Re: iwm: support dynamic queue allocation (DQA)

2019-10-15 Thread Jan Klemkow
On Mon, Oct 14, 2019 at 01:51:02PM +0200, Stefan Sperling wrote: > I have successfully tested this diff on 8265 with our current firmware > image (22.361476.0) as well as a newer -22 firmware image (22.391740.0, > which is *not* in fw_update yet). I have also tested 7265 successfully. Also works f

Re: net80211: increase background scan backoff timer

2019-10-15 Thread Jan Klemkow
On Mon, Oct 14, 2019 at 02:22:24PM +0200, Stefan Sperling wrote: > Is anyone running this already? OK? I run the diff without any noticeable problems. Run this with iwm and wpa-enterprise. iwm0 at pci1 dev 0 function 0 "Intel Dual Band Wireless-AC 8265" rev 0x78, msi Seems to work for me. bye,

Re: iwm: support dynamic queue allocation (DQA)

2019-10-15 Thread Kevin Lo
On Mon, Oct 14, 2019 at 01:51:02PM +0200, Stefan Sperling wrote: > > Newer iwm firmware requires the driver to use a feature known as > dynamic queue allocation (DQA). What matters is that the command queue > index was changed. Newer firmware images have stopped responding to > commands sent with

sysupgrade(8) download directory

2019-10-15 Thread Joe Davis
Prompted by a discussion on irc. While /home/_sysupgrade is a sensible default, it's not always possible to use (for instance if /home is encrypted and mounted via rc.local or even manually over ssh). Previous attempts have had sysupgrade(8) automagically figure out a sensible location based on a

Re: random safety for pbkdf

2019-10-15 Thread Theo de Raadt
Makes sense -- but perhaps justify the arc4random with a comment, explaining what is being done, so that people don't need to look in the commitlog? Ted Unangst wrote: > In the event that a program uses invalid parameters, I think we should > overwrite the key with random data. Otherwise, there'

random safety for pbkdf

2019-10-15 Thread Ted Unangst
In the event that a program uses invalid parameters, I think we should overwrite the key with random data. Otherwise, there's a chance the program will continue with a zero key. It may even appear to work, encrypting and decrypting data, but with a weak key. Random data means it fails closed, and s

Re: iwm: support dynamic queue allocation (DQA)

2019-10-15 Thread Florian Obser
seems to be working on Intel Dual Band Wireless AC 7260 -- I'm not entirely sure you are real.

kill scheduler_fork_hook()

2019-10-15 Thread Martin Pieuchot
The only raison d'etre of scheduler_fork_hook() is because `p_estcpu' is not at the right place in "struct proc". Diff below fixes that. I reordered the fields to not have implicit padding increase the size of the structure. ok? Index: sys/proc.h

tun(4) & switch(4) -> tsleep_nsec(9)

2019-10-15 Thread Martin Pieuchot
Convert the two pseudo-drivers to use tsleep_nsec(9) in their read(2) routine, both drivers block indefinitely. ok? Index: net/if_tun.c === RCS file: /cvs/src/sys/net/if_tun.c,v retrieving revision 1.190 diff -u -p -r1.190 if_tun.c -

tsleep_nsec(9) for revarp

2019-10-15 Thread Martin Pieuchot
The hz/2 idiom means 500ms, so use that. ok? Index: netinet/if_ether.c === RCS file: /cvs/src/sys/netinet/if_ether.c,v retrieving revision 1.240 diff -u -p -r1.240 if_ether.c --- netinet/if_ether.c 17 Jul 2019 16:46:18 - 1.

Re: acpithinkpad: don't take over ws_[gs]et_param on version 2 devices

2019-10-15 Thread Theo Buehler
On Mon, Oct 14, 2019 at 11:00:57PM +0200, Mark Kettenis wrote: > > Date: Sun, 13 Oct 2019 21:37:53 -0500 > > From: joshua stein > > > > Newer ThinkPads have ACPI goo to allow acpivout to control screen > > backlight, so don't take over ws_[gs]et_param from it. This allows > > for 100 levels of