Re: assertion "_kernel_lock_held()" failed, uipc_socket2.c: ipsec

2017-11-13 Thread Stuart Henderson
On 2017/11/13 08:44, Martin Pieuchot wrote: > On 12/11/17(Sun) 22:10, Stuart Henderson wrote: > > On 2017/11/12 22:48, Martin Pieuchot wrote: > > > On 12/11/17(Sun) 21:30, Stuart Henderson wrote: > > > > iked box, GENERIC.MP + WITNESS, -current as of Friday 10th: > > > > > > Weird, did you tweak

Re: assertion "_kernel_lock_held()" failed, uipc_socket2.c: ipsec

2017-11-13 Thread Martin Pieuchot
On 13/11/17(Mon) 10:03, Stuart Henderson wrote: > On 2017/11/13 08:44, Martin Pieuchot wrote: > > On 12/11/17(Sun) 22:10, Stuart Henderson wrote: > > > On 2017/11/12 22:48, Martin Pieuchot wrote: > > > > On 12/11/17(Sun) 21:30, Stuart Henderson wrote: > > > > > iked box, GENERIC.MP + WITNESS,

Re: un-KERNEL_LOCK() TCP/UDP input & co

2017-11-13 Thread Martin Pieuchot
On 13/11/17(Mon) 12:14, Alexandr Nedvedicky wrote: > your change looks good to me. Though I have a comment/question to your diff. > > > Index: net/if_vxlan.c > > === > > RCS file: /cvs/src/sys/net/if_vxlan.c,v > > retrieving revision

Re: try to bundle multiple packets on an ifq before sending

2017-11-13 Thread David Gwynne
> On 13 Nov 2017, at 5:37 pm, Martin Pieuchot wrote: > > On 13/11/17(Mon) 10:56, David Gwynne wrote: >> On Sun, Nov 12, 2017 at 02:45:05PM +0100, Martin Pieuchot wrote: >>> [...] >>> We're currently using net_tq() to distribute load for incoming packets. >>> So I believe you

Re: un-KERNEL_LOCK() TCP/UDP input & co

2017-11-13 Thread Alexandr Nedvedicky
Hello, your change looks good to me. Though I have a comment/question to your diff. > Index: net/if_vxlan.c > === > RCS file: /cvs/src/sys/net/if_vxlan.c,v > retrieving revision 1.63 > diff -u -p -r1.63 if_vxlan.c > ---

Re: README patch

2017-11-13 Thread Andras Farkas
On Mon, Nov 13, 2017 at 8:16 AM, Sebastian Benoit wrote: > http://ftp.openbsd.org/pub/OpenBSD/6.2/README Much more minor, but "This is our 43nd release." should be "This is our 43rd release." Also, "For further details, please visit https://OpenBSD.org/faq; actually ends up

move reading of ifq counters for if_data into ifq.c

2017-11-13 Thread David Gwynne
this moves the reading of counters out of an interfaces ifqs into ifq.c. there's no semantic change, i just want to keep the knowledge about locking in an ifq all in the same place. ok? Index: if.c === RCS file:

Re: README patch

2017-11-13 Thread Ingo Schwarze
Hi, Andras Farkas wrote on Mon, Nov 13, 2017 at 06:21:26PM -0500: > On Mon, Nov 13, 2017 at 8:16 AM, Sebastian Benoit wrote: >> http://ftp.openbsd.org/pub/OpenBSD/6.2/README > Much more minor, It is now known that that file is outdated in a number of respects. However, it

Re: Remove commented out pledge in tsort

2017-11-13 Thread Ori Bernstein
On Sun, 12 Nov 2017 13:31:10 +0100, Marc Espie wrote: > I did this in an eager way a while back, was told in no uncertain terms > "not yet, we're still looking at things". > > Well, see pledge(2). Whenever the dust settles, sure we can kill this line, > or activate it instead.

Re: Set soname for shared libraries

2017-11-13 Thread Philip Guenther
On Mon, Nov 13, 2017 at 12:25 PM, Mark Kettenis wrote: > The diff below changes bsd.lib.mk such that we set the DT_SONAME entry > for shared libraries to the name of the shared library. This brings > us in line with other ELF systems and allows us to use symlinks to >

Re: multiple interface input queues

2017-11-13 Thread Hrvoje Popovski
On 14.11.2017. 5:42, David Gwynne wrote: > this replaces the single mbuf_queue and task in struct ifnet with > a new ifiqueue structure modelled on ifqueues. i've tested this diff with ix, myx, em and bge with down/up interfaces and everything is working fine ...

multiple interface input queues

2017-11-13 Thread David Gwynne
this replaces the single mbuf_queue and task in struct ifnet with a new ifiqueue structure modelled on ifqueues. the main motivation behind this was to show mpsafe input counters. ifiqueues, like ifqueues, allow a driver to configure multiple queueus. this in turn allows a driver with multiple

more ethernet media types

2017-11-13 Thread David Gwynne
apparently active optical cables are distinct from direct attached cables, so here's a bunch of extra media types for them. you can also get LR and ER(?) optics for 25g. ok? Index: if_media.h === RCS file:

Introduce ipsec_sysctl()

2017-11-13 Thread Martin Pieuchot
This move all IPsec tunables to netinet/ipsec_input.c without breaking the "net.inet.ip" sysctl(3) namespace. The reason for this move is to properly separate IPsec and IP globals in order to ease the removal of the NET_LOCK() in these layers. ok? Index: netinet/in.h

Re: assertion "_kernel_lock_held()" failed, uipc_socket2.c: ipsec

2017-11-13 Thread Stuart Henderson
On 2017/11/13 13:17, Martin Pieuchot wrote: > On 13/11/17(Mon) 10:03, Stuart Henderson wrote: > > On 2017/11/13 08:44, Martin Pieuchot wrote: > > > On 12/11/17(Sun) 22:10, Stuart Henderson wrote: > > > > On 2017/11/12 22:48, Martin Pieuchot wrote: > > > > > On 12/11/17(Sun) 21:30, Stuart Henderson

Re: assertion "_kernel_lock_held()" failed, uipc_socket2.c: ipsec

2017-11-13 Thread Martin Pieuchot
On 13/11/17(Mon) 12:33, Stuart Henderson wrote: > On 2017/11/13 13:17, Martin Pieuchot wrote: > > [...] > > So it seems that two of your CPU end up looking at/dealing with > > corrupted memory... > > Is that for sure? 2 does normally print a trace, 3 also drops into ddb. But none of them print:

Re: try to bundle multiple packets on an ifq before sending

2017-11-13 Thread Martin Pieuchot
On 13/11/17(Mon) 20:28, David Gwynne wrote: > > > On 13 Nov 2017, at 5:37 pm, Martin Pieuchot wrote: > > > > On 13/11/17(Mon) 10:56, David Gwynne wrote: > >> On Sun, Nov 12, 2017 at 02:45:05PM +0100, Martin Pieuchot wrote: > >>> [...] > >>> We're currently using net_tq() to

Re: assertion "_kernel_lock_held()" failed, uipc_socket2.c: ipsec

2017-11-13 Thread Sebastien Marie
On Mon, Nov 13, 2017 at 12:33:35PM +, Stuart Henderson wrote: > > Same after an hour or two uptime, but this time I get some "netlock: > lock not held" from some cpu or other, and some functions in the bits of > the trace that get displayed: > > login: panic: kernel diagnostic assertion

Re: README patch

2017-11-13 Thread Sebastian Benoit
Jason McIntyre(j...@kerhand.co.uk) on 2017.11.09 19:45:53 +: > On Thu, Nov 09, 2017 at 08:42:32PM +0100, Ingo Schwarze wrote: > > Hi, > > > > Jason McIntyre wrote on Thu, Nov 09, 2017 at 07:26:32PM +: > > > On Wed, Nov 08, 2017 at 08:14:24PM -0600, Edgar Pettijohn wrote: > > > > >> ---

Re: un-KERNEL_LOCK() TCP/UDP input & co

2017-11-13 Thread Alexandr Nedvedicky
Hello, > > So I'd like to know, what's the plan for NET_ASSERT_LOCKED() macro? > > > > a) are we going to relax it, so it will test if the net_lock is > > locked? > > Yep, that's already done. cool, thanks a lot. I've just noticed the change is there while ago, while

NET_RLOCK in ifioctl()

2017-11-13 Thread Theo Buehler
The diff below pushes the NET_LOCK into ifioctl() and reduces it to NET_RLOCK where possible. In particular, this will allow SIOCG* requests to run in parallel. ok? Index: sys/kern/sys_socket.c === RCS file:

Re: Question about tables in nested anchor on pf since 6.1

2017-11-13 Thread Leonardo Guardati
If I change the load statement in pf.uno using the full path ( /uno/due instead of due ); thens there is no error; but still no table is loaded. /etc/pf.conf: ### block log anchor "uno" load anchor "uno" from "/etc/pf.uno" ### /etc/pf.uno

Re: un-KERNEL_LOCK() TCP/UDP input & co

2017-11-13 Thread Visa Hankala
On Mon, Nov 13, 2017 at 01:25:42PM +0100, Martin Pieuchot wrote: > I'd love is somebody could do the changes in the rwlock API to let us > check if we're holding a lock. Maybe this is already present in > witness(4), visa? witness(4) does maintain data about lock ownership. The patch below might

Re: Introduce ipsec_sysctl()

2017-11-13 Thread Alexander Bluhm
On Mon, Nov 13, 2017 at 01:30:43PM +0100, Martin Pieuchot wrote: > This move all IPsec tunables to netinet/ipsec_input.c without breaking > the "net.inet.ip" sysctl(3) namespace. > > The reason for this move is to properly separate IPsec and IP globals > in order to ease the removal of the

Skip a few more ACPI PNP IDs

2017-11-13 Thread Mark Kettenis
Diff below adds a a few more entries that show up in dmesglog. PNP0303 gets renamed to match the acpica/iasl output. ok? Index: dev/acpi/acpi.c === RCS file: /cvs/src/sys/dev/acpi/acpi.c,v retrieving revision 1.332 diff -u -p

Question about tables in nested anchor on pf since 6.1

2017-11-13 Thread Leonardo Guardati
Hi, there is a confusing error message in 6.1 and 6.2 (not in 6.0) when using a table inside a nested anchor. here the rules: /etc/pf.conf: ### block log anchor "uno" load anchor "uno" from "/etc/pf.uno" ### /etc/pf.uno

ifpromic() change & bridge/switch(4) cleanup

2017-11-13 Thread Martin Pieuchot
In order to reduce the number of ifp->if_ioctl() calls, I'd like to simplify ifpromic(). There's no need for the underlying interface to be UP when calling ifpromic(). If the interface is DOWN, set or remove the flag and return. Next time the interface will be brought UP its multicast

Re: xf86-video-intel patch to test

2017-11-13 Thread Stefan Sperling
On Sat, Nov 11, 2017 at 07:57:16PM +0100, Matthieu Herrb wrote: > Hi, > > the patch below should not affect the intel(4) X.Org driver > functionality. It's sole purpose is to make it compatible with the > future upgrade to the X.Org 1.19 xserver. > > But since I don't have much hardware still

Re: armv7/arm64: enable fifo in com_fdt for "snps,dw-apb-uart"

2017-11-13 Thread Mark Kettenis
> Date: Fri, 10 Nov 2017 00:16:26 +0200 > From: Artturi Alm > > Hi, > > does also remove ti,omapX-uart compatibles from arm64 com_fdt, > as they're never to be found there. > iirc. the lenght is based on minimal lenght i found from any datasheet > for SoCs

Set soname for shared libraries

2017-11-13 Thread Mark Kettenis
The diff below changes bsd.lib.mk such that we set the DT_SONAME entry for shared libraries to the name of the shared library. This brings us in line with other ELF systems and allows us to use symlinks to tell the linker which shared libraries to link against. That in turn would allow us to

Re: ofw_clock: reset_deassert and clock_enable order of use Qs

2017-11-13 Thread Mark Kettenis
> Date: Sun, 12 Nov 2017 23:53:10 +0200 > From: Artturi Alm > > On Sat, Oct 07, 2017 at 10:16:05AM +0300, Artturi Alm wrote: > > Hi, > > > > > > what was the cause of these delays? i just spotted this, so untested on > > likely more affected HW(sunxi A64/H3 and