r323627 tuexen Sept. 15, 2017, 8:41 p.m.
r323626 tuexen Sept. 15, 2017, 8:34 p.m.
Sponsored by:	Netflix, Inc.
r323625 imp Sept. 15, 2017, 8:16 p.m.
Don't call cam_iosched_trim_done or cam_iosched_submit_trim for nda
since its hardware can handle almost an arbitrary number of TRIMs and
we don't have to be careful to only ever do one.

Sponsored by: Netflix
r323624 imp Sept. 15, 2017, 8:15 p.m.
It's intended only for those situations where the periph driver
ones to limit the number of trims active to one and only one.
Also update comments on associated functions.

Sponsored by: Netflix
r323623 emaste Sept. 15, 2017, 8:12 p.m.
Reviewed by:	allanjude
MFC after:	1 week
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D12339
r323622 emaste Sept. 15, 2017, 8:05 p.m.
After r308212 Capsicum permits .. lookups in capability mode, as long as
path component traversal does not escape the directory corresponding to
the provided file descriptor.

We should add a description of the vfs.lookup_cap_dotdot and
vfs.lookup_cap_dotdot_nonlocal sysctls, perhaps as a cross-reference to
capsicum(4). I intend to look at that soon.

Reviewed by:	bjk, cem, kib
MFC after:	1 week
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D12343
r323621 landonf Sept. 15, 2017, 7:56 p.m.
Approved by:	adrian (mentor)
Differential Revision:	https://reviews.freebsd.org/D12300
r323620 bdrewery Sept. 15, 2017, 7:48 p.m.
- The exit probe was not appropriately filtered to only the known pid so it
  was firing on any random process that would exit rather the only the one
  we cared about.
- The dtest script executes the tst.raise*.exe in the background from
  POSIX sh without jobs control.  POSIX mandates that SIGINT be set to
  SIG_IGN in this case.  The test executable never actually tested that
  SIGINT could be caught despite trying to block and delay the signal.
  So the SIGINT sent from raise() is never actually received since it
  is ignored.  This could be fixed by calling 'trap - INT' from dtest
  before running the executable but I've opted to just use SIGUSR1
  instead in these specific tests rather than adding more logic to
  test that SIGINT is not ignored at startup.

These 2 issues meant that the tests would randomly work but only if a process
coincidentally exited during the test.

Reviewed by:	markj
MFC after:	2 weeks
Sponsored by:	Dell EMC Isilon
r323619 kibab Sept. 15, 2017, 7:47 p.m.
* Demote the level of several debug messages to CAM_DEBUG_TRACE
 * Add detection for SDHC cards that can do 1.8V. No voltage switch sequence
   is issued yet;
 * Don't create a separate LUN for each SDIO function. We need just one to make
   pass(4) attach;
 * Remove obsolete mmc_sdio* files. SDIO functionality will be moved into the
   separate device that will manage a new sdio(4) bus;
 * Terminate probing if got no reply to CMD0;
 * Make bcm2835 SDHCI host controller driver compile with 'option MMCCAM'.

Approved by:	imp (mentor)
Differential Revision:	https://reviews.freebsd.org/D12109
r323616 kib Sept. 15, 2017, 4:07 p.m.
free queue mutex lock owning session, same as it was done for the
object termination in r323561.

Reported and tested by:	mjg
Reviewed by:	alc, markj
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
r323615 kevans Sept. 15, 2017, 3:57 p.m.
Start the phasing out of TRE by disabling it by default. r317254 introduced
a BSD_GREP_FASTMATCH knob (defaulting to on) for testing of bsdgrep with and
without TRE enabled. More bugs have cropped up since then, and
WITHOUT_BSD_GREP_FASTMATCH has shown in testing to be more stable than its

Approved by:	emaste (mentor)
Differential Revision:	https://reviews.freebsd.org/D12381
r323614 scottl Sept. 15, 2017, 3:41 p.m.
Sponsored by:	Netflix
r323613 markj Sept. 15, 2017, 2:59 p.m.
MFC after:	1 week
r323612 avg Sept. 15, 2017, 1:57 p.m.
In theory, all data access errors mean that a member is out of sync
at most.  But they were treated as more serious errors to avoid the
situation where a flaky disk gets repeatedly disconnected, re-synchronized,
reconnected and then disconnected again.

ENXIO is a special error that means that the member disk disappeared,
so it should get the same handling as the GEOM orphaning event.
There is a better chance that when the disk is reconnected, it will be
a good member again.

When ENXIO happens on a read we use the exisiting G_MIRROR_BUMP_SYNCID
mechanism which means that the mirror's syncid is increased as soon
as there is a write to the mirror.  That's because no data has got out
of sync yet, but the problematic memeber is disconnected, so the future
write will make it stale.

When ENXIO happens on a write we use a new G_MIRROR_BUMP_SYNCID_NOW
mechanism which means that we update the mirror metadata as soon as
possible because the problematic memeber is already behind.

Reviewed by:	markj, imp
MFC after:	3 weeks
Differential Revision: https://reviews.freebsd.org/D9463
r323611 emaste Sept. 15, 2017, 1:34 p.m.
Reviewed by:	kevans
MFC after:	1 week
Sponsored by:	The FreeBSD Foundation
Differential Revision:	https://reviews.freebsd.org/D12375