route(8): tableid is always valid

tableid is either initialized to the current routing table or to the one specified by the T flag. No need to pass a flag around if the tableid is valid. It always is. This code is now in sync between flushroutes() and p_rttables(). OK? diff --git route.c route.c index

fix route monitor -AF

in the pledge refactoring in 2015 we lost the ability to filter on address family when running route monitor. i.e. route monitor -inet6 would only show IPv6 route messages. This restores that functionality. To see that I'm not globbering some global state I (more or less) machanically turned int

Re: Broken links in fc-*man pages (fontconfig)

On Sun, Apr 29, 2018 at 10:07:29PM +0300, IL Ka wrote: > fc-list(1) and friends man pages > > Have link to > /usr/share/doc/fontconfig/fontconfig-user.html > > Which must be > /usr/X11R6/share/doc/fontconfig/fontconfig-user.html > > Apps that are part of Xorg (like xterm) have special > macros

route(8): let the kernel do the filtering for flushroutes

No need to get the whole routing table from the kernel if we are filtering by address family and / or priority; similar to what p_rttables() is doing. (On a router in the DFZ we need to copy 150+ MB) OK? diff --git route.c route.c index 031ae6c1b33..229cca1491d 100644 --- route.c +++ route.c

Re: [PATCH] mv -P

"Theo de Raadt" wrote: > attila wrote: > > Hi tech@, > > > > I frequently want to move a file from one place to another and shred > > the original via the rm(1) -P option. On several occasions I have > > forgotten about wanting to shred the original

Broken links in fc-*man pages (fontconfig)

fc-list(1) and friends man pages Have link to /usr/share/doc/fontconfig/fontconfig-user.html Which must be /usr/X11R6/share/doc/fontconfig/fontconfig-user.html Apps that are part of Xorg (like xterm) have special macros in their man pages, while fontconfig mans are generated from sgml and seems

Re: fix route monitor -AF

Florian Obser(flor...@openbsd.org) on 2018.04.29 17:10:46 +0200: > in the pledge refactoring in 2015 we lost the ability to filter on > address family when running route monitor. > i.e. route monitor -inet6 would only show IPv6 route messages. > > This restores that functionality. To see that I'm

Re: fix route monitor -AF

On Sun, Apr 29, 2018 at 09:12:36PM +0200, Sebastian Benoit wrote: > Florian Obser(flor...@openbsd.org) on 2018.04.29 17:10:46 +0200: > > in the pledge refactoring in 2015 we lost the ability to filter on > > address family when running route monitor. > > i.e. route monitor -inet6 would only show

Re: avoid uninitialised attr in rasops_scrollback()

On Sun, Apr 29, 2018 at 09:42:00AM +, Miod Vallat wrote: > > > Don't use attr uninitialised. Avoids glitches seen when using > > scrollback with radeondrm. > > That's horrible. The bg attribute should be passed to the function, > rather than computed here with possibly wrong values. > You

Re: [PATCH] mv -P

>"Theo de Raadt" wrote: >> attila wrote: >> > Hi tech@, >> > >> > I frequently want to move a file from one place to another and shred >> > the original via the rm(1) -P option. On several occasions I have >> > forgotten about wanting to shred the

Re: Broken links in fc-*man pages (fontconfig)

> Patches welcome. I don't have time for this currently. Here is patch, It installs new target as maninstall dependency. It patches man pages right in the place using sed. Hope it helps:) diff --git lib/fontconfig/Makefile.inc lib/fontconfig/Makefile.inc index 3075d00d..a5701565 100644 ---

Re: hashinit(): power of 2 fast path

Disregard that, brainfart on my side. Mathieu - wrote: > Anton Lindqvist wrote: > > Hi, > > If the elements argument passed to hashinit() is a power of 2 there's no > > need to find the closest power of 2 that can fit all elements since > > elements == hashsize will always be true. During boot of

Re: [patch] Skip background scan if bssid is set

> On 29 Apr 2018, at 11:43, Stuart Henderson wrote: > >> On 2018/04/29 10:17, Stefan Sperling wrote: >>> On Sun, Apr 29, 2018 at 03:39:07AM +0200, Jesper Wallin wrote: >>> Hi all, >>> >>> I recently learned that my AP behaves badly and I have packet loss when >>> the

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 01:24:09PM +0200, Stefan Sperling wrote: > This diff doesn't look right. You'll want to prevent the timeout_add_msec() > call rather than timeout_del() since your goal is to avoid scheduling > the timeout in the first place. See 'man timeout_add'. Oh, of course. Thanks

Re: [patch] Skip background scan if bssid is set

On 2018/04/29 10:17, Stefan Sperling wrote: > On Sun, Apr 29, 2018 at 03:39:07AM +0200, Jesper Wallin wrote: > > Hi all, > > > > I recently learned that my AP behaves badly and I have packet loss when > > the background scan is running. I had a small chat with stsp@ about it, > > asking if there

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 03:39:07AM +0200, Jesper Wallin wrote: > Hi all, > > I recently learned that my AP behaves badly and I have packet loss when > the background scan is running. I had a small chat with stsp@ about it, > asking if there is a way to disable it. He kindly explained that if

Re: Looking for testers for em(4) quirks patch

On Monday, 2 April 2018 11:52:08 CEST Stefan Fritsch wrote: > It would be nice if people could give it a try on various em(4) hardware > to check if there are any regressions. Somewhat belatedly, this is now committed. Thanks to all the people who tested it. Cheers, Stefan

Re: hashinit(): power of 2 fast path

On Sun, Apr 29, 2018 at 10:52:27AM +0200, Mathieu - wrote: > Disregard that, brainfart on my side. > > Mathieu - wrote: > > Anton Lindqvist wrote: > > > Hi, > > > If the elements argument passed to hashinit() is a power of 2 there's no > > > need to find the closest power of 2 that can fit all

Re: avoid uninitialised attr in rasops_scrollback()

> Don't use attr uninitialised. Avoids glitches seen when using > scrollback with radeondrm. That's horrible. The bg attribute should be passed to the function, rather than computed here with possibly wrong values.

rssi comparison threshold

This diff tries to avoid situations where background scans play ping-pong between different APs with nearly equal RSSI, as observed by phessler. Not all drivers represent RSSI values in dBm or percentage, so the diff includes the possibility for drivers to override the new RSSI comparison

avoid uninitialised attr in rasops_scrollback()

Don't use attr uninitialised. Avoids glitches seen when using scrollback with radeondrm. Index: rasops.c === RCS file: /cvs/src/sys/dev/rasops/rasops.c,v retrieving revision 1.53 diff -u -p -r1.53 rasops.c --- rasops.c27 Apr

Re: [patch] Skip background scan if bssid is set

On 2018/04/29 11:48, Mischa Peters wrote: > > On 29 Apr 2018, at 11:43, Stuart Henderson wrote: > > I'm wondering if changing this may introduce problems when an AP moves > > to a different channel? Either by manual configuration, mechanisms > > like Ruckus' channelfly

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 11:48:47AM +0200, Mischa Peters wrote: > How does this play with roaming protocols on “enterprise” > WiFi equipment, like 802.11k and 802.11v? Those protocols are irrelevant to the problem under discussion because OpenBSD does not support 11k/v yet. OpenBSD clients appear

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 01:47:29PM +0200, Jesper Wallin wrote: > On Sun, Apr 29, 2018 at 01:24:09PM +0200, Stefan Sperling wrote: > > This diff doesn't look right. You'll want to prevent the timeout_add_msec() > > call rather than timeout_del() since your goal is to avoid scheduling > > the

Re: avoid uninitialised attr in rasops_scrollback()

> Date: Sun, 29 Apr 2018 16:46:30 +1000 > From: Jonathan Gray > > Don't use attr uninitialised. Avoids glitches seen when using > scrollback with radeondrm. ok kettenis@ > Index: rasops.c > === > RCS file:

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 10:17:43AM +0200, Stefan Sperling wrote: > I agree, even though it would be nice to understand the underlying > packet loss issue. But I cannot reproduce the problem unforunately :( > Have you verified that the problem only happens on this particular AP? Unfortunately no,

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 10:43:55AM +0100, Stuart Henderson wrote: > It's very common for wifi clients to do background scans so I'd be > interested to know whether non-OpenBSD clients also see packet loss, > or whether OpenBSD with a different client device is any better. What > are the AP and

Re: [patch] Skip background scan if bssid is set

On Sun, Apr 29, 2018 at 12:13:37PM +0200, Jesper Wallin wrote: > I've moved the check to ieee80211_input() now and it seems to work. > However, when I set the bssid and it aborts the scan, the debug message > of "end background scan" appears before the actual list of APs. This diff doesn't look