mlx4_QP_FLOW_STEERING_DETACH_wrapper first removes the steering
rule (which results in freeing the rule structure), and then
references a field in this struct (the qp number) when releasing the
busy-status on the rule's qp.
Since this memory was freed, it could reallocated and changed.
Therefore, the qp number in the struct may be incorrect,
so that we are releasing the incorrect qp. This leaves the rule's qp
in the busy state (and could possibly release an incorrect qp as well).
Fix this by saving the qp number in a local variable, for use after
removing the steering rule.
Linux commit:
3b01fe7f91c8e4f9afc4fae3c5af72c14958d2d8
PR: 264469
Sponsored by: NVIDIA Networking
(cherry picked from commit dd2a8c8f72)
mlx4_delete_all_resources_for_slave() in the resource tracker should free
all memory allocated for a slave. While releasing memory of fs_rule,
it misses releasing memory of fs_rule->mirr_mbox.
Linux commit:
461d5f1b59490ce0096dfda45e10038c122a7892
PR: 264249
Sponsored by: NVIDIA Networking
(cherry picked from commit f29c9901a4)
In function mlx4_opreq_action(), pointer "mailbox" is not released,
when mlx4_cmd_box() return and error, causing a memory leak bug.
Fix this issue by going to "out" label, mlx4_free_cmd_mailbox() can
free this pointer.
Linux commit:
febfd9d3c7f74063e8e630b15413ca91b567f963
PR: 264056
Sponsored by: NVIDIA Networking
(cherry picked from commit 527762b2f7)
External events can cause the backlight level to change (AC adapter
plug/unplug for example) so cache the value there too.
PR: 257796
Sponsored by: Beckhoff Automation GmbH & Co. KG
MFC after: 1 week
(cherry picked from commit e26ef41f79)
Apply 6987c47569 to igc. This is not
expected to have any benefit on current parts with current observed PBA
sizes but will avoid surprises if they are increased in future chips.
Approved by: grehan
Differential Revision: https://reviews.freebsd.org/D35217
(cherry picked from commit 9b88ecd674)
Use a sx lock to permit uiomove directly into/out of the the per-port
buffers. In addition, the sx lock provides a stronger guarantee that
I think this driver wants which is to single-thread read and write
calls even while paused. Finally, replace tsleep with dummy wait
channels with calls to pause_sig to more clearly communicate the
intent.
Differential Revision: https://reviews.freebsd.org/D35081
(cherry picked from commit 51f481235c)
0 should only be returned from probe in legacy drivers that need to
preserve softc state between probe and attach.
Differential Revision: https://reviews.freebsd.org/D35080
(cherry picked from commit 4942e8df47)
The port number is still stored in the unit (si_drv0) but is the
entire unit value now.
While here, removed checks for NULL softc since those can never happen
from cdevsw routines. This also resulted in the close method becoming
a no-op, so it has been removed.
Differential Revision: https://reviews.freebsd.org/D35078
(cherry picked from commit 0db65574bb)
Rather than a detour via the devclass and hardcoding unit 0.
While here, remove a check for sc being NULL. It will never be NULL
when attach is called.
Reviewed by: imp
Differential Revision: https://reviews.freebsd.org/D35010
(cherry picked from commit 1ac10fa429)
Rather than fetching the softc using the controller's unit number as
an index into the devclass.
Reviewed by: imp
Differential Revision: https://reviews.freebsd.org/D35004
(cherry picked from commit a9e5e04e64)
Rather than fetching the softc using the controller's unit number as
an index into the devclass.
Reviewed by: imp
Differential Revision: https://reviews.freebsd.org/D34993
(cherry picked from commit 517a8a715a)
- Make descriptions a bit more human-friendly.
- Don't attempt to probe fan safe speeds, this model does not provide
one.
PR: 260781
(cherry picked from commit dc484aed0a)
ISA sound cards (and ISA itself) are long obsolete. NYCBUG dmesgd has
no entries for any of these devices after 2005.
Mail to -current and -stable revealed some nostalgic interest, but
no offers to test or requests to pospone the deprecation.
Add deprecation notices to device attach routines and man pages for:
snd_ad1816 Analog Devices AD1816 SoundPort
snd_ess Ensoniq ESS
snd_gusc Gravis UltraSound (MAX and newer AFAIK)
snd_mss Microsoft Sound System
snd_sbc Creative Sound Blaster
Reviewed by: cy, mav
MFC after: 3 days
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D34604
(cherry picked from commit e39ec8933b)
It simplifies the declaration of the driver structures a little. There
are no current consumers of this macro, in fact it looks like it was
added for exactly this purpose.
This decreases the scope of some variables, so rework the initialization
in vt_init_logos() such that it doesn't require them.
No functional change intended.
MFC after: 1 week
Differential Revision: https://reviews.freebsd.org/D34820
(cherry picked from commit ff1c8af8c3)
As the PolarFire SoC needs SGMII to connect the PHY, check the
'phy-mode' property of device tree node for ethernet and act on it
appropriately.
Add the compatible strings for the PolarFire SoC device tree.
'microchip,mpfs-mss-gem" is not officially documented but has been
observed in the available firmware for this platform, so it is included
for now.
Also, fix a typo in if_cgem_hw.h.
Reviewed by: mhorne
MFC after: 1 week
Sponsored by: Conclusive Engineering
Differential Revision: https://reviews.freebsd.org/D34764
(cherry picked from commit 34f4c17a2b)
Rather than doing these checks based on the detected hardware variant, allow
quirks to be specified as a set of flags for each compatible string.
This simplifies adding support for new compatible hardware.
Reviewed by: mhorne
MFC after: 1 week
Sponsored by: Conclusive Engineering
Differential Revision: https://reviews.freebsd.org/D34764
(cherry picked from commit 66cc0c61b0)
