danfe [Thu, 1 Dec 2016 14:31:05 +0000 (14:31 +0000)]
- Mention mismatching numbers in MSR vs. ACPI _PSS count warning: seeing
actual numbers would help debugging (also, `MSR' and `ACPI' are standard
abbreviations and thus should be properly capitalized)
- Rephrase unsupported AMD CPUs message and wrap as an overly long line:
`sorry' 1) is wrongly spelled after period (starts with a small letter)
and 2) carries emotional "tinge" that is unnecessary and even bogus in
debug message; `implemented' is not the best word as `supported' suits
better in this context
- Improve readability when reporting resulted P-state transition (debug)
emaste [Thu, 1 Dec 2016 14:28:37 +0000 (14:28 +0000)]
EFI loaders: parse rela relocations on amd64
Prior to this change the loader self relocation code interpreted amd64's
rela relocations as if they were rel relocations, discarding the addend.
This "works" because GNU ld 2.17.50 stores the addend value in both the
r_addend field of the relocation (as expected) and at the target of the
relocation.
Other linkers, and possibly other versions of GNU ld, won't have this
behaviour, so interpret the relocations correctly.
Reported by: George Rimar
Reviewed by: andrew
MFC after: 2 weeks
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D8681
peter [Thu, 1 Dec 2016 07:50:44 +0000 (07:50 +0000)]
Update from subversion 1.9.4 to 1.9.5.
This includes a security fix for a component that we do not build, and
two potentially useful client side fixes for reintegrate merges and tree
conflict handling. See CHANGES for full details.
imp [Thu, 1 Dec 2016 04:35:43 +0000 (04:35 +0000)]
Revert the 'performance' setting to 'NONE' from C2. C2 has issues with
USB in places, as well as having the potential for reducing
performance. Since this is used even when powerd isn't enabled, these
two problems can cause on servers. Supermicro X9 motherboards, for
example, have problems with the virtual IPMI USB keyboards and mice
attaching and detaching repeatedly. Since there are issues on some
CPUs with C2, fail safe by defaulting to not altering it.
pfg [Thu, 1 Dec 2016 01:48:56 +0000 (01:48 +0000)]
indent(1): Don't ignore newlines after comments that follow braces.
indent.c has a special loop that stores tokens from between an if () and
the next statement into a buffer. The loop ignored all newlines, but that
resulted in not calling dump_line() when it was needed to produce the
final line of the buffered up comment.
pfg [Thu, 1 Dec 2016 01:32:13 +0000 (01:32 +0000)]
indent(1): Avoid out of bound access of array in_buffer
Work-around a somewhat complex interaction within the code. From
Piotr's commit [1]:
When pr_comment() calls dump_line() for the first line of a multiline
comment, it doesn't include any indentation - it starts with the "/*".
This is consistent for both boxed and not boxed comments. Where the logic
diverges is in how it treats the rest of the lines of the comment. For box
comments indent assumes that it must not change anything, so lines are
dumped as they were, including the indentation where it exists. For the
rest of comments, it will first remove the indentation to store plain text
of the comment and then add it again where indent thinks it's appropriate
-- this is part of comment re-indenting process.
For continuations of multi-line comments, the code that handles comments
in dump_line() will use pad_output() to create indentation from the
beginning of the line (what indent calls the first column) and then write
string pointed by s_com afterwards. But if it's a box comment, the string
will include original indentation, unless it's the first line of the
comment. This is why tab characters from s_com have to be considered when
calculating how much padding is needed and the "while (*com_st == '\t')
com_st++, target += 8;" does that.
In dump_line(), /target/ is initially set to ps.com_col, so it always
assumes that indentation needs to be produced in this function, regardless
of which line of a box comment it is. But for the first line of a box
comment it is not true, so pr_comment() signals it by setting
ps.n_comment_delta, the negative comment delta, to a negative number which
is then added to /target/ in dump_line() on all lines except the first
one, so that the function produces adequate indentation in this special
case.
The bug was in how that negative offset was calculated: pr_comment() used
count_spaces() on in_buffer, which pr_comment() expected to contain
non-null terminated sequence of characters, originating from whatever
originally was on the left side of the comment. Understanding that
count_spaces() requires a string, pr_comment() temporarily set buf_ptr[-2]
to 0 in hope that it would nul-terminate the right thing in in_buffer and
calling count_spaces() would be safe and do the expected thing. This was
false whenever buf_ptr would point into save_com, an entirely different
char array than in_buffer.
The short-term fix is to recognize whether buf_ptr points into in_buffer
or save_com.
cem [Wed, 30 Nov 2016 21:59:52 +0000 (21:59 +0000)]
ioat(4): Fix 'bogus completion_pending' KASSERT
Fix ioat_release to only set is_completion_pending if DMAs were actually
queued. Otherwise, the spurious flag could trigger an assert in the
reset path on INVARIANTS kernels.
dim [Wed, 30 Nov 2016 19:06:47 +0000 (19:06 +0000)]
Cleanup old debug dirs in delete-old-dirs target
Any .debug or .symbols files under /usr/lib/debug which correspond to
OLD_FILES entries in ObsoleteFiles.inc are also automatically cleaned up
by the delete-old target. Make this also apply to any OLD_DIRS entries.
Reviewed by: emaste
MFC after: 3 days
Differential Revision: https://reviews.freebsd.org/D8683
rwatson [Wed, 30 Nov 2016 14:02:36 +0000 (14:02 +0000)]
Clarify warning message when failing to configure audit on user login:
when au_user_mask() fails, it's not a failure to set the audit mask,
but to calculate the audit mask -- and hence a condfiguration-file
issue (of some sort).
andrew [Wed, 30 Nov 2016 09:47:29 +0000 (09:47 +0000)]
Move the FDT specific parts of the GIC diver softc to the FDT attachment.
This allows the driver to be built in a kernel with no FDT support, e.g.
on arm64 with just ACPI.
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
mav [Wed, 30 Nov 2016 09:27:08 +0000 (09:27 +0000)]
Add `gmirror create` subcommand, alike to gstripe, gconcat, etc.
It is quite specific mode of operation without storing on-disk metadata.
It can be useful in some cases in combination with some external control
tools handling mirror creation and disks hot-plug.
matthew [Wed, 30 Nov 2016 07:16:29 +0000 (07:16 +0000)]
Allow a user-overridable setting 'PKG_CMD' to control the command used
to create a repo during 'make packages'
This would have been useful for a situation I found myself in where
pkg(8) had been upgraded to a version that wanted the FBSD_1.5 ABI
version but libc.so.7 had not been upgraded, and only provided
FBSD_1.4. I found I needed to update libc in order to run pkg, and I
also needed to use pkg to update libc... Which is why pkg-static
exists, but there's currently no way to tell the build system to use
pkg-static instead of pkg.
This creates a variable PKG_CMD, default value 'pkg', that can be
overridden from the command line.
jhibbits [Wed, 30 Nov 2016 02:35:51 +0000 (02:35 +0000)]
Bring the powerpc DDB disassembler into the 21st century
Bring in the most recent copy of NetBSD's db_disasm, to fix bugs and add more
instructions.
* Fix several bugs in the disassembler, most notably the disassembly of the
rlwi* instructions, the original reason for bringing in this change.
* Add more registers to the SPR list
* Add more instructions to the opcode table
emaste [Wed, 30 Nov 2016 02:00:30 +0000 (02:00 +0000)]
netmap: add cast to fix powerpc64 LINT kernel
Attempt to fix powerpc64 LINT kernel broken by r308000. Netmap's use of
a uint64_t wchan seems odd, but in the interest of minimizing this
change just cast through uintptr_t to silence the compiler warning.
Reviewed by: jhb
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D8669
mm [Tue, 29 Nov 2016 22:14:42 +0000 (22:14 +0000)]
MFV r309299:
Sync libarchive with vendor.
Important vendor bugfixes (relevant to FreeBSD):
#821: tar -P cannot extract hardlinks through symlinks
#825: Add sanity check of tar "uid, "gid" and "mtime" fields
Important vendor bugfixes (relevant to FreeBSD):
#821: tar -P cannot extract hardlinks through symlinks
#825: Add sanity check of tar "uid, "gid" and "mtime" fields
mav [Tue, 29 Nov 2016 13:48:36 +0000 (13:48 +0000)]
Make SES status updates more aggressive.
- On control request update all status pages, since they may also be
affected if user enables/disables enclosure slots.
- Periodically update element descriptors too, since there is some
hardware where they are changed dynamically.
avg [Tue, 29 Nov 2016 13:11:00 +0000 (13:11 +0000)]
bhyve: stability and performance improvement for dbgport
The TCP server implementation in dbgport does not track clients, so it
may try to write to a disconected socket resulting in SIGPIPE.
Avoid that by setting SO_NOSIGPIPE socket option.
Because dbgport emulates an I/O port to guest, the communication is done
byte by byte. Reduce latency of the TCP/IP transfers by using
TCP_NODELAY option. In my tests that change improves performance of
kgdb commands with lots of output (e.g. info threads) by two orders of
magnitude.
A general note. Since we have a uart emulation in bhyve, that can be
used for the console and gdb access to guests. So, bvmconsole and bvmdebug
could be de-orbited now. But there are many existing deployments that
still dependend on those.
kadesai [Tue, 29 Nov 2016 13:03:43 +0000 (13:03 +0000)]
This patch will unblock SYNCHRONIZE_CACHE command to firmware, i.e. don't block the SYNCHRONIZE_CACHE command at driver instead of
passing it to firmware for all Gen3 controllers.
For Thunderbolt controller, keep the legacy behavior i.e. return the SYNCHRONIZE_CACHE command with success status from driver itself.
There is Sysctl parameter 'block_sync_cache' is provided to enable customers either to block/unblock these commands to facilitate
legacy behavior if there is a compatibility issue. Default value for module parameter is to unblock this command.
kadesai [Tue, 29 Nov 2016 13:02:48 +0000 (13:02 +0000)]
Wait for AEN task to be completed(if in queue) before resetting the controller
and return without processing event in AEN thread, if controller reset is in progress.
kadesai [Tue, 29 Nov 2016 13:01:31 +0000 (13:01 +0000)]
This patch will add task management support in driver. Below is high level description:
If a SCSI IO times out, then before initiating OCR, now the driver will try to send a
target reset to the particular target for which the IO is timed out. If that also fails,
then the driver will initiate OCR.
kadesai [Tue, 29 Nov 2016 12:58:28 +0000 (12:58 +0000)]
Clean up reference to AEN command if abort AEN is succesful as the command is aborted.
Did the same by setting sc->aen_cmd = NULL when aborting AEN is successful.
kadesai [Tue, 29 Nov 2016 12:49:20 +0000 (12:49 +0000)]
Problem statement:
MFI linked list in megaraid_sas driver is used for mfi-mpt pass-through commands.
This list can be corrupted due to many possible race conditions in driver and
eventually we may see kernel panic.
One example -
MFI frame is freed from calling process as driver send command via polling method and interrupt
for that command comes after driver free mfi frame (actually even after some other context reuse
the mfi frame). When driver receive MPT frame in ISR, driver will be using the index of MFI and
access that MFI frame and finally in-used MFI frames list will be corrupted.
High level description of new solution -
Free MFI and MPT command from same context.
Free both the command either from process (from where mfi-mpt pass-through was called) or from
ISR context. Do not split freeing of MFI and MPT, because it creates the race condition which
will do MFI/MPT list.
jchandra [Tue, 29 Nov 2016 04:32:14 +0000 (04:32 +0000)]
Fix interrupt clear in pl011 uart receive function
Clear the interrupt state before reading the input char from the
input FIFO. In the current code there is a window between the read
to the data register and the write to the the ICR, during which an
input char will not cause an interrupt.
This fixes the issue by which the serial port input on QEMU freezes
when using the emulated pl011 serial port.
imp [Mon, 28 Nov 2016 21:29:01 +0000 (21:29 +0000)]
Trying to autodetect legacy setups lead to problems when people
overrode the disk image creation routine. For now, just always bring
in the legecy defines / routines.
des [Mon, 28 Nov 2016 21:00:19 +0000 (21:00 +0000)]
Use malloc()ed buffers instead of stack buffers in gr_copy() and pw_copy().
This allows pw(8) to operate on passwd and group files with longer lines
than could be accomodated by a stack buffer. It doesn't take more than a
few hundred users to exceed 8192 bytes in /etc/group.
MFC after: 3 weeks
Sponsored by: The University of Oslo
cognet [Mon, 28 Nov 2016 20:27:58 +0000 (20:27 +0000)]
Import Concurrency Kit in the kernel.
CK is a toolkit providing different lockfree algorithms/data structures.
More information can be found here : www.concurrencykit.org
dim [Mon, 28 Nov 2016 20:13:56 +0000 (20:13 +0000)]
Fix packaging for clang, lldb and lld 3.9.0
During the upgrade of clang/llvm etc to 3.9.0 in r309124, the PACKAGE
directive in the usr.bin/clang/*.mk files got dropped accidentally.
Restore it, with a few minor changes and additions:
* Correct license in clang.ucl to NCSA
* Add PACKAGE=clang for clang and most of the "ll" tools
* Put lldb in its own package
* Put lld in its own package
adrian [Mon, 28 Nov 2016 17:54:29 +0000 (17:54 +0000)]
[ath] force wake the hardware if we see a missed beacon.
This adds a workaround to incorrectly behaving APs (ie, FreeBSD APs) which
don't beacon out exactly when they should (at TBTT multiples of beacon
intervals.)
It forces the hardware awake (but leaves it in network-sleep so self
generated frames still state that the hardware is asleep!) and will
remain awake until the next sleep transition driven by net80211.
That way if the beacons are just at the wrong interval, we get a much
better chance of hearing more consecutive beacons before we go to sleep,
thus not constantly disconnecting.
Tested:
* AR9485, STA mode, against a misbehaving FreeBSD AP.
adrian [Mon, 28 Nov 2016 17:06:35 +0000 (17:06 +0000)]
[ath] revert the previous commit, after reading the 802.11-2012 spec a bit more.
The 802.11-2012 spec talks about this - section 10.1.3.2 - Beacon Generation
in Infrastructure Networks. So yes, we should be expecting beacons to be
going out in multiples of intval.
Silly adrian.
So:
* fix the FreeBSD APs that are sending beacons at incorrect TBTTs (target
beacon transmit time); and
* yes indeed we will have to wake up out of network sleep until we sync
a beacon.
mav [Mon, 28 Nov 2016 16:23:32 +0000 (16:23 +0000)]
Process port interrupt even is PxIS register is zero.
ASMedia ASM1062 AHCI chips with some fancy firmware handling PMP inside
seems sometimes forgeting to set bits in PxIS, causing command timeouts.
Removal of this check fixes the issue by the theoretical cost of slightly
higher CPU usage in some odd cases, but this is what Linux does too.
https://www.illumos.org/issues/3821
We recently had nodes with some of the latest zfs bits panic on us in a
rollback-heavy environment. The following is from my preliminary analysis:
Let's look at where we died:
> $C ffffff01ea6b9a10 taskq_dispatch+0x3a(0, fffffffff7d20450, ffffff5551dea920, 1) ffffff01ea6b9a60 zil_clean+0xce(ffffff4b7106c080, 7e0f1) ffffff01ea6b9aa0 dsl_pool_sync_done+0x47(ffffff4313065680, 7e0f1) ffffff01ea6b9b70 spa_sync+0x55f(ffffff4310c1d040, 7e0f1) ffffff01ea6b9c20 txg_sync_thread+0x20f(ffffff4313065680) ffffff01ea6b9c30 thread_start+8()
If we dig in we can find that this dataset corresponds to a zone:
> ffffff4b7106c080::print zilog_t zl_os->os_dsl_dataset->ds_dir->dd_myname
zl_os->os_dsl_dataset->ds_dir->dd_myname = [ "8ffce16a-13c2-4efa-a233- 9e378e89877b" ]
Okay so we have a null taskq pointer. That only happens during the calls to
zil_open and zil_close. If we poke around we can see that we're actually in
midst of a rollback:
> ::pgrep zfs | ::printf "0x%x %s\\n" proc_t . p_user.u_psargs
0xffffff43262800a0 zfs rollback zones/15714eb6-f5ea-469f-ac6d- 4b8ab06213c2@marlin_init
0xffffff54e22a1028 zfs rollback zones/8ffce16a-13c2-4efa-a233- 9e378e89877b@marlin_init
0xffffff4362f3a058 zfs rollback zones/0ddb8e49-ca7e-42e1-8fdc- 4ac4ba8fe9f8@marlin_init
0xffffff5748e8d020 zfs rollback zones/426357b5-832d-4430-953e- 10cd45ff8e9f@marlin_init
0xffffff436b867008 zfs rollback zones/8f36bf37-8a9c-4a44-995c- 6d1b2751e6f5@marlin_init
0xffffff4381ad4090 zfs rollback zones/6c8eca18-fbd6-46dd-ac24- 2ed45cd0da70@marlin_init
Reviewed by: Matthew Ahrens <mahrens@delphix.com>
Reviewed by: Dan Kimmel <dan.kimmel@delphix.com>
Reviewed by: Pavel Zakharov <pavel.zakharov@delphix.com>
Reviewed by: Andriy Gapon <avg@FreeBSD.org>
Approved by: Richard Lowe <richlowe@richlowe.net>
Author: George Wilson <george.wilson@delphix.com>
adrian [Mon, 28 Nov 2016 08:13:20 +0000 (08:13 +0000)]
[ath] wake up the hardware from power-save before doing transmit completion checking.
This was being done in the pre-AR9380 case, but not for AR9380 and later.
When powersave in STA mode is enabled, this may have lead to the transmit
completion code doing this:
* call the task, which doesn't wake up the hardware
* complete the frames, which doesn't touch the hardware
* schedule pending frames on the hardware queue, which DOES touch the
hardware, and this will be ignored
This would show up in the logs like this:
(with debugging enabled):
Nov 27 23:03:56 lovelace kernel: Q1[ 0] (nseg=1) (DS.V:0xfffffe011bd57300 DS.P:0x49b57300) I: 168cc117 L:00000000 F:0005
...
(in general, doesn't require debugging enabled):
Nov 27 23:03:56 lovelace kernel: ath_hal_reg_write: reg=0x00000804, val=0x49b57300, pm=2
That register is a EDMA TX FIFO register (queue 1), and the val is the descriptor
being written.
Whilst here, make sure the software queue gets kicked here.