Re: compare memcmp with 0

2014-06-19 Thread David Gwynne
On 20 Jun 2014, at 7:35, Ted Unangst t...@tedunangst.com wrote: Always explicitly compare memcmp with 0. I find this adds clarity. i agree. ok by me if that has any value in this part of the tree. Index: s3_clnt.c === RCS

em(4) vs long mbuf chains

2014-07-03 Thread David Gwynne
em(4) appears to be susceptible to the same problem that if_bge.c r1.355 addressed. long mbuf chains can cause em_encap to fail, which will cause OACTIVE to be set on the interface, and the problem packet to remain at the start of the send queue. this will cause no packets to end up on the tx

sk(4): jumbo mbufs and rxring accounting

2014-07-09 Thread David Gwynne
this is an update of if_sk.c r1.151, which tried to introduce mclgeti. it updates it to use the if_rxring accounting. does anyone have one they can test this on? it also saves about 2k on amd64. Index: if_sk.c === RCS file:

Re: sk(4): jumbo mbufs and rxring accounting

2014-07-10 Thread David Gwynne
On 11 Jul 2014, at 9:48, Brad Smith b...@comstyle.com wrote: On 10/07/14 1:33 AM, David Gwynne wrote: this is an update of if_sk.c r1.151, which tried to introduce mclgeti. it updates it to use the if_rxring accounting. does anyone have one they can test this on? it also saves about 2k

Re: sk(4): jumbo mbufs and rxring accounting

2014-07-12 Thread David Gwynne
how about this? Index: if_sk.c === RCS file: /cvs/src/sys/dev/pci/if_sk.c,v retrieving revision 1.168 diff -u -p -r1.168 if_sk.c --- if_sk.c 19 Apr 2014 18:29:39 - 1.168 +++ if_sk.c 12 Jul 2014 08:29:20 - @@

Re: sk(4): jumbo mbufs and rxring accounting

2014-07-12 Thread David Gwynne
i think i'll try to find the sk at work and wire it up. its just annoying cos im pretty sure its sr optics with sc connectors. thanks for testing. On 13 Jul 2014, at 4:45 am, Brad Smith b...@comstyle.com wrote: On 12/07/14 4:32 AM, David Gwynne wrote: how about this? Now it attaches

Re: PATCH: fix regress/sys/kern/extent

2014-08-08 Thread David Gwynne
On Tue, Aug 05, 2014 at 08:15:59PM +, Miod Vallat wrote: Hiding a lot of sys/pool.h from userland breaks the build of the extent test in regress/sys/kern/extent on amd64. Additionally, tedu's added size variable in free(9) breaks the userland macros in sys/kern/subr_extent.c. The

Re: Convert ARP llinfo allocation to pool(9)

2014-08-11 Thread David Gwynne
why does it need queue.h? the diffs ok by me. On 11 Aug 2014, at 7:55 pm, Martin Pieuchot mpieuc...@nolizard.org wrote: Some llinfo structures might be allocated in interrupt path when ifa_rtrequest() is called, diff below convert such allocation to pool(9), ok? Index: netinet/if_ether.c

jumbo support for bnx(4)

2014-08-13 Thread David Gwynne
ive had this for 2 years or so. updated to current again. its been tested on the following: bnx0 at pci4 dev 0 function 0 Broadcom BCM5708 rev 0x12: apic 8 int 16 bnx1 at pci13 dev 0 function 0 Broadcom BCM5708 rev 0x12: apic 8 int 16 bnx0: address 00:1e:4f:1b:26:53 brgphy0 at bnx0 phy 1:

Re: jumbo support for bnx(4)

2014-08-14 Thread David Gwynne
the silence is deafening. unless anyone objects im going to commit this tomorrow (~8pm GMT+10) to force some testing. dlg On 13 Aug 2014, at 8:42 pm, David Gwynne da...@gwynne.id.au wrote: ive had this for 2 years or so. updated to current again. its been tested on the following: bnx0

Re: jumbo support for bnx(4)

2014-08-15 Thread David Gwynne
ok i have a 5709 somewhere i can work on. dlg On 15 Aug 2014, at 9:50 am, Brad Smith b...@comstyle.com wrote: On 14/08/14 5:09 PM, Brad Smith wrote: On 13/08/14 6:42 AM, David Gwynne wrote: ive had this for 2 years or so. updated to current again. its been tested on the following

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-18 Thread David Gwynne
On Sun, Jul 13, 2014 at 02:01:15PM +1000, David Gwynne wrote: i think i'll try to find the sk at work and wire it up. its just annoying cos im pretty sure its sr optics with sc connectors. thanks for testing. how's this one? Index: if_sk.c

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-18 Thread David Gwynne
Cool, thanks. On 19/08/2014 8:42 am, Brad Smith b...@comstyle.com wrote: On 18/08/14 6:24 PM, David Gwynne wrote: On Sun, Jul 13, 2014 at 02:01:15PM +1000, David Gwynne wrote: i think i'll try to find the sk at work and wire it up. its just annoying cos im pretty sure its sr optics with sc

replace ti(4) jumbo allocator with MCLGETI

2014-08-18 Thread David Gwynne
this attempts to replace to custom jumbo allocator in ti(4) with MCLGETI. it does not implement struct if_rxring, it just changes where the memory comes from. can someone test this for me? Index: ti.c === RCS file:

remove the jumbo allocator from nfe(4) cos it's not used

2014-08-18 Thread David Gwynne
the jumbo rx stuff wasnt run (look for notyet in the diff), so it's all just dead code. i dont care enough to try and fix it without access to or a use for the hardware, so id rather just remove it and save the bytes in the kernel. can anyone test to make sure i havent broken existing use? dlg

replace lge(4) jumbo allocator with MCLGETI

2014-08-18 Thread David Gwynne
i dont have the hardware, can anyone test? Index: if_lge.c === RCS file: /cvs/src/sys/dev/pci/if_lge.c,v retrieving revision 1.62 diff -u -p -r1.62 if_lge.c --- if_lge.c22 Jul 2014 13:12:11 - 1.62 +++ if_lge.c19 Aug

replace nge(4) jumbo allocator with MCLGETI

2014-08-18 Thread David Gwynne
i dont have the hardware, can anyone test? dlg Index: if_nge.c === RCS file: /cvs/src/sys/dev/pci/if_nge.c,v retrieving revision 1.79 diff -u -p -r1.79 if_nge.c --- if_nge.c22 Jul 2014 13:12:11 - 1.79 +++ if_nge.c19

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-18 Thread David Gwynne
: On Mon, Aug 18, 2014 at 6:24 PM, David Gwynne da...@gwynne.id.au wrote: On Sun, Jul 13, 2014 at 02:01:15PM +1000, David Gwynne wrote: i think i'll try to find the sk at work and wire it up. its just annoying cos im pretty sure its sr optics with sc connectors. thanks for testing. how's

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-19 Thread David Gwynne
I think sthen is referring to the original diff Kettenis had in the tree a few years ago. On 20/08/2014 4:44 am, Brad Smith b...@comstyle.com wrote: On 19/08/14 2:43 PM, Stuart Henderson wrote: From what I remember from last attempt to convert sk(4) to MCLGETI, there were problems which only

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-19 Thread David Gwynne
On 20 Aug 2014, at 7:12 am, Stuart Henderson st...@openbsd.org wrote: On 2014/08/20 07:06, David Gwynne wrote: I think sthen is referring to the original diff Kettenis had in the tree a few years ago. Yes - I just wanted to give ideas of things that might be worth testing that have been

let vlan(4) mtu be limited by the parents hardmtu instead of current mtu

2014-08-19 Thread David Gwynne
this lets you have networks on the native vlan on an interface at 1500, while setting a child vlan interfaces mtu to jumbos. ok? Index: if_vlan.c === RCS file: /cvs/src/sys/net/if_vlan.c,v retrieving revision 1.108 diff -u -p

Re: sk(4): jumbo mbufs and rxring accounting

2014-08-19 Thread David Gwynne
jumbos seem find. nfs seems fine. im going to put this and the other jalloc driver changes in to see what happens. dlg On 20 Aug 2014, at 7:41, Brad Smith b...@comstyle.com wrote: On 19/08/14 2:19 PM, Brad Smith wrote: On 18/08/14 6:24 PM, David Gwynne wrote: On Sun, Jul 13, 2014 at 02:01

Re: let vlan(4) mtu be limited by the parents hardmtu instead of current mtu

2014-08-20 Thread David Gwynne
@ David Gwynne [da...@gwynne.id.au] wrote: this lets you have networks on the native vlan on an interface at 1500, while setting a child vlan interfaces mtu to jumbos. ok? Index: if_vlan.c === RCS file: /cvs/src/sys/net/if_vlan.c,v

Re: let vlan(4) mtu be limited by the parents hardmtu instead of current mtu

2014-08-27 Thread David Gwynne
On Tue, Aug 26, 2014 at 09:11:14PM -0400, Brad Smith wrote: On 20/08/14 8:03 PM, David Gwynne wrote: sthen@ says this is likely a bit optimistic. while most of our drivers unconditionally configure their max mru, there's some stupid ones that still interpret the configured mtu as a what

Re: let vlan(4) mtu be limited by the parents hardmtu instead of current mtu

2014-08-27 Thread David Gwynne
On Tue, Aug 26, 2014 at 09:11:14PM -0400, Brad Smith wrote: On 20/08/14 8:03 PM, David Gwynne wrote: sthen@ says this is likely a bit optimistic. while most of our drivers unconditionally configure their max mru, there's some stupid ones that still interpret the configured mtu as a what

Re: bge(4) Jumbo support for newer chipsets

2014-08-28 Thread David Gwynne
On 28 Aug 2014, at 3:02 am, Mike Belopuhov m...@belopuhov.com wrote: On 27 August 2014 08:25, Brad Smith b...@comstyle.com wrote: Looking for some testing of the following diff to add Jumbo support for the BCM5714 / BCM5780 and BCM5717 / BCM5719 / BCM5720 / BCM57765 / BCM57766 chipsets.

Re: bge(4) Jumbo support for newer chipsets

2014-08-28 Thread David Gwynne
On 28 Aug 2014, at 11:05 pm, Mike Belopuhov m...@belopuhov.com wrote: On 28 August 2014 12:32, David Gwynne da...@gwynne.id.au wrote: On 28 Aug 2014, at 3:02 am, Mike Belopuhov m...@belopuhov.com wrote: On 27 August 2014 08:25, Brad Smith b...@comstyle.com wrote: Looking for some testing

Re: minphys woes

2014-08-28 Thread David Gwynne
are you expecting minphys to appear in that backtrace? On 29 Aug 2014, at 5:07, Stefan Fritsch s...@sfritsch.de wrote: Hi, while debugging some problems on a virtio host that supports only few dma segments per request, I noticed that the minphys function is not called in all code paths.

Re: minphys woes

2014-08-29 Thread David Gwynne
is it something msdos isnt doing that ffs does? On 29 Aug 2014, at 15:55, Stefan Fritsch s...@sfritsch.de wrote: On Friday 29 August 2014 10:34:20, David Gwynne wrote: are you expecting minphys to appear in that backtrace? No. In that trace minphys should have limited the transfer size

Re: minphys woes

2014-08-31 Thread David Gwynne
On 30 Aug 2014, at 6:39, Stefan Fritsch s...@sfritsch.de wrote: On Fri, 29 Aug 2014, Mike Belopuhov wrote: On 29 August 2014 11:26, Stefan Fritsch s...@sfritsch.de wrote: On Fri, 29 Aug 2014, Miod Vallat wrote: sc-sc_xfer_max is computed according to the host's capabilities. What I want

Re: minphys woes

2014-09-01 Thread David Gwynne
On 1 Sep 2014, at 9:22 pm, Mike Belopuhov m...@belopuhov.com wrote: On 1 September 2014 13:06, Stefan Fritsch s...@sfritsch.de wrote: On Mon, 1 Sep 2014, Mike Belopuhov wrote: On 29 August 2014 22:39, Stefan Fritsch s...@sfritsch.de wrote: Yes, that seems to be what happens. But if every

Re: bge(4) Jumbo support for newer chipsets

2014-09-01 Thread David Gwynne
if noone else is going to try this then i think it should go in. On 28 Aug 2014, at 20:32, David Gwynne da...@gwynne.id.au wrote: On 28 Aug 2014, at 3:02 am, Mike Belopuhov m...@belopuhov.com wrote: On 27 August 2014 08:25, Brad Smith b...@comstyle.com wrote: Looking for some testing

mpsafe interrupts for mfi(4)

2014-09-07 Thread David Gwynne
...and some mpsafety in the scsi_cmd path. this is working well for me, but considering how used it is i would appreciate some extra tests. ok? Index: pci/mfi_pci.c === RCS file: /cvs/src/sys/dev/pci/mfi_pci.c,v retrieving revision

mpsafe mpi

2014-09-08 Thread David Gwynne
mpsafe mpi. eyes would be appreciated. tests well to me. ok? Index: ic/mpi.c === RCS file: /cvs/src/sys/dev/ic/mpi.c,v retrieving revision 1.195 diff -u -p -r1.195 mpi.c --- ic/mpi.c3 Sep 2014 00:46:04 - 1.195 +++

deprecate scsi_task usage in arc(4)

2014-09-08 Thread David Gwynne
can someone test this? cheers, dlg Index: arc.c === RCS file: /cvs/src/sys/dev/pci/arc.c,v retrieving revision 1.103 diff -u -p -r1.103 arc.c --- arc.c 13 Jul 2014 23:10:23 - 1.103 +++ arc.c 9 Sep 2014 05:52:14

make top combine cpu lines by default if you have a lot of cpus

2014-09-16 Thread David Gwynne
if you have more than 8 cpus, combine the cpu lines by default. ok? Index: machine.c === RCS file: /cvs/src/usr.bin/top/machine.c,v retrieving revision 1.78 diff -u -p -r1.78 machine.c --- machine.c 4 Jul 2014 05:58:31 -

Re: make top combine cpu lines by default if you have a lot of cpus

2014-09-16 Thread David Gwynne
2014 21:51:00 +1000 From: David Gwynne da...@gwynne.id.au if you have more than 8 cpus, combine the cpu lines by default. Just curious, is there a command line argument to expand the list? /Alexander ok? 8 seems a reasonable number ok kettenis@ Index: machine.c

allow large bpf writes

2014-09-18 Thread David Gwynne
now that we have large clusters, we can easily support packets up to 64k in length. is there doco i have to update? ok? Index: bpf.c === RCS file: /cvs/src/sys/net/bpf.c,v retrieving revision 1.103 diff -u -p -r1.103 bpf.c ---

bpf read timeouts vs kqueues

2014-09-21 Thread David Gwynne
if you configure a read timeout, but then use kqs to wait for stuff to happen, it doesnt happen. you end up waiting for the buffer to fill, which might not be timely... this fixes it by starting the read timer when the kq is set up. works for me. i can provide some example code if anyone wants a

sm(4) is too verbose

2014-09-22 Thread David Gwynne
sm(4) doenst have to bpfdetach cos if_detach will do it for it. also, useless comments are useless. ok? Index: dev/ic/smc91cxx.c === RCS file: /cvs/src/sys/dev/ic/smc91cxx.c,v retrieving revision 1.35 diff -u -p -r1.35 smc91cxx.c

Re: Fix slow DHCP requests in /etc/netstart

2014-09-29 Thread David Gwynne
On 29 Sep 2014, at 19:58, Stuart Henderson st...@openbsd.org wrote: On 2014/09/26 22:49, Reyk Floeter wrote: Taking the interface down shouldn't change the link / media state on Ethernet interfaces (at least with modern drivers). Hmm - that is interesting and definitely not what I would

my screen is too big

2014-10-07 Thread David Gwynne
vi gets upset if i maximise my terminal window. it might need to go even bigger for 4k displays. ok? Index: common/options_f.c === RCS file: /cvs/src/usr.bin/vi/common/options_f.c,v retrieving revision 1.7 diff -u -p -r1.7

replace select with poll in libc

2014-10-08 Thread David Gwynne
this is the only use of select inside libc i could find. it appears pretty straightforward to replace with poll. this compiles, but i have no idea how to test it. any suggestions? or oks? Index: net/rcmd.c === RCS file:

replace select with nanosleep in libc/termios/tcsendbreak.c

2014-10-08 Thread David Gwynne
deraadt points out i suck at grep. ok? Index: termios/tcsendbreak.c === RCS file: /cvs/src/lib/libc/termios/tcsendbreak.c,v retrieving revision 1.6 diff -u -p -r1.6 tcsendbreak.c --- termios/tcsendbreak.c 5 Aug 2005 13:03:00

Re: replace select with poll in libc

2014-10-08 Thread David Gwynne
On 9 Oct 2014, at 15:11, Bryan Steele bry...@gmail.com wrote: On Thu, Oct 09, 2014 at 02:52:50PM +1000, David Gwynne wrote: this is the only use of select inside libc i could find. it appears pretty straightforward to replace with poll. this compiles, but i have no idea how to test

Re: replace select with poll in libc

2014-10-09 Thread David Gwynne
On Thu, Oct 09, 2014 at 05:36:07AM -0600, Todd C. Miller wrote: Two comments inline, otherwise OK. like this? a couple of people also suggested changing select to poll in the error messages. Index: net/rcmd.c === RCS file:

Re: replace select with poll in libc

2014-10-09 Thread David Gwynne
On Fri, Oct 10, 2014 at 02:42:08PM +1000, David Gwynne wrote: On Thu, Oct 09, 2014 at 05:36:07AM -0600, Todd C. Miller wrote: Two comments inline, otherwise OK. like this? a couple of people also suggested changing select to poll in the error messages. theo made me look at where i

Re: improving OpenBSD's gmac.c...

2014-10-12 Thread David Gwynne
dont you need endian.h to get bemtoh64 and htobem64? On 13 Oct 2014, at 7:57, Christian Weisgerber na...@mips.inka.de wrote: Here's a cleaned-up diff. Briefly tested on amd64 sparc64. I'll do some more testing tomorrow. This already has mikeb@'s blessing. Index:

Re: improving OpenBSD's gmac.c...

2014-10-13 Thread David Gwynne
On 13 Oct 2014, at 9:00 pm, Christian Weisgerber na...@mips.inka.de wrote: David Gwynne: dont you need endian.h to get bemtoh64 and htobem64? I went by the man page, which says sys/types.h. and cvs blame says that's my fault... fair enough. ill talk to philip about whether that should

make IPL levels consistent in src/sys/crypto/crypto.c

2014-10-19 Thread David Gwynne
everything else in crypto is protected at IPL_VM (via splvm), i cant see a reason for the taskq to be different. ive read all the users of the crypto api and theyre all at or below IPL_VM. can someone test this with ipsec or softraid crypto? cheers, dlg Index: crypto.c

use M_ZERO and PR_ZERO in crypto.c instead of bzero after allocation

2014-10-19 Thread David Gwynne
ok? Index: crypto.c === RCS file: /cvs/src/sys/crypto/crypto.c,v retrieving revision 1.67 diff -u -p -r1.67 crypto.c --- crypto.c14 Sep 2014 14:17:23 - 1.67 +++ crypto.c20 Oct 2014 00:26:56 - @@ -220,15 +220,12

let crypto.c pools protect themselves

2014-10-19 Thread David Gwynne
pools lock themselves, we just gotta tell them how hard. can someone test this with ipsec or softraid crypto? or ok it? Index: crypto.c === RCS file: /cvs/src/sys/crypto/crypto.c,v retrieving revision 1.68 diff -u -p -r1.68 crypto.c

have softraid crypto call crypto_dispatch appropriately instead of crypto_invoke

2014-10-19 Thread David Gwynne
the distinct impression i get is crypto_invoke is an internal abstraction to src/sys/crypto/crypto.c. it isnt documented in crypto(9), and is only used outside crypto by softraid. softraid should be calling crypto_dispatch with CRYPTO_F_NOQUEUE set if it wants/needs those semantics. can a

Re: have softraid crypto call crypto_dispatch appropriately instead of

2014-10-20 Thread David Gwynne
, mounted the test crypto partition, and sha1'd every hundredth file and compared against originals, nothing seems amiss. is that a sufficient test? yep, perfect. thank you. dlg On Mon, 20 Oct 2014 11:28:34 +1000 David Gwynne da...@gwynne.id.au wrote: the distinct impression i get

atomic.h for arm

2014-10-22 Thread David Gwynne
i think arm is the last arch that lacks the full set of ops advertised by the atomic_foo manpages. this adds support for the lowest common denominator, which in our tree is armv5. armv5 lacks the linked load and store conditional opcodes that armv6 grew. it implements that atomic sequences by

removing select from inetd

2014-10-27 Thread David Gwynne
this replaces the use of select with libevent in inetd. it has been reviewed by a few people, but tests would be appreciated. if people are still using inetd, could they give this a go and let us know if it behaves differently than expected? Index: Makefile

Re: pool page colouring

2014-10-28 Thread David Gwynne
On 29 Oct 2014, at 2:44 am, Mike Belopuhov m...@belopuhov.com wrote: On 28 October 2014 17:02, Ted Unangst t...@tedunangst.com wrote: On Tue, Oct 28, 2014 at 16:49, David Gwynne wrote: when i shuffled the locking in pools around, page colouring was left behind. page colouring is where

Re: wrong mac address used with carp and unnumbered carpdevs

2014-10-29 Thread David Gwynne
ola, while your explanation is good it still makes my head hurt. i think it makes sense, but i will throw it on a box here to try. we use both numbered and unnumbered carpdevs. dlg On 28 Oct 2014, at 23:56, Henning Brauer hb-openbsdt...@ml.bsws.de wrote: so, carp interface with

Re: socket splicing thread

2014-10-30 Thread David Gwynne
On 31 Oct 2014, at 07:10, Alexander Bluhm alexander.bl...@gmx.net wrote: Hi, Some performance measurements showed that socket splicing for TCP can be made faster. The main slowdown was that tcp_output() got called for every incomming packet. When copying through user-land this cannot

Re: socket splicing thread

2014-11-02 Thread David Gwynne
On 31 Oct 2014, at 22:37, Alexander Bluhm alexander.bl...@gmx.net wrote: On Fri, Oct 31, 2014 at 02:50:00PM +1000, David Gwynne wrote: so without splicing, the payloads from multiple tcp packets (at least all of the ones in a single softnet run?) get bundled up into a buffer

use siphash for in_pcb hash key lookups

2014-11-03 Thread David Gwynne
this uses siphash to protect the in_pcb hashes. this is pretty much a textbook example of what siphash should be used for. tests? ok? Index: conf/files === RCS file: /cvs/src/sys/conf/files,v retrieving revision 1.583 diff -u -p

Re: socket splicing thread

2014-11-03 Thread David Gwynne
On 4 Nov 2014, at 09:50, Alexander Bluhm alexander.bl...@gmx.net wrote: On Thu, Oct 30, 2014 at 09:55:35PM -0400, Ted Unangst wrote: On Thu, Oct 30, 2014 at 22:10, Alexander Bluhm wrote: + + /* Avoid user land starvation. */ + yield(); I think this is the responsibility of the

Re: network pool names

2014-11-04 Thread David Gwynne
On 4 Nov 2014, at 10:23 pm, Martin Pieuchot mpieuc...@nolizard.org wrote: Remove pl suffix, ok? ok Index: net/pipex.c === RCS file: /home/ncvs/src/sys/net/pipex.c,v retrieving revision 1.61 diff -u -p -r1.61 pipex.c ---

Re: pool page colouring

2014-11-04 Thread David Gwynne
On 30 Oct 2014, at 07:52, Ted Unangst t...@tedunangst.com wrote: On Wed, Oct 29, 2014 at 07:25, David Gwynne wrote: i dunno. im fine with either removing colouring altogether or setting it from something else completely. i just want a decision to be made cos right now ph_color isnt set

Re: pool page colouring

2014-11-04 Thread David Gwynne
On 5 Nov 2014, at 10:12, Mike Belopuhov m...@belopuhov.com wrote: On 5 November 2014 00:38, David Gwynne da...@gwynne.id.au wrote: On 30 Oct 2014, at 07:52, Ted Unangst t...@tedunangst.com wrote: On Wed, Oct 29, 2014 at 07:25, David Gwynne wrote: i dunno. im fine with either

Re: pool page colouring

2014-11-04 Thread David Gwynne
On 5 Nov 2014, at 10:27, David Gwynne da...@gwynne.id.au wrote: On 5 Nov 2014, at 10:12, Mike Belopuhov m...@belopuhov.com wrote: On 5 November 2014 00:38, David Gwynne da...@gwynne.id.au wrote: On 30 Oct 2014, at 07:52, Ted Unangst t...@tedunangst.com wrote: On Wed, Oct 29, 2014

Re: pool page colouring

2014-11-04 Thread David Gwynne
On 5 Nov 2014, at 10:58, Ted Unangst t...@tedunangst.com wrote: On Wed, Nov 05, 2014 at 09:38, David Gwynne wrote: On 30 Oct 2014, at 07:52, Ted Unangst t...@tedunangst.com wrote: On Wed, Oct 29, 2014 at 07:25, David Gwynne wrote: i dunno. im fine with either removing colouring

Re: Switch mfii(4) to msi, testing required.

2013-08-05 Thread David Gwynne
ok by me. On 05/08/2013, at 7:04 PM, Christiano F. Haesbaert haesba...@openbsd.org wrote: On Thu, Jun 06, 2013 at 12:54:31PM +0200, Christiano F. Haesbaert wrote: Hi, We would like to switch mfii(4) to msi, there is a family of supermicro X9 motherboards with incorrect ioapic routing, so

Re: octeon cnmac phy addresses

2013-09-14 Thread David Gwynne
makes a lot of sense to me, ok. you need oks from people other than me though. dlg On 14/09/2013, at 8:34 PM, Jonathan Matthew jmatt...@openbsd.org wrote: The network interfaces on the ubiquiti edgerouter lite don't work yet. Here's the first step towards fixing that: Mapping port numbers

viomb(4): replace workqs with tasks

2013-10-29 Thread David Gwynne
i dont have the hardware to test this on. could someone give it a go and see if it still works? cheers, dlg Index: viomb.c === RCS file: /cvs/src/sys/dev/pci/viomb.c,v retrieving revision 1.6 diff -u -p -r1.6 viomb.c --- viomb.c

alpha pmap pool tweak

2013-10-30 Thread David Gwynne
can someone with an alpha test this diff for me? if your computer still boots after you're running a kernel with it, i will consider it tested. Index: arch/alpha/alpha/pmap.c === RCS file: /cvs/src/sys/arch/alpha/alpha/pmap.c,v

fd(4): replace disksort() with a bufq

2013-10-30 Thread David Gwynne
this replaces the bare use of disksort with a bufq in fd. this means it will use the new system default of the nscan sorting algorithm, but that shouldnt be a noticable change. it works for me, but i would love to hear from someone with a real floppy drive still... ok? Index: fd.c

fd(4/sparc64): replace disksort() with bufqs

2013-10-30 Thread David Gwynne
this is a port of the same change to the isa/fd.c code to the sparc64 driver. it effectively replaces disksort with nscan. tests? ok? Index: fd.c === RCS file: /cvs/src/sys/arch/sparc64/dev/fd.c,v retrieving revision 1.40 diff -u -p

octcf(4/octeon): switch disksort() to bufqs

2013-10-31 Thread David Gwynne
jmatthew@ tells me this compiles, but we dont have hardware to test with. tests? oks? go for its? cheers, dlg Index: octcf.c === RCS file: /cvs/src/sys/arch/octeon/dev/octcf.c,v retrieving revision 1.12 diff -u -p -r1.12 octcf.c

disksort() vs bufqs

2013-11-02 Thread David Gwynne
once upon a time pretty much all block storage was on spinning disks, so all the drivers for disks tried to order queued io to better suit moving a head across a platter by calling disksort. between then and now a lot of the assumptions that disksort relied on are no longer true (eg, SSDs are a

Re: Fixing an LLVM warning in the i2o code

2013-11-04 Thread David Gwynne
On 5 Nov 2013, at 12:40 am, Kenneth R Westerback kwesterb...@rogers.com wrote: On Sun, Nov 03, 2013 at 10:51:43PM -0500, Brad Smith wrote: LLVM errors out on the i2o code with the following warning.. ../../../../dev/i2o/iop.c:2399:42: error: comparison of unsigned expression 0 is always

Re: disksort() vs bufqs

2013-11-16 Thread David Gwynne
here's a diff to sys/arch/sparc/dev/fd.c. not compiled, and therefore not tested. anyone? ok? On Sun, Nov 03, 2013 at 09:30:56AM +1000, David Gwynne wrote: once upon a time pretty much all block storage was on spinning disks, so all the drivers for disks tried to order queued io to better

Re: convert sppp(4) to taskq

2013-11-18 Thread David Gwynne
On 18 Nov 2013, at 11:59 pm, Martin Pieuchot mpieuc...@nolizard.org wrote: On 18/11/13(Mon) 13:35, Stefan Sperling wrote: On Mon, Nov 18, 2013 at 12:37:53PM +0100, Martin Pieuchot wrote: Even if right now calling task_del() is enough, do you know if there's an easy way to convert this code

Re: disksort() vs bufqs

2013-11-18 Thread David Gwynne
this one is for hd(4) on hp300s. again, i cant compile and therefore cant test. anyone able to compile this for me? does anyone have such hardware? On Sun, Nov 03, 2013 at 09:30:56AM +1000, David Gwynne wrote: once upon a time pretty much all block storage was on spinning disks, so all

Re: disksort() vs bufqs

2013-11-18 Thread David Gwynne
this one is for xy(4) on sparc. again, i cant test, so if anyone could do that for me it would be appreciated. does anyone have one of these? Index: xy.c === RCS file: /cvs/src/sys/arch/sparc/dev/xy.c,v retrieving revision 1.57 diff

Re: disksort() vs bufqs

2013-11-18 Thread David Gwynne
and here's the last one, the hdc stuff on vax. can anyone compile this? test? Index: hdc9224.c === RCS file: /cvs/src/sys/arch/vax/vsa/hdc9224.c,v retrieving revision 1.41 diff -u -p -r1.41 hdc9224.c --- hdc9224.c 14 Oct 2013

remove disksort()

2013-11-20 Thread David Gwynne
the subject says it all really. this is sort of inspired by 5d2ecd5224 in bitrig except this brings all the architectures and device drivers forward (i didnt get to delete any to prepare for this), and maintains an algorithm for trying to order io on spinning rust (nscan). ok? Index:

remove the b_cylinder #define to b_resid in struct buf

2013-11-20 Thread David Gwynne
most hardware and most of the kernel has given up on cylinders meaning anything, we generally work in logical block addresses now. now that disksort is (almost) gone it is no longer used for anything meaningful except in floppy disk drivers. the #define b_cylinder b_resid alias has always grated

uncruft device.h and subr_autoconf.c a bit

2013-11-26 Thread David Gwynne
this removes the idea of multiple cfdata tables. code for it was introduced for some reason for some hotplug stuff 17 years ago, if 0'd 14 years ago, and removed 11 years ago. i dont think we need it. ok? Index: sys/device.h ===

TIMEOUT_INITIALIZER()

2013-12-08 Thread David Gwynne
this lets you init a timeout when its declared, much like how things like MUTEX_INITIALIZER or RWLOCK_INITIALIZER work. ok? Index: sys/sys/timeout.h === RCS file: /cvs/src/sys/sys/timeout.h,v retrieving revision 1.24 diff -u -p

TIMEOUT_INITIALIZER() for src/sys/dev.c to avoid timeout_initialized()

2013-12-08 Thread David Gwynne
this is a demonstration of using TIMEOUT_INITIALIZED(). because we know the timeout is always set up correctly, we dont have to test for it all over the place. Index: rnd.c === RCS file: /cvs/src/sys/dev/rnd.c,v retrieving revision

timeout_pending is not the best

2013-12-09 Thread David Gwynne
timeout_pending followed by timeot_del is pointless, as timeout_del effectively does the pending check itself. now that timeout_del returns whether the timeout was actually removed, you can make decisions based on that rather than race with timeout_pending checking timeout_pending shortly after

Re: TIMEOUT_INITIALIZER() for src/sys/dev.c to avoid timeout_initialized()

2013-12-09 Thread David Gwynne
On 9 Dec 2013, at 6:59 pm, Bret Lambert blamb...@openbsd.org wrote: On Mon, Dec 09, 2013 at 01:55:53PM +1000, David Gwynne wrote: this is a demonstration of using TIMEOUT_INITIALIZED(). because we know the timeout is always set up correctly, we dont have to test for it all over the place

Re: TIMEOUT_INITIALIZER() for src/sys/dev.c to avoid timeout_initialized()

2013-12-09 Thread David Gwynne
On 9 Dec 2013, at 10:01 pm, David Gwynne da...@gwynne.id.au wrote: On 9 Dec 2013, at 6:59 pm, Bret Lambert blamb...@openbsd.org wrote: On Mon, Dec 09, 2013 at 01:55:53PM +1000, David Gwynne wrote: this is a demonstration of using TIMEOUT_INITIALIZED(). because we know the timeout

Re: tftp-proxy(8) with nat-to

2013-12-19 Thread David Gwynne
im glad you wrote a diff rather than simply complain that nat and tftp doesnt work. the moving parts generally look good to me apart from the struct src_addr and getopt chunks. please use sockaddr_storage instead of sockaddr in the src_addr struct. could you split the resolution of the

Re: security(8) check maildir as well as mailbox permissions

2013-12-19 Thread David Gwynne
On 20 Dec 2013, at 2:56 am, Alexander Hall alexan...@beard.se wrote: Henning Brauer lists-openbsdt...@bsws.de wrote: * Craig R. Skinner skin...@britvault.co.uk [2013-12-19 10:18]: On 2013-12-18 Wed 20:48 PM |, J??r??mie Courr??ges-Anglas wrote: skin...@britvault.co.uk (Craig R. Skinner)

Re: tftp-proxy(8) with nat-to

2013-12-20 Thread David Gwynne
wrote: On Fri, Dec 20, 2013 at 01:17:08PM +1000, David Gwynne wrote: im glad you wrote a diff rather than simply complain that nat and tftp doesnt work. the moving parts generally look good to me apart from the struct src_addr and getopt chunks. please use sockaddr_storage instead

Re: bus_dmamap_sync semantics

2013-12-27 Thread David Gwynne
Ok On 27/12/2013 5:28 am, Mark Kettenis mark.kette...@xs4all.nl wrote: Our bus_dmamap(9) man page says: On platforms which implement reordered stores, bus_dmamap_sync() will always cause the store buffer to be flushed. In 2003 NetBSD actually changed that to: On platforms which

sgi/mvme68k tests for sbic(4)

2014-01-17 Thread David Gwynne
this gets rid of XS_NO_CCB in sbic(4) by moving to iopools. i dont have an arch that uses this so i mostly want compile testers, but if someone actually has the hardware that would be great. this change is mildly interesting because it demonstrates the flexibility of iopools at sharing resources

no XS_NO_CCB for vax/ncr(4) or sparc/si(4)

2014-01-17 Thread David Gwynne
can anyone compile or even test this on a sparc or vax for me? cheers, dlg Index: ncr5380sbc.c === RCS file: /cvs/src/sys/dev/ic/ncr5380sbc.c,v retrieving revision 1.30 diff -u -p -r1.30 ncr5380sbc.c --- ncr5380sbc.c17 Jul

Re: em(4): Don't count RX overruns and missed packets as input errros

2014-01-27 Thread David Gwynne
On 26 Jan 2014, at 11:31 am, Brad Smith b...@comstyle.com wrote: On 31/12/13 5:50 AM, Mike Belopuhov wrote: On 31 December 2013 09:46, Brad Smith b...@comstyle.com wrote: On 31/12/13 3:14 AM, Mark Kettenis wrote: Date: Tue, 31 Dec 2013 01:28:04 -0500 From: Brad Smith b...@comstyle.com

Re: hppa mutex ipl diff

2014-01-29 Thread David Gwynne
is there anything in particular you're supposed to look for to see if it is or isnt working? either it locks up the machine (deadlock) or doesnt? On 29 Jan 2014, at 8:06 pm, Mark Kettenis mark.kette...@xs4all.nl wrote: Here is a diff that makes mutexes raise their ipl to the highest level

Re: /usr/src/sys/dev/pci/mfii.c compile failures

2014-03-28 Thread David Gwynne
i fixed this in src/sys/dev/pci/mfii.c r1.14. cheers, dlg On 28 Mar 2014, at 7:43 am, kspill...@acm.org wrote: Is anyone else seeing this error building -current? /usr/src/sys/dev/pci/mfii.c: In function 'mfii_initialise_firmware': /usr/src/sys/dev/pci/mfii.c:1004: error: 'struct

Re: Dell R310 - H200 Raid performance problem

2011-02-17 Thread David Gwynne
this diff implements the disk cache ioctl handling in mpii so sd(4) can drive the change rather than have mpii(4) whack everything. modelled on the same functionality in mpi(4) and mikeb's code... could someone test this please? Index: mpii.c

Re: Dell R310 - H200 Raid performance problem

2011-02-20 Thread David Gwynne
i believe the diff below should work out of the box. it pulls in all mikeb's fixes. On Fri, Feb 18, 2011 at 07:54:09PM +0100, ??ukasz Czarniecki wrote: With following Mike's suggestions it worked. # scsi -f /dev/rsd0c -m 8 IC: 0 ABPF: 0 CAP: 0 DISC: 0 SIZE: 0 WCE: 1 MF: 0

  1   2   3   4   5   6   7   8   9   10   >