromname() and
zvol_rename_minors() ever did anything during the promote operation as
the old name was not initialized.
kaktus [Thu, 27 Feb 2020 13:12:14 +0000 (13:12 +0000)]
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (18 of many)
r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are
still not MPSAFE (or already are but aren’t properly marked).
Use it in preparation for a general review of all nodes.
This is non-functional change that adds annotations to SYSCTL_NODE and
SYSCTL_PROC nodes using one of the soon-to-be-required flags.
Mark all obvious cases as MPSAFE. All entries that haven't been marked
as MPSAFE before are by default marked as NEEDGIANT
jeff [Thu, 27 Feb 2020 08:23:10 +0000 (08:23 +0000)]
A pair of performance improvements.
Swap buckets on free as well as alloc so that alloc is always the most
cache-hot data.
When selecting a zone domain for the round-robin bucket cache use the
local domain unless there is a severe imbalance. This does not affinitize
memory, only locks and queues.
jeff [Thu, 27 Feb 2020 02:37:27 +0000 (02:37 +0000)]
Add unlocked grab* function variants that use lockless radix code to
lookup pages. These variants will fall back to their locked counterparts
if the page is not present.
projects / FreeBSD / FreeBSD.git / log