mjg [Thu, 19 Oct 2017 01:38:31 +0000 (01:38 +0000)]
sysctl: only take mem lock if oldlen is > 4 * PAGE_SIZE
The previous limit of just one page is hit by ps.
The entire mechanism should be reworked, if not whacked. It seems the intent
is to reduce kernel dos-ability - some handlers wire the amount of memory
passed here. Handlers should probably stop wiring in the first place or in
the worst case indicate they are doing so so that the check is done only if
necessary. It should also probably be a counter, not a lock.
mjg [Wed, 18 Oct 2017 22:00:44 +0000 (22:00 +0000)]
Don't take Giant for SMP status and cpu topology sysctls.
Not only this lock doesn't play any role here, dirtying it slows down
other things a little bit as giant-held checks (e.g. DROP_GIANT) are
spread all over the kernel.
rlibby [Wed, 18 Oct 2017 19:28:28 +0000 (19:28 +0000)]
ql*_def.h: fix QL_ALIGN parenthesization
QL_ALIGN is a set of copies of roundup2, but it was missing an outer set
of parentheses, which began to matter with r324538. Now, fully copy the
parenthesization of roundup2.
ed [Wed, 18 Oct 2017 19:22:53 +0000 (19:22 +0000)]
Import the latest CloudABI definitions, version 0.16.
The most important change in this release is the removal of the
poll_fd() system call; CloudABI's equivalent of kevent(). Though I think
that kqueue is a lot saner than many of its alternatives, our
experience is that emulating this system call on other systems
accurately isn't easy. It has become a complex API, even though I'm not
convinced this complexity is needed. This is why we've decided to take a
different approach, by looking one layer up.
We're currently adding an event loop to CloudABI's C library that is API
compatible with libuv (except when incompatible with Capsicum).
Initially, this event loop will be built on top of plain inefficient
poll() calls. Only after this is finished, we'll work our way backwards
and design a new set of system calls to optimize it.
Interesting challenges will include integrating asynchronous I/O into
such a system call API. libuv currently doesn't aio(4) on Linux/BSD, due
to it being unreliable and having undesired semantics.
jhb [Wed, 18 Oct 2017 17:23:16 +0000 (17:23 +0000)]
Remove CPU_HAVEFPU.
Instead, use a runtime decision to handle COP1 traps. If floating point
support is present in the current CPU, enable saving of the floating point
state. If support is not present, fail with SIGILL.
markj [Wed, 18 Oct 2017 15:38:05 +0000 (15:38 +0000)]
Move kernel dump offset tracking into MI code.
All of the kernel dump implementations keep track of the current offset
("dumplo") within the dump device. However, except for textdumps, they
all write the dump sequentially, so we can reduce code duplication by
having the MI code keep track of the current offset. The new
dump_append() API can be used to write at the current offset.
This is needed to implement support for kernel dump compression in the
MI kernel dump code.
Also simplify dump_encrypted_write() somewhat: use dump_write() instead
of duplicating its bounds checks, and get rid of the redundant offset
tracking.
jonathan [Wed, 18 Oct 2017 00:33:20 +0000 (00:33 +0000)]
Improve computation of {BC,LL}OBJS.
Now that OBJS has grown an OBJS_SRCS_FILTER variable, use this variable
in the computation of BCOBJS and LLOBJS too. Also move BCOBJS and LLOBJS
computation to be next to the OBJS computation: this should both make
the parallel structure clearer and serve to remind people changing OBJS
that parallel changes are required in BCOBJS and LLOBJS.
A side effect of this change is that BCOBJS and LLOBJS will be available
even when LLVM_LINK has not been defined, but that seems like a positive
change: there's no reason we can't ask "what bitcode files would you
generate" just because we can't link those files together into a
complete bitcode representation of a binary or library.
jonathan [Wed, 18 Oct 2017 00:30:15 +0000 (00:30 +0000)]
Improve logic of CLEANFILES+=${PROG_FULL}.{bc,ll}.
The build rule describing how to create ${PROG_FULL}.{bc,ll} is only
dependent on LLVM_LINK being defined, not on MK_DEBUG_FILES being "yes".
Move the addition of ${PROG_FULL}.{bc,ll} out of the conditional block
under `.if ${MK_DEBUG_FILES} != "no"` and up next to where the build
rules for ${PROG_FULL}.{bc,ll} are defined.
imp [Tue, 17 Oct 2017 23:38:27 +0000 (23:38 +0000)]
Revert "Unify boot1 with loader" change r324646
Back out the unification commit to boot1. There's some issues on the
arm and arm64 platforms that need to be addressed with code
changes. There's also a discussion on arch@ about the future of
boot1.efi vs just using loader.efi that needs to play out. So take a
pause on these changes until the arm issues can be fixed and it's
clear boot1.efi will survive into FreeBSD 12.
emaste [Tue, 17 Oct 2017 21:13:26 +0000 (21:13 +0000)]
embed_mfs: add error handling, usage
Ensure that we are called with two arguments, and that the output file
is writable. Also, if we cannot find the mfs section report the output
file name rather than "kernel", as this script may be used with other
than kernels.
rlibby [Tue, 17 Oct 2017 20:37:31 +0000 (20:37 +0000)]
cxgbe: delete now-redundant vnet decls
r324539 gathered some vnet decls into netinet/tcp_var.h, so that they
are now redundant in dev/cxgbe/tom/{t4_cpl_io.c,t4_ddp.c}. This triggers
gcc -Wredundant-decls.
markj [Tue, 17 Oct 2017 19:41:45 +0000 (19:41 +0000)]
Fix a racy VI_DOOMED check in MNT_VNODE_FOREACH_ALL().
MNT_VNODE_FOREACH_ALL() is supposed to avoid returning doomed vnodes,
but the VI_DOOMED check it used was done without the vnode interlock
held, so it could race with a concurrent vgone().
emaste [Tue, 17 Oct 2017 19:11:29 +0000 (19:11 +0000)]
loader.mk: clean md.o even if MD_IMAGE_SIZE not defined
We don't normally provide special handling for optionally-included src
files, but md.o depends on both md.c and the value of ${MD_IMAGE_SIZE}.
Previously if one built with MD_IMAGE_SIZE, executed "make clean", and
then built with a different MD_IMAGE_SIZE md.o would not be rebuilt.
Reported by: Zakary Nafziger
Sponsored by: The FreeBSD Foundation
gordon [Tue, 17 Oct 2017 17:22:36 +0000 (17:22 +0000)]
Update wpa_supplicant/hostapd for 2017-01 vulnerability release.
hostapd: Avoid key reinstallation in FT handshake
Prevent reinstallation of an already in-use group key
Extend protection of GTK/IGTK reinstallation of WNM-Sleep Mode cases
Fix TK configuration to the driver in EAPOL-Key 3/4 retry case
Prevent installation of an all-zero TK
Fix PTK rekeying to generate a new ANonce
TDLS: Reject TPK-TK reconfiguration
WNM: Ignore Key Data in WNM Sleep Mode Response frame if no PMF in use
WNM: Ignore WNM-Sleep Mode Response if WNM-Sleep Mode has not been used
WNM: Ignore WNM-Sleep Mode Response without pending request
FT: Do not allow multiple Reassociation Response frames
TDLS: Ignore incoming TDLS Setup Response retries
avg [Tue, 17 Oct 2017 16:03:59 +0000 (16:03 +0000)]
never retry oustanding requests when terminating iscsi session
CAM_REQ_ABORTED sounds natural for aborting outstanding requests when
tearing down a session, but that status actually causes eligible
requests to be tried again. That's completely useless, so let's use
CAM_DEV_NOT_THERE instead. Perhaps there is a better status, but this
should be good enough. The change should affect only the session
termination.
avg [Tue, 17 Oct 2017 15:39:38 +0000 (15:39 +0000)]
iscsi: do not hold the global lock while tearing down a session
It should be sufficient to hold the lock just for removing the session
from the session list. Everything else should be covered by the session
specific lock.
On top of that, at present we can get a deadlock caused by waiting on
the CAM SIM reference count while holding the global lock. A specific
scenario involving ZFS is this:
- concurrent termination of two sessions, S1 and S2
- session S1 completed all I/Os and sleeps in CAM waiting for device
close by ZFS;
- session S2 is also dead now, but can not forcefully complete
outstanding requests by calling iscsi_session_cleanup() from
iscsi_maintenance_thread_terminate(), since it can't get the same
global sc_lock;
- as soon as there are unfinished requests, ZFS can not do
spa_config_enter() as writer, and so can not close the device for
session S1;
- deadlock.
emaste [Tue, 17 Oct 2017 02:51:45 +0000 (02:51 +0000)]
write.2: correct maximum nbytes size for EINVAL error
In FreeBSD 11 and later debug.iosize_max_clamp defaults to 0, and the
maximum nbytes count for write(2) is SSIZE_MAX. Update the man page to
document this, and mention the sysctl that can be set to obtain the
previous behaviour.
PR: 196666
MFC after: 1 week
Sponsored by: The FreeBSD Foundation
rlibby [Tue, 17 Oct 2017 01:12:17 +0000 (01:12 +0000)]
gdb kernel server: fixup Search:memory style
This is a NFC patch to move around the Search:memory implementation so
that it doesn't exceed the standard column width and doesn't take so
much vertical space in gdb_trap.
rmacklem [Mon, 16 Oct 2017 23:28:12 +0000 (23:28 +0000)]
Use taskqueue(9) to do writes/commits to mirrored DSs concurrently.
When the NFSv4.1 pNFS client is using a Flexible File Layout specifying
mirrored Data Servers, it must do the writes and commits to all mirrors.
This patch modifies the client to use a taskqueue to perform these writes
and commits concurrently.
The number of threads can't be changed for taskqueue(9), so it is set
to 4 * mp_ncpus by default, but this can be overridden by setting the
sysctl vfs.nfs.pnfsiothreads.
avos [Mon, 16 Oct 2017 21:46:11 +0000 (21:46 +0000)]
mbuf(9): unbreak m_fragment()
- Fix it by replacing m_cat() with m_prev->m_next = m_new
(m_cat() will try to append data - as a result, there will be no
fragmentation).
- Move some constants out of the loop.
kib [Mon, 16 Oct 2017 20:21:51 +0000 (20:21 +0000)]
Re-evaluate thread' signal mask after ptracestop().
The stop drops process lock, which allows the signal mask to be
changed and our selected signal might become blocked, i.e. should be
returned to the process queue instead of delivery.
Also, for the existing check of the process no longer having an
attached debugger, we should not loose the signal, but requeue it.
Reported and tested by: bdrewery
Reviewed by: jhb
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
romain [Mon, 16 Oct 2017 17:21:52 +0000 (17:21 +0000)]
Add a quick description of the geom_getxml(3), geom_xml2tree(3),
geom_gettree(3) and geom_deletetree(3) functions provided by libgeom and are
not documented in libgeom(3).
mjoras [Mon, 16 Oct 2017 16:14:50 +0000 (16:14 +0000)]
Properly reset the fields in clean_unrhdr.
In r324542 I neglected to reset the first and last fields of struct
unrhdr. This causes a tmpfs to fail the unr(9) consistency checks with
DIAGNOSTIC on. Fix this by resetting the fields by calling init_unrhdr.
While here, change a loop to use TAILQ_FOREACH_SAFE since it is more
readable and equally fast.
kib [Mon, 16 Oct 2017 15:16:24 +0000 (15:16 +0000)]
Fix the pv_chunks pc_lru tailq handling in reclaim_pv_chunk().
For processing, reclaim_pv_chunk() removes the pv_chunk from the lru
list, which makes pc_lru linkage invalid. Then the pmap lock is
released, which allows for other thread to free the last pv entry
allocated from the chunk and call free_pv_chunk(), which tries to
modify the invalid linkage.
Similarly, the chunk is inserted into the private tailq new_tail
temporary. Again, free_pv_chunk() might be run and corrupt the
linkage for the new_tail after the pmap lock is dropped.
This is a consequence of r299788 elimination of pvh_global_lock, which
allowed for reclaim to run in parallel with other pmap calls which
free pv chunks.
As a fix, do not remove the chunk from pc_lru queue, use a marker to
remember the position in the queue iteration. We can safely operate
on the chunks after the chunk's pmap is locked, we fetched the chunk
after the marker, and we checked that chunk pmap is same as we have
locked, because chunk removal from pc_lru requires both pv_chunk_mutex
and the pmap mutex owned.
Note that the fix lost an optimization which was present in the
previous algorithm. Namely, new_tail requeueing rotated the pv chunks
list so that reclaim didn't scan the same pv chunks that couldn't be
freed (because they contained a wired and/or superpage mapping) on
every invocation. An additional change is planned which would improve
this.
Reported and tested by: pho
Reviewed by: alc
Sponsored by: The FreeBSD Foundation
MFC after: 1 week
kp [Mon, 16 Oct 2017 15:03:45 +0000 (15:03 +0000)]
pf tests: Basic IPv6 forwarding tests
Pass/block packets in the forwarding path with pf.
Introduce the pft_set_rules() helper function, because we need to
remember to flush states between individual tests. If not we can get
packets passing despite rules blocking them because they match states
created in a previous test.
Extend pft_ping.py to be able to send IPv6 echo requests.
mmel [Mon, 16 Oct 2017 12:53:54 +0000 (12:53 +0000)]
Save VFP state in getcontext(3) on ARM.
This is a last followup of r315974, which fixes userland part
of VFP save/restore problems described in PR 217611.
imp [Mon, 16 Oct 2017 03:59:28 +0000 (03:59 +0000)]
Move common/Makefile.inc to sys/boot/loader.mk.
Makefile.inc has a specific meaning in the tree, and
common/Makefile.inc doesn't quite fit into that. Rename it to
loader.mk and it will be a place to collect common things to all
/boot/loader programs there.
imp [Sun, 15 Oct 2017 23:53:55 +0000 (23:53 +0000)]
Closer examination shows that nvme and CAM both normally zero-fill
allocations (for req and ccb, which ultimately contain the
nvme_cmd). As such, we can micro-optimize these routines. Add a
comment to this effect, and bzero the ccb used to make the requests
for the nda dump rotuine so it more closely matches a ccb allocated
with xpt_get_ccb().
rmacklem [Sun, 15 Oct 2017 22:22:27 +0000 (22:22 +0000)]
Fix the client IP address reported by nfsdumpstate for 64bit arch and NFSv4.1.
The client IP address was not being reported for some NFSv4 mounts by
nfsdumpstate. Upon investigation, two problems were found for mounts
using IPv4. One was that the code (originally written and tested on i386)
assumed that a "u_long" was a "uint32_t" and would exactly store an
IPv4 host address. Not correct for 64bit arches.
Also, for NFSv4.1 mounts, the field was not being filled in. This was
basically correct, because NFSv4.1 does not use a callback address.
However, it meant that nfsdumpstate could not report the client IP addr.
This patch should fix both of these issues.
For IPv6, the address will still not be reported. The original NFSv4 RFC
only specified IPv4 callback addresses. I think this has changed and, if so,
a future commit to fix reporting of IPv6 addresses will be needed.
tijl [Sun, 15 Oct 2017 19:28:14 +0000 (19:28 +0000)]
Add information needed by Linux libdrm 2.4.74 (shipped with CentOS 7.4).
Create a config file for PCI devices that exposes their configuration
space. Only fields needed by libdrm are filled in (vendor, device,
revision, subvendor and subdevice).
Link /sys/class/drm/card%d/device to the PCI device directory.
tijl [Sun, 15 Oct 2017 18:53:21 +0000 (18:53 +0000)]
When a Linux program tries to access a /path the kernel tries
/compat/linux/path before /path. Stop following symbolic links when
looking up /compat/linux/path so dead symbolic links aren't ignored.
This allows syscalls like readlink(2) and lstat(2) to work on such links.
And open(2) will return an error now instead of trying /path.
imp [Sun, 15 Oct 2017 16:19:09 +0000 (16:19 +0000)]
Use nvme_ctrlr_poll instead of nvme_ctrlr_intx_handler since it is
more general and doesn't try to access registers that may be undefined
when the card is in MSIX mode.
This change, along with r324630, r324631, r324632, makes nda crash
dumps work again. Previously, they only worked on CPU 0 when the stack
garbage was just so.
Sponsored by: Netflix
Suggested by: scottl@ (who provided earlier version of the patch)
imp [Sun, 15 Oct 2017 16:18:08 +0000 (16:18 +0000)]
Create general polling function for the nvme controller. Use it when
we're doing the various pin-based interrupt modes. Adjust
nvme_ctrlr_intx_handler to use nvme_ctrlr_poll.
imp [Sun, 15 Oct 2017 16:18:03 +0000 (16:18 +0000)]
Be nicer on the dump stack by allocating only a ccb_nvmeio rather than
a full ccb. This saves a few hundre bytes, which might be important
during a crash dump...
imp [Sun, 15 Oct 2017 16:17:59 +0000 (16:17 +0000)]
Explicitly set reserved fields and 'fuse' to 0. This prevents us from
acidentally sending bogus values in these fields, which some drives
may reject with an error or worse (undefined behavior).
This is especially needed for the ndadump routine which allocates the
cmd from stack garbage....
cy [Sun, 15 Oct 2017 02:40:13 +0000 (02:40 +0000)]
Sync (make same) the offsetof macro definition in include/ with the
definition of the same in sys/sys/. The problem was discovered while
working on implementing a new C11 gets_s() for libc. (The new gets_s()
requires rsize_t found in include/stddef.h.) The solution to sync the two
definitions was suggested by ed@ while discussing D12667.
brooks [Sat, 14 Oct 2017 18:38:36 +0000 (18:38 +0000)]
Switch procstat from subcommand flags to verbs
- Use an enumerated value instead of separate flags for commands
- Look for a verb if no command flag is set
- Lookup the "xocontainer" value based on the command
- Document the new command verbs in the man-page
mjg [Fri, 13 Oct 2017 21:54:34 +0000 (21:54 +0000)]
Reduce traffic on vm_cnt.v_free_count
The variable is modified with the highly contended page free queue lock.
It unnecessarily shares a cacheline with purely read-only fields and is
re-read after the lock is dropped in the page allocation code making the
hold time longer.
Pad the variable just like the others and store the value as found with
the lock held instead of re-reading.
Provides a modest 1%-ish speed up in concurrent page faults.
projects / FreeBSD / FreeBSD.git / log