Re: em(4): enable TCP/UDP checksum offload

2013-07-10 Thread Hrvoje Popovski
On 7.11.2012. 12:57, Hrvoje Popovski wrote: 82546GB pf (with henning's patch) + ipforward and with or without em tcp/udp checksum - not working 82541GI pf (with henning's patch) + ipforward + em tcp/udp checksum - works like charm I don't know if it's a good time, but now with new

Re: use time_uptime for various pf expirations

2012-07-29 Thread Hrvoje Popovski
On 28.7.2012. 21:56, Florian Obser wrote: Benno and me were looking into why pflow is sending flows with starttime after endtime. We believe this was introduced with this commit: CVSROOT:/cvs Module name:

Re: use time_uptime for various pf expirations

2012-07-29 Thread Hrvoje Popovski
On 29.7.2012. 18:09, Mike Belopuhov wrote: On Sun, Jul 29, 2012 at 5:47 PM, Hrvoje Popovski wrote: Hello, i have tested your patches with torrent box behind 2 firewalls (nat, carp,pfsync). One firewall is patched and second box isn't. Both of them quite often export flows with duration

Re: use time_uptime for various pf expirations

2012-07-29 Thread Hrvoje Popovski
On 29.7.2012. 18:09, Mike Belopuhov wrote: On Sun, Jul 29, 2012 at 5:47 PM, Hrvoje Popovski wrote: Hello, i have tested your patches with torrent box behind 2 firewalls (nat, carp,pfsync). One firewall is patched and second box isn't. Both of them quite often export flows with duration

Re: use time_uptime for various pf expirations

2012-07-30 Thread Hrvoje Popovski
On 30.7.2012. 12:32, Florian Obser wrote: On Mon, Jul 30, 2012 at 12:07:15AM +0200, Hrvoje Popovski wrote: On 29.7.2012. 18:09, Mike Belopuhov wrote: On Sun, Jul 29, 2012 at 5:47 PM, Hrvoje Popovski wrote: Hello, i have tested your patches with torrent box behind 2 firewalls (nat, carp

Re: use time_uptime for various pf expirations

2012-08-24 Thread Hrvoje Popovski
cameild@ noticed this problem last year. Hrvoje Popovski tested an older version without the pfsync fix. Since this changes the semantics of a field in pfsync both pfsync machines need to be updated (I think it's ok to import 5.1 states into this version but the other way around will cause

Re: use time_uptime for various pf expirations

2012-09-07 Thread Hrvoje Popovski
cameild@ noticed this problem last year. Hrvoje Popovski tested an older version without the pfsync fix. Since this changes the semantics of a field in pfsync both pfsync machines need to be updated (I think it's ok to import 5.1 states into this version but the other way around will cause

Re: em(4): enable TCP/UDP checksum offload

2012-11-05 Thread Hrvoje Popovski
On 5.11.2012. 14:38, Hrvoje Popovski wrote: Hello, it's working here. Please ignore working here part. It is enabled here. Will play with combination of 82546GB and 82541GI interfaces.

Re: em(4): enable TCP/UDP checksum offload

2012-11-07 Thread Hrvoje Popovski
On 5.11.2012 23:45, Hrvoje Popovski wrote: On 5.11.2012. 14:38, Hrvoje Popovski wrote: Hello, it's working here. Please ignore working here part. It is enabled here. Will play with combination of 82546GB and 82541GI interfaces. 82546GB pf (with henning's patch) + ipforward

Re: VPLS patch [0/3]: introduction

2014-11-14 Thread Hrvoje Popovski
On 14.11.2014. 18:24, Rafael Zalamena wrote: On Fri, Nov 14, 2014 at 05:41:32PM +0100, Mike Belopuhov wrote: On 14 November 2014 17:26, Rafael Zalamena rzalam...@gmail.com wrote: On Sun, Sep 14, 2014 at 11:48:11PM -0300, Rafael Zalamena wrote: The following mails will contain patchs that

Re: Dell R630 high interrupts on acpi0

2014-12-16 Thread Hrvoje Popovski
On 16.12.2014. 6:16, Jonathan Matthew wrote: On Sun, Dec 14, 2014 at 06:22:37PM +0100, Hrvoje Popovski wrote: Hi all, I have got two new Dell R630 and have current on them from Sun Dec 14 15:07:17. Installation went great and very fast. The problem is that I see around 11k interrupts on acpi0

Re: Dell R630 high interrupts on acpi0

2014-12-16 Thread Hrvoje Popovski
On 16.12.2014. 6:16, Jonathan Matthew wrote: We just got some r630s too, so I spent some time last week figuring out what's going on here. Something in the AML wants to talk to the intel MEI device. Normally this works, but on the new generation of dell machines (we've seen it on r630s and

Re: Dell R630 high interrupts on acpi0

2014-12-19 Thread Hrvoje Popovski
On 17.12.2014. 6:34, Philip Guenther wrote: Uh, ACPI *requires* that C1 exist. The halt instruction is defined as entering C1, so not having C1 would mean your CPU lacks a basic manadatory ia32 instruction. Hopefully the BIOS docs explain that you're just disabling deep C-states or something

typo in pcidevs ... ?

2015-03-19 Thread Hrvoje Popovski
Hi all, I think that in /usr/src/sys/dev/pci/pcidevs these 3 records product INTEL E5V2_ERR_20x0be3 E5 v2 Error product INTEL E5V2_ERR_30x0be6 E5 v2 Error product INTEL E5V2_ERR_40x0be7 E5 v2 Error should be product INTEL E5V2_ERR_20x0eb3 E5 v2 Error product

Re: typo in pcidevs ... ?

2015-04-27 Thread Hrvoje Popovski
On 19.3.2015. 16:52, Hrvoje Popovski wrote: Hi all, I think that in /usr/src/sys/dev/pci/pcidevs these 3 records product INTEL E5V2_ERR_20x0be3 E5 v2 Error product INTEL E5V2_ERR_30x0be6 E5 v2 Error product INTEL E5V2_ERR_40x0be7 E5 v2 Error should

Re: Loosing characters on Dell R420 Drac console

2015-07-15 Thread Hrvoje Popovski
On 9.1.2015. 18:24, Matthieu Herrb wrote: Hi, I've an issue on a Dell R420 server running OpenBSD. I tried both 5.6-stable and -current on it with the same behavior. I'm using a DRAC 7 Enterprise remote access card to access the virtual console through their Java web start software. I'm

Re: acpi fix needs testing

2015-07-19 Thread Hrvoje Popovski
On 20.7.2015. 0:06, Mark Kettenis wrote: The acpi code that reads and writes pci config space is quite busted. It always does byte-sized reads and writes even if the aml specifies an access size of four bytes. This made vmware unhappy, because it expetcs to see a magic value being written

Re: acpi fix needs testing

2015-07-19 Thread Hrvoje Popovski
On 20.7.2015. 0:53, Bob Beck wrote: I'm pretty sure that's a different problem. But thanks for pointing this out. -ml no, just update your tree please, I just committed the fix thank you, everything is working after your fix

Re: acpi fix needs testing

2015-07-19 Thread Hrvoje Popovski
On 20.7.2015. 0:06, Mark Kettenis wrote: The acpi code that reads and writes pci config space is quite busted. It always does byte-sized reads and writes even if the aml specifies an access size of four bytes. This made vmware unhappy, because it expetcs to see a magic value being written

Re: acpi fix needs testing

2015-07-19 Thread Hrvoje Popovski
On 20.7.2015. 0:41, Mike Larkin wrote: On Mon, Jul 20, 2015 at 12:38:19AM +0200, Hrvoje Popovski wrote: On 20.7.2015. 0:06, Mark Kettenis wrote: The acpi code that reads and writes pci config space is quite busted. It always does byte-sized reads and writes even if the aml specifies an access

Re: Unlocking ix(4) a bit further

2015-10-20 Thread Hrvoje Popovski
he rx mutex that claudio@ put in recently. >>>> >>>> I don't have any ix(4) hardware. So the only guarantee is that this >>>> compiles ;). >>> >>> Based on the experience with em(4), here is an updated diff. >> >> What's happening to this? Mark did you ge

Re: first step of ix(4) without kernel lock

2015-09-10 Thread Hrvoje Popovski
On 10.9.2015. 9:47, Peter Hessler wrote: > On 2015 Sep 09 (Wed) at 23:57:57 +0200 (+0200), Hrvoje Popovski wrote: > :On 9.9.2015. 10:10, Claudio Jeker wrote: > :> This is a port of the em(4) IPL_MPSAFE changes made by kettenis@ to ix(4). > :> Seems to work for me but don't e

Re: first step of ix(4) without kernel lock

2015-09-10 Thread Hrvoje Popovski
On 10.9.2015. 9:57, Claudio Jeker wrote: > There are still some issues with this first version. Here a 2nd version > that pushes the kernel lock even a bit further and tries to solve some of > the problems between the interrupt handler and the ioctl code. > There is still something not fully right

Re: first step of ix(4) without kernel lock

2015-09-10 Thread Hrvoje Popovski
On 10.9.2015. 9:57, Claudio Jeker wrote: > On Wed, Sep 09, 2015 at 11:57:57PM +0200, Hrvoje Popovski wrote: >> On 9.9.2015. 10:10, Claudio Jeker wrote: >>> This is a port of the em(4) IPL_MPSAFE changes made by kettenis@ to ix(4). >>> Seems to work for me bu

Re: first step of ix(4) without kernel lock

2015-09-09 Thread Hrvoje Popovski
On 9.9.2015. 10:10, Claudio Jeker wrote: > This is a port of the em(4) IPL_MPSAFE changes made by kettenis@ to ix(4). > Seems to work for me but don't expect any miracles. > > Please test > Hi, i am testing your patch with bridged and routed setup and everything works nice. ix card is dual

Re: first step of ix(4) without kernel lock

2015-09-10 Thread Hrvoje Popovski
On 10.9.2015. 17:02, Hrvoje Popovski wrote: > On 10.9.2015. 9:57, Claudio Jeker wrote: >> On Wed, Sep 09, 2015 at 11:57:57PM +0200, Hrvoje Popovski wrote: >>> On 9.9.2015. 10:10, Claudio Jeker wrote: >>>> This is a port of the em(4) IPL_MPSAFE changes made by kettenis

Re: Unlocking ix(4) a bit further

2015-09-30 Thread Hrvoje Popovski
On 30.9.2015. 14:30, Mark Kettenis wrote: > Since people seemed to like my diff for em(4), here is one for ix(4). > In addition to unlocking the rx completion path, this one also uses > intr_barrier() and removes the rx mutex that claudio@ put in recently. > > I don't have any ix(4) hardware. So

Re: Unlocking ix(4) a bit further

2015-09-30 Thread Hrvoje Popovski
On 30.9.2015. 19:20, Hrvoje Popovski wrote: > On 30.9.2015. 14:30, Mark Kettenis wrote: >> Since people seemed to like my diff for em(4), here is one for ix(4). >> In addition to unlocking the rx completion path, this one also uses >> intr_barrier() and removes the rx mu

tcpdump bpf_catchpacket crash

2015-09-30 Thread Hrvoje Popovski
Hi, I have tested mp ix patches and at one point very naively, i did tcpdump on interface on which i generate traffic :) and everything is fine until i quit tcpdump with ^C, then crash happened. traffic is generated at high rate (10Mpps). this crash is 100% reproducible. in ddb console i can't do

Re: tcpdump bpf_catchpacket crash

2015-10-01 Thread Hrvoje Popovski
On 1.10.2015. 11:24, Martin Pieuchot wrote: > On 01/10/15(Thu) 11:11, Hrvoje Popovski wrote: >> On 1.10.2015. 0:43, Stuart Henderson wrote: >>> In /sys/ddb/db_trap.c line 72, change if(panicstr != NULL) to if(1), it >>> should give you a backtrace and short ps (show

Re: tcpdump bpf_catchpacket crash

2015-10-01 Thread Hrvoje Popovski
tries (which includes uvm faults). tnx .. here's screenshot http://kosjenka.srce.hr/~hrvoje/zaprocvat/crash2.jpg > On 30 September 2015 23:30:02 BST, Hrvoje Popovski <hrv...@srce.hr> wrote: >> Hi, >> I have tested mp ix patches and at one point very naively, i did >

Re: sleep in interface detach until all refs are released

2015-09-27 Thread Hrvoje Popovski
On 27.9.2015. 7:51, David Gwynne wrote: > this uses the refcnt api to do the heavy lifting. > > i think we have all the if_get/if_put calls we need, so this should > be safe at this point. > > if anyone wants to test, can you try detaching or destroying > interfaces and check that the ifconfig

Re: Unlocking em(4) a bit further

2015-09-29 Thread Hrvoje Popovski
On 29.9.2015. 1:00, Hrvoje Popovski wrote: > On 28.9.2015. 23:53, Mark Kettenis wrote: >> This diff makes the tx completion path run without the kernel lock >> held. With this change, the interrupt handler will not grab the >> kernel lock under normal circumstances. The

Re: Unlocking em(4) a bit further

2015-09-28 Thread Hrvoje Popovski
On 28.9.2015. 23:53, Mark Kettenis wrote: > This diff makes the tx completion path run without the kernel lock > held. With this change, the interrupt handler will not grab the > kernel lock under normal circumstances. The diff follows the same > approach as dlg@ took with vmx(4). > > The diff

Re: Make em(4) more mpsafe again

2015-12-07 Thread Hrvoje Popovski
On 5.12.2015. 15:41, Claudio Jeker wrote: > So Mark and I spent some time to figure out what the issue was with ix(4) > based on that info I resurected the em(4) mpsafe diff that got backed out > and I applied the same fix. It is somewhat unclear if this fixes the > watchdog timeouts since in

Re: mpsafe bnx(4)

2015-12-03 Thread Hrvoje Popovski
On 2.12.2015. 12:14, Jonathan Matthew wrote: > This is mostly a backout of if_bnx.c r1.77, which introduced lists of tx > descriptors, allocated on demand, in order to avoid allocating space per ring > slot. These days I think we can afford a few kb of memory overhead if it > makes the packets go

Re: Make ix(4) mpsafe: take 2

2015-12-04 Thread Hrvoje Popovski
On 4.12.2015. 12:47, Mark Kettenis wrote: > Here is a new diff to make ix(4) mpsafe. Should now longer get stuck > in the OACTIVE state. Tests more than welcome. > Hi, i have tested this patch with 82599 and x540 while sending 6Mpps for cca 3 hours and ifconfig down/up and everything is

Re: taskctx and revisiting if_start serialisation

2015-12-06 Thread Hrvoje Popovski
On 6.12.2015. 5:00, David Gwynne wrote: > the current code for serialising if_start calls for mpsafe nics does what it > says. > > however, kettenis realised it doesnt help us much when we're trying > to coordinate between the start and txeof side of a driver when > setting or clearing oactive.

Re: taskctx and revisiting if_start serialisation

2015-12-06 Thread Hrvoje Popovski
On 6.12.2015. 15:56, Hrvoje Popovski wrote: > On 6.12.2015. 5:00, David Gwynne wrote: >> the current code for serialising if_start calls for mpsafe nics does what it >> says. >> >> however, kettenis realised it doesnt help us much when we're trying >> to coord

Re: Make ix(4) mpsafe: take 2

2015-12-05 Thread Hrvoje Popovski
On 4.12.2015. 17:35, Hrvoje Popovski wrote: > On 4.12.2015. 12:47, Mark Kettenis wrote: >> Here is a new diff to make ix(4) mpsafe. Should now longer get stuck >> in the OACTIVE state. Tests more than welcome. >> > > > Hi, > > i have tested this patch wit

Re: taskctx and revisiting if_start serialisation

2015-12-07 Thread Hrvoje Popovski
On 6.12.2015. 15:56, Hrvoje Popovski wrote: > On 6.12.2015. 5:00, David Gwynne wrote: >> the current code for serialising if_start calls for mpsafe nics does what it >> says. >> >> however, kettenis realised it doesnt help us much when we're trying >> to coord

Re: Make ix(4) mpsafe: take 2

2015-12-05 Thread Hrvoje Popovski
On 5.12.2015. 16:35, Peter Hessler wrote: > On 2015 Dec 05 (Sat) at 16:16:50 +0100 (+0100), Hrvoje Popovski wrote: > :On 4.12.2015. 17:35, Hrvoje Popovski wrote: > :> On 4.12.2015. 12:47, Mark Kettenis wrote: > :>> Here is a new diff to make ix(4) mpsafe. Should

Re: em watchdog timeout

2016-01-04 Thread Hrvoje Popovski
On 3.1.2016. 21:26, Hrvoje Popovski wrote: > Hi all, > > i accidentally managed to trigger em watchdog timeout -- resetting on my > test box which is clean -current from today at 20:00 CET. > Test box is IBM X3550M4 with 4 onboard em I350 (em0,em1,em2,em3) and 2 > ix (ix2,ix3

Re: Make em(4) more mpsafe again

2016-01-02 Thread Hrvoje Popovski
On 31.12.2015. 15:08, David Gwynne wrote: > On Sat, Dec 05, 2015 at 03:41:24PM +0100, Claudio Jeker wrote: >> So Mark and I spent some time to figure out what the issue was with ix(4) >> based on that info I resurected the em(4) mpsafe diff that got backed out >> and I applied the same fix. It is

Re: Another mpsafe ix(4) diff

2015-12-29 Thread Hrvoje Popovski
On 29.12.2015. 17:49, Mark Kettenis wrote: >> Date: Tue, 22 Dec 2015 23:45:49 +0100 >> > >> > On 22.12.2015. 22:08, Mark Kettenis wrote: >>> > > Anybody willing to give this a spin? I don't have access to hardware >>> > > currently... >>> > > >>> > > Thanks, >>> > > >>> > > Mark >> > >> > Hi,

Re: rework em(4) tx ring handling

2016-01-08 Thread Hrvoje Popovski
On 8.1.2016. 6:18, David Gwynne wrote: > i found em_txeof hard to read, so i did this. > > i was originally going to follow myx/re and use a separate prod/cons > for a ring of packet info which would point into the descriptor > ring, but came to the conclusion that the "end of packet" pointer >

Re: Make em(4) more mpsafe again

2015-12-23 Thread Hrvoje Popovski
On 11.12.2015. 10:47, Martin Pieuchot wrote: > On 05/12/15(Sat) 15:41, Claudio Jeker wrote: >> So Mark and I spent some time to figure out what the issue was with ix(4) >> based on that info I resurected the em(4) mpsafe diff that got backed out >> and I applied the same fix. It is somewhat

Re: Another mpsafe ix(4) diff

2015-12-30 Thread Hrvoje Popovski
On 29.12.2015. 22:23, Hrvoje Popovski wrote: > On 29.12.2015. 17:49, Mark Kettenis wrote: >>> Date: Tue, 22 Dec 2015 23:45:49 +0100 >>>> >>>> On 22.12.2015. 22:08, Mark Kettenis wrote: >>>>>> Anybody willing to give this a s

Re: make IFF_OACTIVE mpsafe

2015-11-23 Thread Hrvoje Popovski
On 23.11.2015. 15:09, David Gwynne wrote: > by replacing it. > > hrvoje popovski was prescient in predicting this. > > anyway. there are two things shared between the network stack and > drivers in the send path: the send queue and the IFF_OACTIVE flag. > the send q

Re: Unlocking ix(4) a bit further

2015-11-28 Thread Hrvoje Popovski
he rx mutex that claudio@ put in recently. >>>> >>>> I don't have any ix(4) hardware. So the only guarantee is that this >>>> compiles ;). >>> >>> Based on the experience with em(4), here is an updated diff. >> >> What's happening to this? Mark did

Re: serialise if_start calls

2015-11-30 Thread Hrvoje Popovski
On 30.11.2015. 12:55, David Gwynne wrote: > this tweaks the guts of if_start so it guarantees that there's only > ever one call to ifp->if_start running in the system at a time. > previously this was implicit because it could only be called with > the KERNEL_LOCK held. > > as we move forward it

Re: trunk vs busy ports

2015-11-20 Thread Hrvoje Popovski
On 20.11.2015. 6:36, David Gwynne wrote: > IFF_OACTIVE means the hardware ring is full, not if it is busy. > > perhaps a better check is to see whether there are pending packets > on the send queue? > > i could also argue we dont need the check at all, but this is less > of a semantic change.

Re: trunk vs busy ports

2015-11-20 Thread Hrvoje Popovski
On 20.11.2015. 13:26, Hrvoje Popovski wrote: > On 20.11.2015. 6:36, David Gwynne wrote: >> IFF_OACTIVE means the hardware ring is full, not if it is busy. >> >> perhaps a better check is to see whether there are pending packets >> on the send queue? >> >> i

snapshot from 26-Jan-2016 - pfsync panic

2016-01-28 Thread Hrvoje Popovski
Hi all, i have pf,carp,pfsync and dhcpd setup with 2 Dell R610. today i updated my secondary firewall with latest snapshot from http://ftp2.eu.openbsd.org/ install59.iso 26-Jan-2016 04:00 and after update secondary firewall panic http://kosjenka.srce.hr/~hrvoje/crash2.jpg i couldn't type

Re: DDB causing lost keystrokes on Dell iDRAC console (not inside ddb)

2016-01-24 Thread Hrvoje Popovski
On 23.1.2016. 23:29, Adam McDougall wrote: > Hello, > > I have a few Dell servers which I've installed OpenBSD for testing > but ran into a problem with keystroke loss on the console when used > through the Dell iDRAC remote graphical console. Surprisingly it > operates perfectly fine in the

renesas unknown product on dell r630

2016-02-20 Thread Hrvoje Popovski
hi, this patch adds renesas product on dell r630 dmesg on dell r630 without patch: pci6 at ppb5 bus 6 ppb6 at pci6 dev 0 function 0 vendor "Renesas", unknown product 0x001d rev 0x00 pci7 at ppb6 bus 7 ppb7 at pci7 dev 0 function 0 vendor "Renesas", unknown product 0x001d rev 0x00 pci8 at ppb7

Re: New scheduler for OpenBSD

2016-03-20 Thread Hrvoje Popovski
On 18.3.2016. 20:00, Mark Kettenis wrote: > One other important case to test is network packet forwarding. Some > of our network stack is now running inside a kernel thread. And any > changes in the scheduling behaviour have the potential of having a > significant impact there. I've done some

Re: MSI-X support

2016-05-04 Thread Hrvoje Popovski
On 4.5.2016. 8:50, Mark Kettenis wrote: >> Date: Tue, 3 May 2016 15:23:07 -0700 >> From: Mike Larkin >> >> On Tue, May 03, 2016 at 09:40:28PM +0200, Mark Kettenis wrote: >>> Today mpi@ reminded me that I had written support for MSI-X some time >>> ago. Since he is

Re: ix(4): enable checksum offload

2016-04-16 Thread Hrvoje Popovski
On 9.9.2013. 22:07, Mike Belopuhov wrote: > On 9 September 2013 21:48, Brad Smith wrote: >> Here is a diff to enable the checksum offload support for ix(4). >> >> Looking for any testing. >> > > last time i checked this broke ospf traffic. please make sure at least > ip/tcp,

Re: Moving away from softnet interrupts

2016-04-18 Thread Hrvoje Popovski
On 18.4.2016. 10:50, Martin Pieuchot wrote: > The current goal of the Network SMP effort is to have a single CPU > process the IP forwarding path in a process context without holding > the KERNEL_LOCK(). To achieve this goal we're progressively moving > code from the softnet interrupt context to

Re: Moving away from softnet interrupts

2016-04-18 Thread Hrvoje Popovski
On 18.4.2016. 15:31, Hrvoje Popovski wrote: > On 18.4.2016. 10:50, Martin Pieuchot wrote: >> The current goal of the Network SMP effort is to have a single CPU >> process the IP forwarding path in a process context without holding >> the KERNEL_LOCK(). To achieve this goa

Re: Help with the NET_LOCK()

2017-02-01 Thread Hrvoje Popovski
On 31.1.2017. 21:35, David Hill wrote: > On Tue, Jan 31, 2017 at 09:11:37PM +0100, Alexander Bluhm wrote: >> On Tue, Jan 31, 2017 at 12:14:35PM -0500, David Hill wrote: >>> with mpi@'s suggestion to pass a struct mbuf * >> We call mbuf variables m and mbuf pointer mp. So you should rename >> *mp

Re: Help with the NET_LOCK()

2017-01-31 Thread Hrvoje Popovski
On 31.1.2017. 18:14, David Hill wrote: > On Tue, Jan 31, 2017 at 10:43:26AM +0100, Martin Pieuchot wrote: >> On 27/01/17(Fri) 14:33, David Hill wrote: >>> [...] >>> Forgot a file... Try this: >> Is it now possible to pass a 'struct mbuf *' instead of a 'struct mbuf **' >> to the pr_ctloutput()

Re: Help with the NET_LOCK()

2017-02-03 Thread Hrvoje Popovski
On 25.1.2017. 7:32, Martin Pieuchot wrote: > I just enabled the NET_LOCK() again and I'm looking for test reports. > Please go build a kernel from sources or wait for the next snapshot, > run it and report back. > > If you're looking for some small coding tasks related to the NET_LOCK() > just

Re: tcpdump(63969): syscall 54 "tty"

2017-01-24 Thread Hrvoje Popovski
On 24.1.2017. 19:03, Sebastien Marie wrote: > On Tue, Jan 24, 2017 at 03:32:25PM +0100, Hrvoje Popovski wrote: >> Hi all, >> >> every time when quitting tcpdump with ^C i see that log on console. >> Source is fetched few minutes ago ... >> >> Don't know is

Re: Help with the NET_LOCK()

2017-01-27 Thread Hrvoje Popovski
On 27.1.2017. 19:14, David Hill wrote: >> splassert: yield: want 0 have 1 >> Starting stack trace... >> yield() at yield+0xac >> pool_get() at pool_get+0x1ca >> m_get() at m_get+0x28 >> ip_ctloutput() at ip_ctloutput+0x4bf >> sogetopt() at sogetopt+0x7e >> sys_getsockopt() at sys_getsockopt+0xbf

Re: Help with the NET_LOCK()

2017-01-30 Thread Hrvoje Popovski
On 25.1.2017. 7:32, Martin Pieuchot wrote: > I just enabled the NET_LOCK() again and I'm looking for test reports. > Please go build a kernel from sources or wait for the next snapshot, > run it and report back. > > If you're looking for some small coding tasks related to the NET_LOCK() > just

Re: Help with the NET_LOCK()

2017-01-27 Thread Hrvoje Popovski
On 27.1.2017. 20:33, David Hill wrote: > On Fri, Jan 27, 2017 at 08:09:36PM +0100, Hrvoje Popovski wrote: >> On 27.1.2017. 19:14, David Hill wrote: >>>> splassert: yield: want 0 have 1 >>>> Starting stack trace... >>>> yield() at yield+0xac >>&g

Re: Help with the NET_LOCK()

2017-01-25 Thread Hrvoje Popovski
On 25.1.2017. 7:32, Martin Pieuchot wrote: > I just enabled the NET_LOCK() again and I'm looking for test reports. > Please go build a kernel from sources or wait for the next snapshot, > run it and report back. > > If you're looking for some small coding tasks related to the NET_LOCK() > just

Re: Test wanted: IPv4 forwarding w/o KERNEL_LOCK()

2017-02-23 Thread Hrvoje Popovski
On 23.2.2017. 13:24, Alexander Bluhm wrote: > On Wed, Feb 22, 2017 at 06:22:56PM +0100, Martin Pieuchot wrote: >> I'd appreciate if you could test this diff and report regressions. > > I did run the regression tests with it. Everything worked fine. > >> This cannot be tested if you're using

Re: Test wanted: IPv4 forwarding w/o KERNEL_LOCK()

2017-02-23 Thread Hrvoje Popovski
On 23.2.2017. 13:24, Alexander Bluhm wrote: > On Wed, Feb 22, 2017 at 06:22:56PM +0100, Martin Pieuchot wrote: >> I'd appreciate if you could test this diff and report regressions. > > I did run the regression tests with it. Everything worked fine. > >> This cannot be tested if you're using

Re: NET_LOCK() for bpf(4)

2017-01-24 Thread Hrvoje Popovski
On 24.1.2017. 10:59, Martin Pieuchot wrote: > ok? > > Index: net/bpf.c > === > RCS file: /cvs/src/sys/net/bpf.c,v > retrieving revision 1.158 > diff -u -p -r1.158 bpf.c > --- net/bpf.c 9 Jan 2017 19:15:01 - 1.158 > +++

tcpdump(63969): syscall 54 "tty"

2017-01-24 Thread Hrvoje Popovski
Hi all, every time when quitting tcpdump with ^C i see that log on console. Source is fetched few minutes ago ... Don't know is this good or bad so i'm sending it here .. OpenBSD 6.0-current (GENERIC.MP) #15: Tue Jan 24 15:09:53 CET 2017

Re: NET_LOCK() pr_sysctl

2017-01-16 Thread Hrvoje Popovski
On 16.1.2017. 23:53, Alexander Bluhm wrote: > Hrvoje Popovski has tested the diff and found some ugly > pmap_unwire: wiring for pmap 0xff00075f5210 va 0x7f7d5000 didn't > change! > kernel printfs. The happens when sysctl(8) writes a value. > > If oldp and newp are i

Re: NET_LOCK() take 2, tests wanted!

2017-01-20 Thread Hrvoje Popovski
On 20.1.2017. 3:04, Martin Pieuchot wrote: > Diff below enables the NET_LOCK(), again. > > What's new? > > - The lock ordering problem with fdplock() should now be fixed. It is >also documented, fdplock() should be taken first if both locks are >needed. > > - Page faults involving

Re: bigger mbuf clusters for sosend()

2016-08-21 Thread Hrvoje Popovski
On 13.8.2016. 10:59, Claudio Jeker wrote: > This diff refactors the uio to mbuf code to make use of bigger buffers (up > to 64k) and also switches the MCLGET to use M_WAIT like the MGET calls in > the same function. I see no point in not waiting for a cluster and instead > chain lots of mbufs

Re: bigger mbuf clusters for sosend()

2016-08-22 Thread Hrvoje Popovski
On 22.8.2016. 8:20, David Gwynne wrote: > >> On 22 Aug 2016, at 03:36, Hrvoje Popovski <hrv...@srce.hr> wrote: >> >> On 13.8.2016. 10:59, Claudio Jeker wrote: >>> This diff refactors the uio to mbuf code to make use of bigger buffers (up >>> to 64k) an

Re: MSI-X support

2016-09-01 Thread Hrvoje Popovski
On 4.5.2016. 16:32, Mark Kettenis wrote: >>> This is great, thanks for doing this! I'm a bit surprised that >>> we don't need to the same suspend/resume dance in ppb(4) as with >>> MSI. >>> >> >> That is an excellent point I overlooked. Kettenis, do we? > > Almost certainly. I committed the

Re: network performance fun

2016-10-24 Thread Hrvoje Popovski
On 24.10.2016. 23:36, Mike Belopuhov wrote: > On Mon, Oct 24, 2016 at 19:04 +0200, Hrvoje Popovski wrote: >> Hi all, >> >> OpenBSD box acts as transit router for /8 networks without pf and with >> this sysctls >> >> ddb.console=1 >>

Re: network performance fun

2016-10-24 Thread Hrvoje Popovski
On 25.10.2016. 0:22, Hrvoje Popovski wrote: > On 24.10.2016. 23:36, Mike Belopuhov wrote: >> On Mon, Oct 24, 2016 at 19:04 +0200, Hrvoje Popovski wrote: >>> Hi all, >>> >>> OpenBSD box acts as transit router for /8 networks without pf and with &g

network performance fun

2016-10-24 Thread Hrvoje Popovski
Hi all, OpenBSD box acts as transit router for /8 networks without pf and with this sysctls ddb.console=1 kern.pool_debug=0 net.inet.ip.forwarding=1 net.inet.ip.ifq.maxlen=8192 netstat 11/8 192.168.11.2 UGS0 114466419 - 8 ix0 12/8 192.168.12.2

Re: Intel 10GbE (ix) driver update - Looking for tests

2016-11-17 Thread Hrvoje Popovski
On 16.11.2016. 23:04, Mike Belopuhov wrote: > Hi, > > I've done a massive update of our ix(4) driver that brings > support for X550 family of controllers including those > integrated into new Xeon chips as well as QSFP support for > X520 (82599) but this needs thorough testing. If you're > using

Re: request for test: mfii

2016-10-28 Thread Hrvoje Popovski
On 26.10.2016. 5:50, YASUOKA Masahiko wrote: > On Wed, 26 Oct 2016 10:26:19 +1100 > Jonathan Gray wrote: >> On Tue, Oct 25, 2016 at 05:29:55PM +0900, YASUOKA Masahiko wrote: >>> I'm working on making mfii(4) bio(4) capable. >>> >>> If you have a machine which has mfii(4), I'd like

Xeon-D R2PCIe Agent pcidevs

2016-12-08 Thread Hrvoje Popovski
Hi all, i have this supermicro box: https://www.supermicro.nl/products/motherboard/Xeon/D/X10SDV-TP8F.cfm dmesg without this patch shows : vendor "Intel", unknown product 0x6f34 (class DASP subclass Time and Frequency, rev 0x03) at pci12 dev 16 function 1 not configured with this patch: "Intel

E5v4 pciids

2016-12-15 Thread Hrvoje Popovski
Hi all, patch in attachment adds some E5v4 pciids that i'm seeing in supermicro 1018R-WR box with E5-1650 v4. before patch: vendor "Intel", unknown product 0x6f20 (class system subclass miscellaneous, rev 0x01) at pci0 dev 4 function 0 not configured vendor "Intel", unknown product 0x6f21

Re: Xeon-D R2PCIe Agent pcidevs

2016-12-11 Thread Hrvoje Popovski
On 8.12.2016. 14:32, Hrvoje Popovski wrote: > Hi all, > > i have this supermicro box: > https://www.supermicro.nl/products/motherboard/Xeon/D/X10SDV-TP8F.cfm > > dmesg without this patch shows : > vendor "Intel", unknown product 0x6f34 (class DASP subclass

Re: BFD: route get and route monitor

2017-01-12 Thread Hrvoje Popovski
On 23.12.2016. 16:57, Hrvoje Popovski wrote: > On 21.12.2016. 23:15, Sebastian Benoit wrote: >>> Hi, >>> >>> it seems that bfd is working with Force10 S4810 and Extreme Networks >>> x460 switches. I can test it with cisco c6k5 if you want? >> >&g

Re: pfkey vs splsoftnet()

2017-01-13 Thread Hrvoje Popovski
On 12.1.2017. 18:27, Hrvoje Popovski wrote: > On 12.1.2017. 16:20, Martin Pieuchot wrote: >> On 10/01/17(Tue) 10:37, Martin Pieuchot wrote: >>> In pfkey_sendup() we call sorwakeup() which asserts for NET_LOCK(), so >>> we are already at IPL_SOFTNET. >>&

Re: pfkey vs splsoftnet()

2017-01-12 Thread Hrvoje Popovski
On 12.1.2017. 16:20, Martin Pieuchot wrote: > On 10/01/17(Tue) 10:37, Martin Pieuchot wrote: >> In pfkey_sendup() we call sorwakeup() which asserts for NET_LOCK(), so >> we are already at IPL_SOFTNET. >> >> pfkeyv2_send() is called via pfkey_output() which is also called with >> the NET_LOCK()

Re: BFD: route get and route monitor

2016-12-21 Thread Hrvoje Popovski
On 17.12.2016. 14:05, Peter Hessler wrote: > Updated output, requested by Theo. A normal get will show just the bfd > state, use "-bfd" to get all of the information. > > OK? > > $ route -n get 203.0.113.9 >route to: 203.0.113.9 > destination: 203.0.113.9 >mask: 255.255.255.255 >

Re: BFD: route get and route monitor

2016-12-23 Thread Hrvoje Popovski
On 21.12.2016. 23:15, Sebastian Benoit wrote: >> Hi, >> >> it seems that bfd is working with Force10 S4810 and Extreme Networks >> x460 switches. I can test it with cisco c6k5 if you want? > > Hei, > > i'm sure phessler (who might not read this for a couple of days) is happy > about any test you

Re: Reducing NET_LOCK() contention

2017-08-02 Thread Hrvoje Popovski
On 2.8.2017. 11:00, Alexandr Nedvedicky wrote: > Hello, > > On Wed, Aug 02, 2017 at 10:10:51AM +0200, Martin Pieuchot wrote: >> On 18/07/17(Tue) 15:55, Martin Pieuchot wrote: >>> When forwarding a lot of traffic with 10G interfaces contention on the >>> NET_LOCK() is "visible". Each time you

Re: Reducing NET_LOCK() contention

2017-08-02 Thread Hrvoje Popovski
On 2.8.2017. 10:10, Martin Pieuchot wrote: > On 18/07/17(Tue) 15:55, Martin Pieuchot wrote: >> When forwarding a lot of traffic with 10G interfaces contention on the >> NET_LOCK() is "visible". Each time you type "ifconfig" you can go grab >> a coffee... >> >> The problem has a name:

Re: NET_LOCK() w/o argument

2017-08-11 Thread Hrvoje Popovski
On 11.8.2017. 19:56, Martin Pieuchot wrote: > Two weeks ago I remove the splsoftnet()/splx() dance inside the > NET_LOCK(). Turns out we found a single bug, a missing splx() > in net/if_spppsubr.c. > > I believe it's time to move forward and completely remove the > argument. This will allow us

Re: 1M routes or 1M arp entries

2017-08-14 Thread Hrvoje Popovski
On 14.8.2017. 16:48, Simon Mages wrote: > Hi, > > you may want to take a look into /etc/login.conf > login.conf(5), cap_mkdb(1) > > In this file you can fiddle with you limit maxima > for login classes. > > BR > Simon > Thank you, i will do that ...

1M routes or 1M arp entries

2017-08-14 Thread Hrvoje Popovski
Hi all, when openbsd imports cca 1M routes or more and if i want to see them with "netstat -rn" i'm getting "Cannot allocate memory". bgpd can see all routes. i don't think that this is real problem but full bgp table is cca 700K routes. # bgpctl show ip bgp mem RDE memory statistics 1245184

Re: 1M routes or 1M arp entries

2017-08-14 Thread Hrvoje Popovski
On 14.8.2017. 16:03, Alexander Bluhm wrote: > On Mon, Aug 14, 2017 at 03:52:56PM +0200, Hrvoje Popovski wrote: >> # netstat -rnf inet >> netstat: Cannot allocate memory > > Have you tried to increase ulimit -d ? it seems that i can decrease it but not increase it, or i

/etc/rc: kernel relinking failed

2017-07-18 Thread Hrvoje Popovski
Hi all, i have checkout cvs tree few minutes ago and i'm seeing this log. Jul 18 22:47:36 x3550m4 /etc/rc: kernel relinking failed; see /usr/share/compile/GENERIC.MP/relink.log here it is: http://kosjenka.srce.hr/~hrvoje/zaprocvat/relink.log OpenBSD 6.1-current (GENERIC.MP) #8: Tue Jul 18

Re: /etc/rc: kernel relinking failed

2017-07-18 Thread Hrvoje Popovski
On 18.7.2017. 22:59, Theo Buehler wrote: > On Tue, Jul 18, 2017 at 10:55:59PM +0200, Hrvoje Popovski wrote: >> Hi all, >> >> i have checkout cvs tree few minutes ago and i'm seeing this log. >> >> Jul 18 22:47:36 x3550m4 /etc/rc: kernel relinking failed; see

Re: serial console and ddb

2017-07-03 Thread Hrvoje Popovski
On 3.7.2017. 23:42, Stuart Henderson wrote: > The phrase "break sequence" is often used, but it's a bit of a misnomer. > When a serial port is connected but not actively transmitting data the tx > line is usually held high. A "break" is when that line is low for more > than a frame duration (the

  1   2   3   >