7d4dceec erj Feb. 4, 2021, 11:33 p.m.
X700 family of controllers has limited number of available VLAN
HW filters. Driver did not handle properly a case when user
assigned more VLANs to the interface which had all filters
already in use. Fix that by disabling HW filtering when
it is impossible to create filters for all requested VLANs.
Keep track of registered VLANs using bitstring to be able
to re-enable HW filtering when number of requested VLANs
drops below the limit.

Also switch all allocations to use M_IXL malloc type
to ease detecting memory leaks in the driver.

Reviewed by:	erj
Tested by:	gowtham.kumar.ks@intel.com
MFC after:	1 week
Differential Revision:	https://reviews.freebsd.org/D28137
cgit
2bd4ff2d tsoome Feb. 4, 2021, 9:29 p.m.
Conout does contian the default output device name.
ConOutDev does contain all possible output device names, so we can
use it as fallback, when there is no ConOut.

PR: 253253
cgit
aa776623 bofh Feb. 4, 2021, 9:05 p.m.
Although if RJ-45 interface is not being used the miibus is not required
but miibus is a build time dependency.

Reviewed by:    imp, manu, rajesh1.kumar@amd.com
Approved by:    imp, manu, rajesh1.kumar@amd.com
Differential Revision:  https://reviews.freebsd.org/D28465
cgit
8a06ca2f rstone Feb. 4, 2021, 6:58 p.m.
When OFED was upgraded to Linux v4.9, a bunch of Linux-specific
netlink changes were dropped.  Unfortunately, there was a mismerge
in this process and as a result ib_sa_cancel_query() would fail to
cancel an outstanding MAD.

This was causing rdma_destroy_id() to hang indefinitely waiting
for the MAD to complete and release the final reference.

Sponsored by: Dell Inc.
Differential Revision:	https://reviews.freebsd.org/D28421
Reviewed by: hselasky, kib
MFC after: 2 months
cgit
b58cf1cb rstone Feb. 4, 2021, 6:54 p.m.
Consider the following scenario:

1. A delayed_work struct in the WORK_ST_TIMER state.
2. Thread A calls mod_delayed_work()
3. Thread B (a callout thread) simultaneously calls
linux_delayed_work_timer_fn()

The following sequence of events is possible:

A: Call linux_cancel_delayed_work()
A: Change state from TIMER TO CANCEL
B: Change state from CANCEL to TASK
B: taskqueue_enqueue() the task
A: taskqueue_cancel() the task
A: Call linux_queue_delayed_work_on().  This is a no-op because the
state is WORK_ST_TASK.

As a result, the delayed_work struct will never be invoked.  This is
causing address resolution in ib_addr.c to stop permanently, as it
never tries to reschedule a task that it thinks is already scheduled.

Fix this by introducing locking into the cancel path (which
corresponds with the lock held while the callout runs).  This will
prevent the callout from changing the state of the task until the
cancel is complete, preventing the race.

Differential Revision:	https://reviews.freebsd.org/D28420
Reviewed by: hselasky
MFC after: 2 months
cgit
1eec5861 arichardson Feb. 4, 2021, 5:57 p.m.
It appears this test no longer fails after c203bd70b5957f85616424b6fa374479372d06e3.

PR:		215690
cgit
72692dfd arichardson Feb. 4, 2021, 5:56 p.m.
These tests create jails with the same name, so they cannot be run in
parallel.

Reviewed By:	lwhsu
Differential Revision: https://reviews.freebsd.org/D28482
cgit
58de61b9 arichardson Feb. 4, 2021, 5:56 p.m.
The output now contains http-alt instead of 8080 and personal-agent
instead of 5555.
This was probably caused by 228e2087a32847fa51168f3f0c58f931cb2cb0f8.

Reviewed By:	kp
Differential Revision: https://reviews.freebsd.org/D28481
cgit
c203bd70 arichardson Feb. 4, 2021, 3:03 p.m.
This includes improvements to the atf-sh helper functions that
significantly reduce the number of spawned processes for each test
and therefore speeds up running the testsuite noticeably.
cgit
cb7cc72c jhb Feb. 3, 2021, 10:59 p.m.
OpenSSL BIO classes provide an abstraction for dealing with I/O.
OpenSSL provides BIO classes for commonly used I/O primitives backed
by file descriptors, sockets, etc. as well as permitting consumers
of OpenSSL to define custom BIO classes.

One of the methods BIO classes implement is a control method invoked
by BIO_ctrl() for various ancilliary tasks somewhat analgous to
fcntl() and ioctl() on file descriptors.  According to the BIO_ctrl(3)
manual page, control methods should return 0 for unknown control
requests.

KTLS support in OpenSSL adds new control requests.  Two of those new
requests are queries to determine if KTLS is enabled for either
reading or writing.  These control reuquest return 1 if KTLS is
enabled and 0 if it is not.

serf includes two custom BIO classes for wrapping I/O requests from
files and from a buffer in memory.  These BIO classes both use a
custom control method.  However, this custom control method was
returning 1 for unknown or unsupported control requests instead of 0.
As a result, OpenSSL with KTLS believed that these BIOs were using
KTLS and were thus adding headers and doing encryption/decryption in
the BIO.  Correcting the return value removes this confusion.

PR:		253135
Reported by:	Guido Falsi <mad@madpilot.net>
Reviewed by:	emaste
MFC after:	3 days
Sponsored by:	Netflix
Differential Revision:	https://reviews.freebsd.org/D28472
cgit
9fac05ba alfredo Feb. 4, 2021, 1:06 a.m.
Submitted by:   Andre Fernando da Silva <andre.silva@eldorado.org.br>
Reviewed by:    luporl, alfredo, kadesai (on email)
Sponsored by:   Eldorado Research Institute (eldorado.org.br)
Differential Revision:  https://reviews.freebsd.org/D26531
cgit
e34a057c alfredo Feb. 4, 2021, 1:06 a.m.
Add endiannes conversions in order to support big-endian platforms

Submitted by:	Andre Fernando da Silva <andre.silva@eldorado.org.br>
Reviewed by:	luporl, alfredo, kadesai (on email)
Sponsored by:	Eldorado Research Institute (eldorado.org.br)
Differential Revision:	https://reviews.freebsd.org/D26531
cgit
847dfd28 emaste Feb. 3, 2021, 9:24 p.m.
PR:		246015
Reviewed by:	cem
Sponsored by:	The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D28461
cgit
e79b51e2 emaste Feb. 3, 2021, 9:17 p.m.
MFC after:	1 week
Sponsored by:	The FreeBSD Foundation
cgit
660344ca rstone Feb. 3, 2021, 9:16 p.m.
If a M_WAITOK contig alloc fails, the VM subsystem will try to
reclaim contiguous memory twice before actually failing the
request.  On a system with 64GB of RAM I've observed this take
400-500ms before it finally gives up, and I believe that this
will only be worse on systems with even more memory.

In certain contexts this delay is extremely harmful, so add a flag
that will skip reclaim for allocation requests to allow those
paths to opt-out of doing an expensive reclaim.

Sponsored by: Dell Inc
Differential Revision:	https://reviews.freebsd.org/D28422
Reviewed by: markj, kib
cgit