1. The only place in the tree which calls getnewvnode with mp == NULL does it
for vp_crossmp which will never execute this codepath. Any vnode which legally
has ->v_mount == NULL is also doomed, which once more wont execute this code.
2. Remove an assertion for v_holdcnt from production kernels. It gets taken care
of by refcount macros in debug kernels.
Any code which would want to pass NULL mp can construct a fake one instead.
Reviewed by: kib (previous version)
Differential Revision: https://reviews.freebsd.org/D22722
If any non-static modules are loaded (and mac_ntpd tends to be), the lock is
taken all the time al over the kernel. On platforms like arm64 this results in
an avoidable significant performance degradation. Since write-locking is almost
never needed, use a primitive optimized towards read-locking.
Sample result of building the kernel on tmpfs 11 times:
stock 11142.80s user 6704.44s system 4924% cpu 6:02.42 total
patched 11118.95s user 2374.94s system 4547% cpu 4:56.71 total
To be used when like rmlocks, except when sleeping for readers needs to be
allowed. See the manpage for more information.
Reviewed by: kib (previous version)
Differential Revision: https://reviews.freebsd.org/D22823
* Fix a couple of format errors.
* Add some extra compiler flags needed to force clang to build SPE code.
(These are temporary until the target triple is fixed)
Chase the removal of dev from gpioths_dht_readbytes() in r355540.
Reviewed by: ian
Approved by: will (mentor)
Differential Revision: https://reviews.freebsd.org/D22926
To improve reliability of kernel modules after the clang switch, switch to
-fPIC when building for now.
This bypasses some limitations to the way clang and LLD handle relocations,
and is a more robustly tested compilation regime than the
"static shared object" mode that we were previously attempting to convince
the compiler stack to use.
The kernel linker was recently augmented to be able to handle this mode.
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D22798
Summary:
As a transition aide, implement an alternative elfN_freebsd_fixup which
is called for old powerpc binaries. Similarly, add a translation to rtld to
convert old values to new ones (as expected by a new rtld).
Translation of old<->new values is incomplete, but sufficient to allow an
installworld of a new userspace from an old one when a new kernel is running.
Test Plan:
Someone needs to see how a new kernel/rtld/libc works with an old
binary. If if works we can probalby ship this. If not we probalby need
some more compat bits.
Submitted by: brooks
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D20799
This changes the LLVM default powerpc64 ABI to ELFv2, if target OS is
FreeBSD >= 13.0
This will also be sent upstream.
Submitted by: alfredo.junior_eldorado.org.br
Reviewed by: dim, luporl
Relnotes: YES
Differential Revision: https://reviews.freebsd.org/D20383
This enables LLVM as the default compiler for powerpc, powerpc64, and
powerpcspe, as well as LLD as the default linker for powerpc64.
LLD is not yet ready for prime time for powerpc and powerpcspe, but work is
continuing on it.
Submitted by: alfredo.junior_eldorado.org.br
Relnotes: YES
Differential Revision: https://reviews.freebsd.org/D20378
allocate them with VM_ALLOC_NOOBJ which means they are not busy. For now
move the busy assert for the new page in vm_page_replace into the public
api and out of the private api used by contig reclaim. Fix another issue
where we would leak busy if the page could not be removed from pmap.
Reported by: pho
Discussed with: markj
gvinum was the only GEOM class, using consumer nstart/nend fields. Making
it do its own accounting for orphanization purposes allows in perspective
to remove burden of that expensive for SMP accounting from GEOM.
Also the previous implementation spinned in a tight event loop, waiting
for all active BIOs to complete, while the new one knows exactly when it
is possible to close the consumer.
MFC after: 1 month
Sponsored by: iXsystems, Inc.
- Only take an ifaddr ref in in rt_exportinfo() if the caller explicitly
requests it. Take care to release it in this case.
- Don't unconditionally take a ref in rtrequest1_fib(). rt_getifa_fib()
will acquire a reference, in which case we would previously acquire
two references.
- Stop taking a reference in rtinit1() before calling rtrequest1_fib().
rtrequest1_fib() will acquire a reference for the RTM_ADD case.
PR: 242746
Reviewed by: melifaro (previous version)
Tested by: ghuckriede@blackberry.com
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D22912
Per the University California Regents letter, drop the so-called
"advertisement" clause.
Discussed with: bde, kargl (2017)
Differential Revision: https://reviews.freebsd.org/D22928
lock/lock_free decisions in compiled time
Summary:
Enables atomic.c in compiler_rt and forces clang to not emit a call for runtime
decision about lock/lock_free. At compiling time, if clang can't decide if
atomic operation can be lock free, it emits calls to external functions like
`__atomic_is_lock_free`, `__c11_atomic_is_lock_free` and
`__atomic_always_lock_free`, postponing decision to a runtime check. According
to LLVM code documentation, the mechanism exists due to differences between
x86_64 processors that can't be decided at runtime.
On PowerPC and PowerPCSPE (32 bits), we already know in advance it can't be lock
free, so we force the decision at compile time and avoid having to implement it
in an external library.
This patch was made after 32 bit users testing the PowePC32 bit ISO reported
llvm could not be compiled with in-base llvm due to `__atomic_load8` not
implemented.
Submitted by: alfredo.junior_eldorado.org.br
Reviewed by: jhibbits, dim
Differential Revision: https://reviews.freebsd.org/D22549
While the mailer is normally opened/set if the mailto is set, this is not
the case if the grandchild actually didn't produce any output. This change
corrects the situation to only attempt to kill/close the mail process if it
was actually opened in the first place.
The reporter initially stumbled on the -n (suppress mail on success) flag
leading to a SIGKILL of the process group, but simultaneously
discovered/reported the behavior with !-n jobs if MAILTO was set and no
output happened.
All of these places that are checking mailto should actually be checking
whether mail is set, so do that for consistency+correctness.
This set of bugs were introduced by r352668.
Submitted by: sigsys@gmail.com
Reported by: sigsys@gmail.com
When mount_nfs calls nmount(2), certain NFSv4 specific errors such as
NFSERR_MINORVERMISMATCH can be returned.
Without this patch, 10021 is reported as an unknown error.
This is not particulcarily serious, but make it difficult for sysadmins
to figure out why the mount attempt is failing.
This patch uses nfsv4_errstr.h to convert 10021 and similar to error strings
that can be printed out.
A positive side effect of this patch is the removal of a reference to
sys/nfsclient/nfs.h, which should no longer be used, since it is
part of the old NFS client.
This patch should only affect reporting of failed mount attempts and not the
semantics of NFS mount attempts.
[PowerPC] Only use PLT annotations if using PIC relocation model
Summary:
The default static (non-PIC, non-PIE) model for 32-bit powerpc does
not use @PLT annotations and relocations in GCC. LLVM shouldn't use
@PLT annotations either, because it breaks secure-PLT linking with
(some versions of?) GNU LD.
Update the available-externally.ll test to reflect that default mode
should be the same as the static relocation, by using the same check
prefix.
Reviewed by: sfertile
Differential Revision: https://reviews.llvm.org/D70570
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D22913
r356084 added error strings for NFSv4.1 and NFSv4.2, with the first
character capitalized. Since the other error strings were not capitalized
and these strings would usually be imbedded in an error, I decided to
make the first characters lower cased.
No real effect but more consistent.
srandom(9) is meaningless on SMP systems or any system with, say,
interrupts. One could never rely on random(9) to produce a reproducible
sequence of outputs on the basis of a specific srandom() seed because the
global state was shared by all kernel contexts. As such, removing it is
literally indistinguishable to random(9) consumers (as compared with
retaining it).
Mark random(9) as deprecated and slated for quick removal. This is not to
say we intend to remove all fast, non-cryptographic PRNG(s) in the kernel.
It/they just won't be random(9), as it exists today, in either name or
implementation.
Before random(9) is removed, a replacement will be provided and in-tree
consumers will be converted.
Note that despite the name, the random(9) interface does not bear any
resemblance to random(3). Instead, it is the same crummy 1988 Park-Miller
LCG used in libc rand(3).
Simplify RANDOM_LOADABLE by removing the ability to unload a LOADABLE
random(4) implementation. This allows one-time random module selection
at boot, by loader(8). Swapping modules on the fly doesn't seem
especially useful.
This removes the need to hold a lock over the sleepable module calls
read_random and read_random_uio.
init/deinit have been pulled out of random_algorithm entirely. Algorithms
can run their own sysinits to initialize; deinit is removed entirely, as
algorithms can not be unloaded. Algorithms should initialize at
SI_SUB_RANDOM:SI_ORDER_SECOND. In LOADABLE systems, algorithms install
a pointer to their local random_algorithm context in p_random_alg_context at
that time.
Go ahead and const'ify random_algorithm objects; there is no need to mutate
them at runtime.
LOADABLE kernel NULL checks are removed from random_harvestq by ordering
random_harvestq initialization at SI_SUB_RANDOM:SI_ORDER_THIRD, after
algorithm init. Prior to random_harvestq init, hc_harvest_mask is zero and
no events are forwarded to algorithms; after random_harvestq init, the
relevant pointers will already have been installed.
Remove the bulk of random_infra shim wrappers and instead expose the bare
function pointers in sys/random.h. In LOADABLE systems, read_random(9) et
al are just thin shim macros around invoking the associated function
pointer. We do not provide a registration system but instead expect
LOADABLE modules to register themselves at SI_SUB_RANDOM:SI_ORDER_SECOND.
An example is provided in randomdev.c, as used in the random_fortuna.ko
module.
Approved by: csprng(markm)
Discussed with: gordon
Differential Revision: https://reviews.freebsd.org/D22512
In the event of a MOD_LOAD failure, MOD_UNLOAD will be invoked to unwind
module load. Most of the reversion in MOD_LOAD can just be deferred to
normal MOD_UNLOAD cleanup, rather than duplicating the effort.
A NULL return of kbd_get_switch in the MOD_UNLOAD handler has been
downgraded from a panic to a successful return, as that certainly just means
that kbd_add_driver failed (not possible at the moment) and we have no work
to do.
r356087 made it rather innocuous to double-register built-in keyboard
drivers; we now set a flag to indicate that it's been registered and only
act once on a registration anyways. There is no misleading here, as the
follow-up kbd_delete_driver will actually remove the driver as needed now
that the linker set isn't also consulted after kbdinit.
A weak symbol here is decidedly cleaner than any #ifdef soup or relocating
kbdinit, the former leading to maintenance required on addition of any
console/keyboard drivers and the latter pushing kbd init bits away from
where they're used.
This leads to the revert of r355806; this reduces duplication in keyboard
registration and driver switch lookup and leaves us with one authoritative
source for currently registered drivers. The reduced duplication later is
nice as we have more procedure involved in keyboard setup.
keyboard_driver->flags is used to more quickly detect bogus adds/removes.
From KPI consumers' perspective, nothing changes- kbd_add_driver of an
already-registered driver will succeed, and a single kbd_delete_driver will
later remove it as expected. In contrast to historical behavior,
kbd_delete_driver on a driver registered via linker set will now actually
de-register the driver so that it may not be used -- e.g. if kbdmux's
MOD_LOAD handler fails somewhere.
Detection for already-registered drivers in kbd_add_driver has improved, as
the previous SLIST_NEXT(driver) != NULL check would not have caught a driver
that's at the tail end.
kbdinit is now called from cninit() rather than via SYSINIT so that keyboard
drivers are available as early as console drivers. This is particularly
important as cnprobe will, in both syscons and vt, attempt to do any early
configuration of keyboard drivers built-in (see: kbd_configure).
Reviewed by: imp (earlier version, pre-cninit change)
Differential Revision: https://reviews.freebsd.org/D22835
other files.
Arm and mips systems need to replace the SYSTEM_LD variable because they
need to create intermediate files which are post-processed with objcopy to
create the final .TARGET file. Previously they did so by pasting the full
expansion of SYSTEM_LD with the output filename replaced. This means
changing SYSTEM_LD in kern.pre.mk means you need to chase down anything that
replaces it and figure out how it differs so you can paste your changes in
there too.
Now there is a SYSTEM_LD_BASECMD variable that holds the entire basic kernel
linker command without the input and output files. This will allow arm and
mips makefiles to create their custom versions by refering to
SYSTEM_LD_BASECMD, which then becomes the one place where you have to make
changes to the basic linker command args.
Differential Revision: https://reviews.freebsd.org/D22921
nfsv4_errstr.h only had strings for NFSv4.0 errors. This patch adds the
errors for NFSv4.1 and NFSv4.2. At this time, this file is not used by
any sources in the tree, so the change is not significant.
I do plan on using nfsv4_errstr.h in a future patch to mount_nfs.c.
Since I am doing this patch so that "minor version mismatch" will be
recognized, I made that string less abbreviated.
the zone size and flags fields in the per-cpu caches. This allows fast
alloctions to proceed only touching the single per-cpu cacheline and
simplifies the common case when no ctor/dtor is specified.
Reviewed by: markj, rlibby
Differential Revision: https://reviews.freebsd.org/D22826
cache area. This allows us to check on bucket space for all per-cpu
buckets with a single cacheline access and fewer branches.
Reviewed by: markj, rlibby
Differential Revision: https://reviews.freebsd.org/D22825
SYSTEM_LD variable. This avoids duplicating the contents of SYSTEM_LD
from kern.pre.mk just to add the -N flag to it. If the basic linker command
ever needs to be changed, this will be one less place that has to be found
and fixed.
Some testing by kp@ indicates that the -N flag may not be needed at all,
so a comment to that effect is also added, and the -N flag may be removed
in a followup commit.
Differential Revision: https://reviews.freebsd.org/D22920
Proper locking for atkbdc will likely replace the kbdc_lock mechanism
entirely with a mutex in atkbdc_softc, so that other consumers can also
properly ensure locking protocol is followed (e.g. psm.c:doinitialize).
The first step to doing this neatly is making KBDC less opaque so that
others don't have to jump through weird casting hoops to address the mutex.
No functional change intended; this diff effectively just removes a bunch of
casting. A future change may remove the KBDC typedef entirely and just opt
for using `atkbdc_softc_c *` directly, but this was decidedly a good
intermediate step to make these changes simple to audit.
If nfsrpc_getdirpath() returns NFSERR_MINORVERMISMATCH, it would erroneously
get mapped to EIO. This was not particularily harmful, but would make it
hard for sysadmins to diagnose why an NFSv4 mount is failing.
mount_nfs.c still needs to be fixed so that it does not report
NFSERR_MINORVERMISMATCH as an unknown error 10021.
MFC after: 1 week
USE JOURNAL? [yn]
when the journal timestamp does not match the filesystem mount time
as we are just going to print an error and fall through to a full fsck.
Instead, just run a full fsck.
Requested by: Bjoern A. Zeeb (bz)
MFC after: 7 days
in clang HEAD.
There was an invisible space in the middle of the tabs, and that apprently
was enough to throw off clang's column counting.
Even if clang is "incorrect" here, it's still a style(9) violation.
This is a prerequisite for anything IFUNC in the ELFv2 / clang switch.
Since probing cpu info on powerpc is a privileged operation, define that we
pass AT_HWCAP / AT_HWCAP2 through as cpu_features and cpu_features2 to ifunc
resolvers.
This is particularly important when dealing with non-PLT GNU IFUNC, which is
not allowed to PLT call from resolvers and therefore can't access global
variables.
The naming convention "cpu_features"/"cpu_features2" is an existing FreeBSD
PowerPC convention and matches the way we treat these variables in
machine/cpu.h.
The underlying variables are u_long, however, as per the commit message for
r332868, only the low 32 bits are ever used, so the underlying flags are
compatible across all of PowerPC.
The resolver prototype is defined to reserve the maximum number of
register-passed parameters the various PowerPC ABIs allow. This leaves
plenty of room for growth without needing to resort to passing via the
stack in the future.
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D22787
This is based on DragonFly's implementation from about 2019-09-13. It
only contains the basic code and header information to identify the
disks.
Relnotes: yes
Differential Revision: https://reviews.freebsd.org/D13369
_sleep(9), wakeup(9), sleepqueue(9), et al do not dereference or modify the
channel pointers provided in any way; they are merely used as intptrs into a
dictionary structure to match waiters with wakers. Correctly annotate this
such that _sleep() and wakeup() may be used on const pointers without
invoking ugly patterns like __DECONST(). Plumb const through all of the
underlying sleepqueue bits.
No functional change.
Reviewed by: rlibby
Discussed with: kib, markj
Differential Revision: https://reviews.freebsd.org/D22914
As PowerPC is moving to clang, we can finally start taking advantage of
IFUNC.
Implement the MD parts of IFUNC handling for rtld.
Currently, it is necessary to look for R_PPC_IRELATIVE in the PLT in
addition to RELA. This is an ABI violation, but LLD9 has some .iplt bugs
that require this as a workaround.
Reviewed by: jhibbits
Differential Revision: https://reviews.freebsd.org/D22855
As PowerPC is moving to clang, we can finally start taking advantage of
IFUNC.
Implement the MD parts of IFUNC handling for rtld.
Currently, it is necessary to look for R_PPC_IRELATIVE in the PLT in
addition to RELA. This is an ABI violation, but LLD9 has some .iplt bugs
that require this as a workaround.
Reviewed by: kib
Differential Revision: https://reviews.freebsd.org/D22789
