Andrew Turner [Tue, 17 May 2016 12:46:50 +0000 (12:46 +0000)]
Clean up the GICv3 intrng code:
* In gic_v3_attach free the correct data on failure.
* Implement gic_v3_teardown_intr.
* Update the panic string when enabling/disabling an invalid interrupt.
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
Add implementation of robust mutexes, hopefully close enough to the
intention of the POSIX IEEE Std 1003.1TM-2008/Cor 1-2013.
A robust mutex is guaranteed to be cleared by the system upon either
thread or process owner termination while the mutex is held. The next
mutex locker is then notified about inconsistent mutex state and can
execute (or abandon) corrective actions.
The patch mostly consists of small changes here and there, adding
neccessary checks for the inconsistent and abandoned conditions into
existing paths. Additionally, the thread exit handler was extended to
iterate over the userspace-maintained list of owned robust mutexes,
unlocking and marking as terminated each of them.
The list of owned robust mutexes cannot be maintained atomically
synchronous with the mutex lock state (it is possible in kernel, but
is too expensive). Instead, for the duration of lock or unlock
operation, the current mutex is remembered in a special slot that is
also checked by the kernel at thread termination.
Kernel must be aware about the per-thread location of the heads of
robust mutex lists and the current active mutex slot. When a thread
touches a robust mutex for the first time, a new umtx op syscall is
issued which informs about location of lists heads.
The umtx sleep queues for PP and PI mutexes are split between
non-robust and robust.
Somewhat unrelated changes in the patch:
1. Style.
2. The fix for proper tdfind() call use in umtxq_sleep_pi() for shared
pi mutexes.
3. Removal of the userspace struct pthread_mutex m_owner field.
4. The sysctl kern.ipc.umtx_vnode_persistent is added, which controls
the lifetime of the shared mutex associated with a vnode' page.
Reviewed by: jilles (previous version, supposedly the objection was fixed)
Discussed with: brooks, Martin Simmons <martin@lispworks.com> (some aspects)
Tested by: pho
Sponsored by: The FreeBSD Foundation
Randall Stewart [Tue, 17 May 2016 09:53:22 +0000 (09:53 +0000)]
This small change adopts the excellent suggestion for using named
structures in the add of a new tcp-stack that came in late to me
via email after the last commit. It also makes it so that a new
stack may optionally get a callback during a retransmit
timeout. This allows the new stack to clear specific state (think
sack scoreboards or other such structures).
Sponsored by: Netflix Inc.
Differential Revision: http://reviews.freebsd.org/D6303
Andriy Gapon [Tue, 17 May 2016 07:56:05 +0000 (07:56 +0000)]
zfs_ioc_rename: fix a reversed condition
FreeBSD zfs_ioc_rename() has an option, not present upstream, that
allows to rename snapshots without unmounting them first. I am not sure
what is a rationale for that option, but its actual behavior was the
opposite of the intended behavior. That is, by default the snapshots
were not unmounted.
The option was introduced as part of a large update from upstream in
r248498.
One of the consequences was a havoc under .zfs/snapshot after the rename.
The snapshots got new names but were mounted on top of directories with
old names, so readdir would list the new names, but lookup would still
find the old mounts.
Add initial support for negotiating iSER parameters to iscsid(8). Some
rework might be needed to support asymetrical limits, but this should be
ok for now.
Obtained from: Mellanox Technologies (earlier version)
MFC after: 1 month
Sponsored by: The FreeBSD Foundation
Make named objects set-aware. Now it is possible to create named
objects with the same name in different sets.
Add optional manage_sets() callback to objects rewriting framework.
It is intended to implement handler for moving and swapping named
object's sets. Add ipfw_obj_manage_sets() function that implements
generic sets handler. Use new callback to implement sets support for
lookup tables.
External actions objects are global and they don't support sets.
Modify eaction_findbyname() to reflect this.
ipfw(8) now may fail to move rules or sets, because some named objects
in target set may have conflicting names.
Note that ipfw_obj_ntlv type was changed, but since lookup tables
actually didn't support sets, this change is harmless.
Adrian Chadd [Tue, 17 May 2016 07:15:25 +0000 (07:15 +0000)]
[bwn] add in bwn n-phy linking.
* The default kernel and options won't build the GPL PHY bits;
* bwn(4) defaults to building as a module anyway!;
* If BWN_GPL_PHY is specified in the config file, and you uncomment
the GPL PHY bits in the module Makefile, you'll get a working
N-PHY.
This is specifically designed to be obtuse for now, as I don't want
to flip it on by default. It's easy enough for people to flip on
and build, and it's a module so the default GENERIC kernel won't be
GPL tainted.
I'll have to add an actual HAL layer that allows the GPL PHY to be loaded
before if_bwn so it can be "magic", but that'll come later.
Adrian Chadd [Tue, 17 May 2016 06:52:53 +0000 (06:52 +0000)]
[bhnd] Finish bhnd(4) PCI/PCIe-G1 hostb support.
Now that we've got access to SPROM and can access board identification,
this implements all known remaining hardware work-arounds for the bhnd(4)
PCI and PCIe-G1 cores operating endpoint mode.
Additionally, this adds an initial set of skeleton PCIe-G2 hostb and pcib
drivers, required by fullmac and newer softmac devices.
PCIe PHY needs different initialization on MT7628/MT7688 SoCs than it does
on MT7620.
However, LEDE (and OpenWRT) dts files have the PCIe node for MT7628/MT7688
as compatible with mt7620-pci.
We already can handle this properly in our driver, so we just need to add
compat strings to fbsd-mt7628an.dtsi and the PCIe driver.
Approved by: adrian (mentor)
Sponsored by: Smartcom - Bulgaria AD
Differential Revision: https://reviews.freebsd.org/D6395
This is an import of the reworked LEDE dts files. Besides other things
they make it easier for us to reuse.
The only diffs left are for the following SoCs:
MT7620A (fbsd-mt7620a.dtsi)
MT7621 (fbsd-mt7621.dtsi)
MT7628 (fbsd-mt7628an.dtsi)
RT3883 (fbsd-rt3883.dtsi)
So we include the fbsd-*.dtsi files at the end of the original LEDE dtsi
files, using '#include "fbsd-xxxx.dtsi"'.
For example, for MT7621, the LEDE dtsi file is mt7621.dtsi. At the end of
it we add:
#include "fbsd-mt7621.dtsi"
Approved by: adrian (mentor)
Obtained from: LEDE project
Sponsored by: Smartcom - Bulgaria AD
Differential Revision: https://reviews.freebsd.org/D6394
Andrew Rybchenko [Tue, 17 May 2016 06:28:03 +0000 (06:28 +0000)]
sfxge(4): only raise an exception after MC assert or reboot in the common code
Fix efx_mcdi_request_poll so it only raises an exception if EIO is
reported from a detected MC assert or reboot. This prevents
an unnecessary exception being raised if an MCDI response error code
is trandlated to EIO.
Submitted by: Andy Moreton <amoreton at solarflare.com>
Sponsored by: Solarflare Communications, Inc.
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D6392
Andrew Rybchenko [Tue, 17 May 2016 06:26:02 +0000 (06:26 +0000)]
sfxge(4): fix Medford timer quantum calculation in common code
The event/timer block used sysclk in Huntington, but has been
moved to the dpcpu clock domain for Medford. Fix the computed
timer quantum to use the right clock.
Submitted by: Andy Moreton <amoreton at solarflare.com>
Sponsored by: Solarflare Communications, Inc.
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D6389
Andrew Rybchenko [Tue, 17 May 2016 06:25:00 +0000 (06:25 +0000)]
sfxge(4): query and use current MTU if setting the MTU fails
This allows the driver to fall back to the largest usable MTU if a
user attempts to configure an unprivileged function with an MTU higher
than that of the attached port.
Submitted by: Mark Spender <mspender at solarflare.com>
Sponsored by: Solarflare Communications, Inc.
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D6387
Adrian Chadd [Tue, 17 May 2016 00:23:46 +0000 (00:23 +0000)]
[bhnd] Centralize fetching of board information
Centralizes fetching of board information (vendor, type, revision, etc),
and adds support for matching quirks against board identification info.
* Adds a BHND_BUS_READ_BOARD_INFO(), allowing bhnd bus/bus parent(s) to
handle implementation-specific fetching of board info.
* Integrates board type constants from the latest Broadcom ISC-licensed
bcmdevs.h included in dd-wrt's Broadcom driver source drops.
* Adds support for matching on chip/board quirks to bhnd_device_quirks()/
bhnd_chip_quirks().
* Use the new board/chip quirk matching to match Apple devices that failed
to set BFL2_PCIEWAR_OVR in SROM.
Adrian Chadd [Mon, 16 May 2016 23:56:31 +0000 (23:56 +0000)]
[mips] Improve MIPS trampoline code
This patch fix trampoline build. inckern.S increases stack and calls _startC
of elf_trampoline, so inckern.S should be called before elf_trampoline. gcc4.2 puts
text (code) into image according to order of source files in this call, so order has changed.
In addition make will install trampoline kernel. It allows to use kernel for firmware build.
Submitted by: Michael Zhilin <mizhka@gmail.com>
Differential Revision: https://reviews.freebsd.org/D6242
Adrian Chadd [Mon, 16 May 2016 23:40:32 +0000 (23:40 +0000)]
[bhnd] Add logging macros to BHND.
There are 5 logging levels:
* ERROR
* WARN
* INFO
* DEBUG
* TRACE
There are 2 logging context:
* with
* without device
DEBUG and TRACE records are printed only if bootverbose.
Logging records are printed with source code line information if acceptable
logging level is DEBUG or TRACE.
Submitted by: Michael Zhilin <mizhka@gmail.com>
Differential Revision: https://reviews.freebsd.org/D6247
Don Lewis [Mon, 16 May 2016 23:29:04 +0000 (23:29 +0000)]
Set ai2 to NULL in in find_host() before the loop and after calling
freeaddrinfo() on it to indicate that it doesn't point to a valid
addrinfo list. This fixes this Coverity issues: 1006368 Uninitialized pointer read 1018506 Double free 1305590 Resource leak
that can be triggered in the hp->hostname[0] != '\0' case.
Don't treat a character as a boolean.
Fix these Coverity issues: 1009293 Unchecked return value from library 1194246 Wrong size argument
by tweaking the status file extend code.
Don Lewis [Mon, 16 May 2016 23:00:48 +0000 (23:00 +0000)]
Actually use the loop interation limit so carefully computed on the
previous line to prevent buffer overflow. This turns out to not be
important because the upstream xdr code already capped the object
size at the proper value. Using the correct limit here looks a lot
less scary and should please Coverity.
John Baldwin [Mon, 16 May 2016 21:33:31 +0000 (21:33 +0000)]
Use polling spin loops for timeouts during early boot.
Some ACPI operations such as mutex acquires and event waits accept a
timeout. The ACPI OSD layer implements these timeouts by using regular
sleep timeouts. However, this doesn't work during early boot before
event timers are setup. Instead, use polling combined with DELAY()
to spin.
This fixes booting on upcoming Intel systems with Kaby Lake processors.
Adrian Chadd [Mon, 16 May 2016 20:26:30 +0000 (20:26 +0000)]
[bwn] check DUALPHY; add BCM4322 to the don't-override list.
* DUALPHY in TGSHIGH tells us there's a phy that is dualband, rather than
two separate PHYs/MACs (which we almost but don't quite yet support.)
Use it.
* Add the BCM4322 PCI ID to the list of devices we don't override.
This means the 2g/5g flags are preserved, and thus we get 5GHz
operation (with N-PHY, of course.)
Tested:
* BCM4311, STA mode (11bg)
* BCM4312, STA mode (11bg)
* BCM4321, STA mode (11abg)
Adrian Chadd [Mon, 16 May 2016 20:22:52 +0000 (20:22 +0000)]
[siba] fix up debugging.
* unbreak non-debug builds - don't default to debugging SCAN; that was
left-over from my testing.
* include opt_siba.h, now that it's generated as appropriate.
* stick the debug enum outside the debug block, just so it's there for
any code that wants to set siba_debug for some reason (like say,
my debugging muckup.)
* make DPRINTF() use __VA_ARGS__ for formatting too, so it correctly
handles printing w/ no args.
* Make DPRINTF() use device_printf().
Pedro F. Giffuni [Mon, 16 May 2016 20:04:38 +0000 (20:04 +0000)]
dev/ow: Tag an unreachable switch default.
Coverity reports an uninitialized "dir" in case the switch defaults
without hitting any case. Respect the original intent and quell the
false positive with the relatively new __unreachable() builtin.
Pedro F. Giffuni [Mon, 16 May 2016 20:00:09 +0000 (20:00 +0000)]
libefi: Tag an unreachable switch default.
Coverity reports an uninitialized "len" in case the switch defaults
without hitting any case. Respect the original intent and quell the
false positive with the relatively new __unreachable() builtin.
Andriy Gapon [Mon, 16 May 2016 15:48:56 +0000 (15:48 +0000)]
do not destroy 'snapdir' when it becomes inactive
That was just wrong. In fact, we can safely keep this static entry when
it's inactive.
Now the destructive action is moved to the reclaim method and the
function is renamed from zfsctl_snapdir_inactive(0 to
zfsctl_snapdir_reclaim().
Also, we can use gfs_vop_reclaim() instead of gfs_dir_inactive() +
kmem_free().
Lastly, we can just assert that the node does not any children when it
is reclaimed, even on the force unmount. That's because zfs_umount()
does an extra vflush() pass which should destroy all snapshot-mountpoint
vnodes that are the snapdir's children.
Andriy Gapon [Mon, 16 May 2016 15:37:41 +0000 (15:37 +0000)]
try to recycle "snap" vnodes as soon as possible
Those vnodes should not linger. "Stale" nodes may get out of
synchronization with actual snapshots. For example if we destroy a
snapshot and create a new one with the same name. Or when we rename a
snapshot.
While there fix the argument type for zfsctl_snapshot_reclaim().
Also, its original argument can be passed to gfs_vop_reclaim() directly.
Bug 209093 could be related although I have not specifically verified
that. Referencing just in case.
Andriy Gapon [Mon, 16 May 2016 15:28:39 +0000 (15:28 +0000)]
fix locking in zfsctl_root_lookup
Dropping the root vnode's lock after VFS_ROOT() didn't really help the
fact that we acquired the lock while holding its child's, .zfs, lock
while performing the operaiton.
So, directly use zfs_zget() to get the root vnode.
While there simplify the code in zfsctl_freebsd_root_lookup.
We know that .zfs is always exclusively locked.
We know that there is already a reference on *vpp, so no need for an
extra one.
Account for the fact that .. lookup may ask for a different lock type,
not necessarily LK_EXCLUSIVE. And handle a possible failure to acquire
the lock given the lock flags.
Andriy Gapon [Mon, 16 May 2016 15:13:16 +0000 (15:13 +0000)]
gfs_lookup_dot() does not have to acquire any locks
In fact, that was dangerous. For example, zfsctl_snapshot_reclaim()
calls gfs_dir_lookup() on ".." path and that ends up calling
gfs_lookup_dot() which violated locking order by acquiring the parent's
directory vnode lock after the child's vnode lock.
Also, the previous behavior was inconsistent as gfs_dir_lookup()
returned a locked vnode for . and .. lookups, but not for any other.
Now gfs_lookup_dot() just references a resulting vnode and the locking
is done in its consumers, where necessary.
Note that we do not enable shared locking support for any gfs / zfsctl
vnodes.
Andriy Gapon [Mon, 16 May 2016 15:03:52 +0000 (15:03 +0000)]
avoid deadlock between zfsctl_snapdir_lookup and zfsctl_snapshot_reclaim
The former acquired a snap vnode lock while holding sd_lock while the
latter does the opposite.
The solution is drop sd_lock before acquiring the vnode lock. That
should be okay as we are still holding a lock on the 'snapshot'
directory in the exclusive mode. That lock ensures that there are no
concurrent lookups in the directory and thus no concurrent mount attempts.
But now we have to account for the possibility that the snap vnode
might get reclaim after we drop sd_lock and before we can get
the node lock. So, check for that case and retry.
Andrew Turner [Mon, 16 May 2016 14:07:43 +0000 (14:07 +0000)]
Add intrng support to the GICv3 driver. It lacks ITS support so won't handle
MSI or MSI-X interrupts, however this is enought to boot FreeBSD under the
ARM Foundation Model with a GICv3 interrupt controller.
Approved by: ABT Systems Ltd
Relnotes: yes
Sponsored by: The FreeBSD Foundation
Jilles Tjoelker [Mon, 16 May 2016 12:56:28 +0000 (12:56 +0000)]
install: Revert utimensat usage (r299850).
This should fix the build on older stable/10, since install is a bootstrap
tool.
Pending a decision how to fix this properly, revert utimensat usage. Copies
with the -p option will again appear older than the original almost always,
but -p is not commonly used.
Andriy Gapon [Mon, 16 May 2016 12:15:19 +0000 (12:15 +0000)]
fix a vnode reference leak caused by illumos compat traverse()
This commit partially reverts r273641 which introduced the leak.
It did so to accomodate for some consumers of traverse() that expected
the starting vnode to stay as-is. But that introduced the leak in the
case when a mounted filesystem was found and its root vnode was
returned.
r299914 removed the troublesome consumers and now there is no reason to
keep the starting vnode. So, now the new rules are:
- if there is no mounted filesystem, then nothing is changed
- otherwise the starting vnode is always released
- the root vnode of the mounted filesystem is returned locked and
referenced in the case of success
Andrew Turner [Mon, 16 May 2016 09:31:44 +0000 (09:31 +0000)]
Add a pcib interface for use by interrupt controllers that need to
translate the pci rid to a controller ID. The translation could be based
on the 'msi-map' OFW property, a similar ACPI option, or hard-coded for
hardware lacking the above options.
Reviewed by: wma
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
Andrew Turner [Mon, 16 May 2016 09:15:50 +0000 (09:15 +0000)]
Re-commit r299467 having fixed the build:
Add a new get_id interface to pci and pcib. This will allow us to both
detect failures, and get different PCI IDs.
For the former the interface returns an int to signal an error. The ID is
returned at a uintptr_t * argument.
For the latter there is a type argument that allows selecting the ID type.
This only specifies a single type, however a MSI type will be added
to handle the need to find the ID the hardware passes to the ARM GICv3
interrupt controller.
A follow up commit will be made to remove pci_get_rid.
Reviewed by: jhb, rstone (previous version)
Obtained from: ABT Systems Ltd
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D6239
Andrew Turner [Mon, 16 May 2016 09:11:40 +0000 (09:11 +0000)]
Introduce MSI and MSI-X support to intrng. This adds a new msi device
interface with 5 methods to mirror the 5 MSI/MSI-X methods in the pcib
interface. The pcib driver will need to perform a device specific lookup
to find the MSI controller and pass this to intrng as the xref. Intrng
will finally find the controller and have it handle the requested operation.
Obtained from: ABT Systems Ltd
MFH: yes
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D5985
Don Lewis [Mon, 16 May 2016 08:34:17 +0000 (08:34 +0000)]
Hoist the getpwnam() call outside the first if/else block in
pam_sm_chauthtok(). Set user = getlogin() inside the true
branch so that it is initialized for the following PAM_LOG()
call. This is how it is done in pam_sm_authenticate().