Switch to DuckDuckGo Search
   October 24, 2019  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >

Toggle Join/Part | bottom
[00:00:11] <andyf> I can't believe somebody else has modified usr/src/cmd/cron/
[00:01:44] *** andy_js <andy_js!~andy@97e29e78.skybroadband.com> has quit IRC (Quit: andy_js)
[00:05:03] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has joined #illumos
[00:08:29] <jlevon> andyf: it calls it that but really it just means you need to rebase
[00:11:51] *** TwoADay <TwoADay!~hemi770@unaffiliated/hemi770> has quit IRC (Quit: Oh my god, it's full of stars!)
[00:12:07] *** TwoADay <TwoADay!~hemi770@208.79.89.189> has joined #illumos
[00:18:57] <andyf> So you learn to ignore it? :)
[00:20:03] <jlevon> yes
[00:21:00] *** CaptainTobin <CaptainTobin!~tobin@c-68-38-10-41.hsd1.in.comcast.net> has joined #illumos
[00:26:44] <gitomat> [illumos-gate] 11691 ptree could show service FMRIs -- John Levon <john.levon at joyent dot com>
[01:03:58] <Reinhilde> i assume i inspired somebody
[01:04:18] *** jlevon <jlevon!~jlevon@movementarian.org> has quit IRC (Quit: Coyote finally caught me)
[01:08:41] *** xenol <xenol!~xenol@2001:470:5bc7:1a6c::dead:beef> has quit IRC (Ping timeout: 246 seconds)
[01:15:43] *** jlevon <jlevon!~jlevon@movementarian.org> has joined #illumos
[01:22:20] *** xenol <xenol!~xenol@2001:470:5bc7:1a6c::dead:beef> has joined #illumos
[01:33:03] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Remote host closed the connection)
[01:34:04] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has joined #illumos
[01:38:09] <LeftWing> andyf: You're building a commit basically, in the system, which includes a specific parent as it usually does in git -- so it's just flagging that we can't update the master ref to point to your commit without a rebase basically
[01:38:38] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Ping timeout: 265 seconds)
[01:38:42] <LeftWing> Because it's configured to maintain a linear history like we do with the gate today
[01:38:54] <LeftWing> (no merge commits etc)
[01:45:44] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-npcepdwkfdthaoql> has quit IRC (Ping timeout: 268 seconds)
[02:01:15] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[02:01:28] *** tjg <tjg!tjg@gateway/vpn/privateinternetaccess/tjg> has quit IRC (Quit: Leaving)
[02:05:21] *** danmcd <danmcd!~danmcd@208.184.5.170> has quit IRC ()
[02:12:00] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-flwlpfmpgvjozbqu> has joined #illumos
[02:38:05] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-flwlpfmpgvjozbqu> has quit IRC (Ping timeout: 265 seconds)
[02:59:25] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-sdiynjuymchimatg> has joined #illumos
[03:10:42] <Reinhilde> i know this is an illumos channel and not a debian channel, but does anyone know if debian's netinst ISO has a serial console?
[03:11:56] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[03:16:52] *** patdk-lap <patdk-lap!~patrickdk@208.95.164.6> has quit IRC (Ping timeout: 264 seconds)
[03:17:24] *** patdk-lap <patdk-lap!~patrickdk@208.95.164.6> has joined #illumos
[03:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[03:20:07] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[03:28:22] <Smithx10> #include <byteswap.h> , Anything I should know? Stuck on trying to compile something
[03:30:25] <Reinhilde> i don't know
[03:30:28] <Reinhilde> what's the issue?
[03:31:26] <Smithx10> src/hll.c:25:22: fatal error: byteswap.h: No such file or directory
[03:34:22] <Smithx10> https://github.com/citusdata/postgresql-hll/blob/master/src/hll.c https://www.illumos.org/issues/1777
[04:00:54] <Reinhilde> would anyone happen to know what the illumos equivalent of FreeBSD's vmm.ko is?
[04:01:19] *** Guest94707 <Guest94707!jstoker@claire.jcs.me.uk> has quit IRC (Remote host closed the connection)
[04:05:57] <Reinhilde> nvm
[04:12:31] *** JStoker <JStoker!jstoker@unaffiliated/jstoker> has joined #illumos
[04:15:18] <Reinhilde> what would cause loading drv/vmm to not create a /devices/pseudo/vmm ?
[04:15:46] *** hadfl <hadfl!~hadfl@omniosce.ee.ethz.ch> has quit IRC (Quit: nn o/)
[04:15:46] *** andyf <andyf!~andyf@omniosce.ee.ethz.ch> has quit IRC (Quit: nn)
[04:16:12] *** andyf <andyf!~andyf@omniosce.ee.ethz.ch> has joined #illumos
[04:16:38] *** hadfl <hadfl!~hadfl@omniosce.ee.ethz.ch> has joined #illumos
[05:01:51] *** funnyguy_ <funnyguy_!funnyguy@2600:3c02::f03c:91ff:fe56:7f51> has quit IRC (Remote host closed the connection)
[05:07:15] <Reinhilde> surely someone would understand this gem.
[05:07:17] <Reinhilde> >open /dev/kvm: No such device or address
[05:12:35] *** funnyguy <funnyguy!~funnyguy@74.207.236.96> has joined #illumos
[05:25:14] <Reinhilde> LeftWing: does any of this ring a bell to you?
[06:30:42] <LeftWing> Reinhilde: Perhaps virtualisation isn't enabled (in the BIOS) or supported on this system?
[06:30:57] <LeftWing> Was there a message in the system log from kvm
[06:31:36] <LeftWing> Is this in your vultr guest? They might not do nested virtualisation
[06:41:19] <Reinhilde> LeftWing: the only message in the system log from kvm is as follows
[06:41:21] <Reinhilde> [perihelion ellenor]~ $ dmesg | grep kvm
[06:41:23] <Reinhilde> Oct 23 19:40:51 perihelion.nj.us.umbrellix.net pseudo: [ID 129642 kern.info] pseudo-device: kvm0
[06:41:25] <Reinhilde> Oct 23 19:40:51 perihelion.nj.us.umbrellix.net genunix: [ID 936769 kern.info] kvm0 is /pseudo/kvm@0
[06:41:27] <Reinhilde> [perihelion ellenor]~ $ grep kvm /var/adm/messages
[06:41:29] <Reinhilde> Oct 23 19:40:51 perihelion.nj.us.umbrellix.net pseudo: [ID 129642 kern.info] pseudo-device: kvm0
[06:41:31] <Reinhilde> Oct 23 19:40:51 perihelion.nj.us.umbrellix.net genunix: [ID 936769 kern.info] kvm0 is /pseudo/kvm@0
[06:41:33] <Reinhilde> [perihelion ellenor]~ $ grep kvm /var/log/syslog
[06:42:56] <LeftWing> What sort of machine is this
[06:43:02] <Reinhilde> and yes, this is in the vultr guest
[06:43:06] <LeftWing> Ah ok
[06:43:19] <Reinhilde> it should be throwing some other error
[06:43:25] <LeftWing> I wouldn't necessarily expect hardware virtualisation to be enabled in there
[06:43:51] <Reinhilde> I wouldn't either, but I would not expect the kernel modules to load at all were it not
[06:43:53] <Reinhilde> $ pfexec bhhwcompat -cv
[06:43:55] <Reinhilde> 32
[06:43:57] <Reinhilde> Cannot open /dev/vmmctl: No such file or directory
[06:44:12] <Reinhilde> someone with some idea of bhhwcompat's output would be able to tell me if that's a good or bad sign
[06:45:17] <Reinhilde> does oracle still sell sparc boxen?
[06:49:30] *** kujira <kujira!~null@mobile-166-177-251-248.mycingular.net> has joined #illumos
[06:56:27] <tsoome> yes
[06:58:01] <Reinhilde> tsoome: wtf
[06:58:40] <Reinhilde> LeftWing: should i email vultr and ask if nested virtualisation is available (i'll tell them it's so I can run a program that's not compatible with the 1st layer guest OS, because that's true)
[07:11:56] *** amrfrsh <amrfrsh!~Thunderbi@95.174.67.172> has quit IRC (Ping timeout: 240 seconds)
[07:16:10] <LeftWing> Reinhilde: Why not just create another guest with the target operating system?
[07:17:38] <Smithx10> LeftWing: any idea about compiling on illumos and trying to use #include <byteswap.h> ?
[07:17:57] <LeftWing> Do we have that file?
[07:18:07] <LeftWing> I think we have a different one
[07:18:10] <Smithx10> Been poking around the interwebs and i dont think so
[07:18:14] <Smithx10> https://github.com/Sereal/Sereal/issues/139
[07:18:25] <LeftWing> I feel like we have endian.h
[07:18:32] <Smithx10> There was a illumos issue
[07:18:39] <Smithx10> https://www.illumos.org/issues/1777
[07:19:11] <Smithx10> trying to compile https://github.com/citusdata/postgresql-hll/blob/master/src/hll.c this sucker
[07:19:39] <Smithx10> /msg LeftWing
[07:19:42] <Smithx10> ERRRrRr
[07:19:57] <LeftWing> I'm pretty sure we have an endian.h now
[07:20:02] <tsoome> we do.
[07:20:15] <rmustacc> I did it a while back and even said so on the ticket.
[07:20:22] <tsoome> however, as usual, please update your build hosts…
[07:20:37] <rmustacc> That wouldn't help this case.
[07:20:39] <Smithx10> I'm running into byteswap.h
[07:20:41] <rmustacc> They're using sys/endian.h.
[07:20:45] <rmustacc> Not endian.h.
[07:21:05] <rmustacc> You'll need to edit it to have an appropriate ifdef to #include <endian.h> on us.
[07:22:37] <rmustacc> There's no sys/endian.h since that'd be for the kernel (though I'd kind of like that over the current macros)
[07:24:12] <rmustacc> Smithx10: Does that make sense?
[07:24:29] <Smithx10> i can ifdef the ednian
[07:24:33] <Smithx10> https://gist.github.com/Smithx10/2b1b09aaf0aa29db4236ec00190132b3
[07:24:53] <Smithx10> Currently tho, can't find byteswap.h,
[07:24:58] <Smithx10> They are separate right?
[07:25:21] <rmustacc> The system is looking for one or the other.
[07:25:24] <Smithx10> I really need to learn C, and more about all of this
[07:25:26] <rmustacc> We don't have a byteswap.h.
[07:25:51] <rmustacc> It says on OS X do one thing, on others use sys/endian.h, on others do byteswap.h.
[07:25:58] <Smithx10> ahhh yes
[07:26:04] <Smithx10> I see now
[07:26:45] <rmustacc> Where everyone else doesn't actually have it either.
[07:27:18] <Smithx10> Stupid Question incoming, where do they get __OS__ from?
[07:27:39] <rmustacc> In the #ifdef sections?
[07:28:21] <Smithx10> are they passed in?
[07:28:31] <rmustacc> So, compiler configurations basically have a default set of macros that they define and values.
[07:28:54] <rmustacc> So it's basically from the compiler definitions for the system.
[07:29:07] <Reinhilde> LeftWing: i don't need more resources to do what I'm doing
[07:29:08] <Smithx10> so gcc grabs those at runtime?
[07:29:28] <rmustacc> Not really.
[07:29:38] <rmustacc> Most are hardcoded into a particular build of gcc.
[07:29:45] <rmustacc> Based on what gcc has been asked to target.
[07:30:21] <rmustacc> You can list all of them by running something like: 'gcc -dM -E - < /dev/null'
[07:31:04] *** CaptainTobin <CaptainTobin!~tobin@c-68-38-10-41.hsd1.in.comcast.net> has quit IRC (Ping timeout: 268 seconds)
[07:31:13] <yuripv> it doesn't looks like we provide bswap* in endian.h though? (looking at linux' byteswap.h contents)
[07:33:00] <Smithx10> so.... defined(__Illumos__) aint a thing
[07:33:08] <Smithx10> Guessing its sun or something from that output
[07:33:18] <arekinath> __sun works ok
[07:34:55] <rmustacc> yuripv: No, we don't.
[07:35:02] <rmustacc> It's not part of what it's supposed to define.
[07:35:18] <rmustacc> (At least based on what everyone else has)
[07:35:42] <rmustacc> So if it needs the actual bswap macros, that's not going to help.
[07:35:56] <Smithx10> rmustacc: that worked, but ran into errors around bswap
[07:36:03] <yuripv> well, I'm looking at byteswap.h in debian, and it has only bswap_* macros
[07:36:17] <rmustacc> OK.
[07:36:31] <rmustacc> Looks like it's just unconditional swapping?
[07:36:46] <Smithx10> https://gist.github.com/Smithx10/729357da686c9a3abe657df1e4f47363
[07:37:03] <rmustacc> Yes, we don't have the function there.
[07:37:14] <rmustacc> I misunderstood what it was doing based on the use of sys/endian.h on other systems.
[07:37:53] <Smithx10> ahhhh ok
[07:38:42] <Smithx10> no HLL for the short term
[07:38:52] <rmustacc> You're only using this on i386 right?
[07:38:55] <rmustacc> Erm, x86.
[07:39:18] <Smithx10> its a 64 bit smartos zone
[07:39:30] <rmustacc> You can work around it by doing a #define bswap_64 htobe64
[07:39:42] <rmustacc> Add that after the #include of endian.h.
[07:39:47] <rmustacc> That should give you a logical equivalent.
[07:40:07] <rmustacc> That'll basically always cause a swap.
[07:40:12] <rmustacc> As long as you're on x86.
[07:40:45] <Smithx10> brilliant
[07:40:53] <Smithx10> looks like it compiled
[07:41:03] <Smithx10> ill see if they have some tests I can run
[07:41:22] <Smithx10> Thank You so much!
[07:41:34] <rmustacc> Actually, there's probably a better way to do it.
[07:41:47] <rmustacc> Use the kernel sys/byteorder.h
[07:42:09] <rmustacc> And just do a #define bswap_64 to BSWAP_64
[07:42:28] <rmustacc> Maybe?
[07:44:14] <rmustacc> Dunno. Either should work for the time being and at least get you off the ground.
[07:44:43] <Smithx10> for sys/byteorder.h htats just a normal include correct?
[07:45:18] <rmustacc> Yes. As '#include <sys/byteorder.h>'
[07:46:37] <Smithx10> that built also
[07:46:57] <Smithx10> You suggest using the bswap64 BSWAP_64 and byteorder?
[07:47:28] <Smithx10> How did you know those things? Did you just look it up real qucik using like cscope or something?
[07:47:47] <rmustacc> I wrote the endian.h suppport?
[07:47:51] <Smithx10> hahahahaha
[07:48:00] <Smithx10> I figured it was you just knew it
[07:48:17] <Smithx10> hahhahahaha, how do you keep that all up there/?!$?!$
[07:48:22] <rmustacc> I use the sys/byteorder.h stuff for kernel drivers since it's all we have kernel side.
[07:48:25] <rmustacc> Not very well.
[07:49:20] <rmustacc> Proably should go and actually optimize all the endian.h stuff some day like we have the other bits.
[07:50:31] <Smithx10> Thanks again! I'm gonna get these changes added so we can have this build. I've been using Postgres on illumos, and the citusdb stuff looks like it could help our BUs out
[07:50:55] <Smithx10> Just gotta get it all compiling :) 1 more to go I think :P
[07:51:05] <rmustacc> If you hit other isues, please reach out.
[07:51:21] <rmustacc> Probably should add a byteswap.h.
[07:51:36] <Smithx10> yea, I'm hoping that if we become a customer of citusdb that they will test against illumos
[08:00:19] *** glasspelican <glasspelican!~quassel@2607:5300:201:3100::325> has quit IRC (Quit: http://quassel-irc.org - Chat comfortably. Anywhere.)
[08:02:16] *** glasspelican <glasspelican!~quassel@2607:5300:201:3100::325> has joined #illumos
[08:07:55] <jbk> IIRC if you -D_ASM_DEFINES (or something like that) it'll use the specific x86 instruction to do the swap instead of calling a function
[08:08:22] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[08:11:37] *** CaptainTobin <CaptainTobin!~tobin@c-68-38-10-41.hsd1.in.comcast.net> has joined #illumos
[08:24:21] *** kujira <kujira!~null@mobile-166-177-251-248.mycingular.net> has quit IRC ()
[08:24:29] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[08:25:36] *** patdk-lap <patdk-lap!~patrickdk@208.95.164.6> has quit IRC (Ping timeout: 265 seconds)
[08:26:08] *** patdk-lap <patdk-lap!~patrickdk@208.95.164.6> has joined #illumos
[08:44:24] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[09:21:18] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[09:27:38] *** v_a_b <v_a_b!~volker@p57A272B1.dip0.t-ipconnect.de> has joined #illumos
[10:05:24] <Reinhilde> is it weird that I think 1m should be kept around instead of using 8
[10:05:31] <Reinhilde> but instead of scrapping 8
[10:05:37] <Reinhilde> use 8 for stuff that only root would run
[10:05:46] <Reinhilde> and 1m for stuff that functions as intended as a normal user
[10:06:03] <Reinhilde> or i guess that'd be 1d (daemons) mostly
[10:12:18] <Reinhilde> jbk, rmustacc, LeftWing: on cron and @reboot - how would I name the enums and variables in cron.c that would be required to support this?
[10:17:07] *** lblume <lblume!~lblume@greenviolet/laoyijiehe/lblume> has joined #illumos
[10:22:52] <Reinhilde> i'm preliminarily calling it "timetype", the enum itself _service (given that @reboot is generally used for non-SMF-managed services), and the values CRON_TIME (for a minute hour daymonth month dayweek entry), CRON_REBOOT (for a @reboot entry), and CRON_CRONBOOT (for an entry type I want to create that will replicate the older behavior of @reboot - essentially being "@reboot without checking
[10:22:55] <Reinhilde> /var/run/cron.booted").
[10:26:30] *** edef <edef!edef@NixOS/user/edef> has quit IRC (Quit: leaving)
[10:26:51] *** edef <edef!edef@NixOS/user/edef> has joined #illumos
[10:27:30] *** andy_js <andy_js!~andy@97e29e78.skybroadband.com> has joined #illumos
[10:29:39] *** man_u <man_u!~manu@manu2.gandi.net> has joined #illumos
[10:37:17] *** amrfrsh <amrfrsh!~Thunderbi@185.212.171.68> has joined #illumos
[10:44:30] *** lblume <lblume!~lblume@greenviolet/laoyijiehe/lblume> has quit IRC (Quit: Leaving.)
[10:49:35] <Reinhilde> i'm also thinking of mangling the dayweek handling so that 7 can refer to sunday
[10:49:45] <Reinhilde> if indeed it is not already so mangled
[11:04:12] *** MarcelT <MarcelT!~marcel@tyhog-gua01a-216.cronon.net> has joined #illumos
[11:15:42] *** yuripv_ <yuripv_!~yuripv@176.113.72.212> has joined #illumos
[11:16:19] *** yuripv <yuripv!~yuripv@85.172.54.114> has quit IRC (Disconnected by services)
[11:16:22] *** yuripv_ is now known as yuripv
[11:17:01] <Reinhilde> how would I compile a tool independently of the rest of the gate?
[11:22:40] *** hugoAgogo <hugoAgogo!~hugoAgogo@148.252.250.12> has joined #illumos
[11:23:58] *** lblume <lblume!~lblume@greenviolet/laoyijiehe/lblume> has joined #illumos
[11:26:20] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[11:30:52] <Reinhilde> stupid question: has illumos totally freed itself yet? (reference to /opt/onbld/env/illumos)
[11:31:14] *** yuripv <yuripv!~yuripv@176.113.72.212> has quit IRC (Remote host closed the connection)
[11:37:59] <Reinhilde> i assume not
[11:39:33] <Reinhilde> jbk, rmustacc, LeftWing: do you know if the proprietary binaries (other than firmwares) are still technically necessary to create a working illumos distribution?
[11:43:18] <Reinhilde> so basically if someone wanted an all-libre illumos, can they have that in this day and age?
[11:44:56] *** yuripv <yuripv!~yuripv@184.170.246.167> has joined #illumos
[11:45:21] <jperkin> I think that's something that unleashed is working on
[11:45:33] <Reinhilde> jperkin: who or what is "unleashed"?
[11:45:49] <jperkin> https://unleashed-os.org
[11:46:57] <Reinhilde> ah
[11:49:44] <andyf> Reinhilde, since I happen to be working on cron just now, I can tell you exactly how to build it
[11:50:03] <Reinhilde> andyf: thanks. now please tell me how to build it :-)
[11:50:22] <andyf> What OS are you using?
[11:50:42] <Reinhilde> Omni
[11:50:53] <andyf> Do you have a .env file set up for building?
[11:51:07] <andyf> If not, grab the right one from https://downloads.omniosce.org/env/
[11:54:00] <Reinhilde> andyf: shit
[11:54:02] <Reinhilde> done
[11:54:12] <Reinhilde> pardon my saxon
[11:54:13] <andyf> Then change in to the top directory of your git checkout, and run
[11:54:44] <andyf> /opt/onbld/bin/bldenv /path/to/build-gate-on...env
[11:54:59] <andyf> That will get you into the build environment, hopefully with everything set up properly
[11:57:24] <Reinhilde> In the email sent by `nightly`, I get a tonne of crud like this: dmake: Warning: Command failed for target `pics/ctf_hash.o'
[11:58:06] <andyf> To just build a single component like cron, you don't need to use nightly
[11:58:11] <Reinhilde> k
[11:58:21] <andyf> use `bldenv` to drop into a build environment shell
[11:58:27] <andyf> which should say things like:
[11:58:44] <andyf> Build type is non-DEBUG
[11:58:47] <andyf> VERSION is gate-master-2cd405a338
[11:58:53] <Reinhilde> yes
[11:58:56] <andyf> ok..
[11:59:01] <andyf> now cd usr/src; dmake setup
[11:59:18] <andyf> that will build all of the tools you need, and set up an initial prototype area with headers etc.
[11:59:23] <Reinhilde> a dependency is clearly missing.
[11:59:36] <Reinhilde> what have i forgot to install? sh: line 1: /opt/gcc-7//bin/gcc: not found
[11:59:42] <andyf> pkg install illumos-tools
[12:08:02] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 240 seconds)
[12:14:29] <tsoome> jlevon: /code/illumos-gate/usr/src/tools/proto/root_i386-nd/opt/onbld/bin/i386
[12:14:30] <tsoome> 0000000000509a8d match_assign () + da
[12:14:48] <jlevon> a core?
[12:15:12] <tsoome> yep:)
[12:15:12] <jlevon> is it reproducible
[12:15:48] <tsoome> should be but I am not entirely sure - let me see...
[12:16:34] <tsoome> hm, i got it yesterday, and not from the nightly run I just did...
[12:17:41] <tsoome> nop, not getting it now...
[12:18:52] <jlevon> hrm
[12:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[12:20:07] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[12:30:30] *** hugoAgogo <hugoAgogo!~hugoAgogo@148.252.250.12> has quit IRC (Remote host closed the connection)
[12:32:36] <andyf> Reinhilde, still there? I'm off for lunch soon
[12:34:11] <Reinhilde> andyf: I'm slowly going insane, and my build system crapped out
[12:35:03] <andyf> The `dmake setup`?
[12:35:12] <Reinhilde> yes andyf
[12:35:25] <Reinhilde> i'm now running in -j1
[12:35:54] <andyf> right - I was about to suggest running `make setup` as that is serial make
[12:36:57] <Reinhilde> andyf: what's thedifference?
[12:37:29] <andyf> make and dmake are the same program, but dmake runs jobs in parallel
[12:38:00] <Reinhilde> https://umbrellix.net/~elleno/notreleasingw95yet.txt (the name is a reference to a William Gates quote that came to mind)
[12:38:03] <Reinhilde> errr..
[12:38:04] <andyf> although `dmake -m serial` and `make -m parallel` are possible, it's easier to just use dmake or make
[12:38:07] <Reinhilde> https://umbrellix.net/~ellenor/notreleasingw95yet.txt (the name is a reference to a William Gates quote that came to mind)
[12:40:30] <andyf> That's strange. If you haven't already, try exiting the build environment and going back in now that more tools are installed.
[12:43:42] <andyf> then maybe doing a complete clear out and setup again, like this:
[12:43:52] <andyf> cd usr/src; dmake clobber; dmake setup
[12:44:40] <andyf> Then, assuming that works, build any required libraries and then build cron itself, so I have been doing:
[12:44:47] <andyf> dmake -C lib/libcustr install
[12:44:53] <andyf> cd cmd/cron
[12:44:54] <andyf> make
[12:46:01] <andyf> and since you are looking at cron too, if you have time please can you review my change at https://www.illumos.org/rb/r/2407/ ?
[12:51:39] <Reinhilde> i'm a layperson and I don't believe my voucher would mean anything
[12:53:02] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:53:19] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:54:56] <Reinhilde> andyf: ... so this would be a new parser for something?
[12:55:10] <andyf> All reviews help, and we're both looking to make changes to cron
[12:55:33] <andyf> Have a look at the issue - https://www.illumos.org/issues/11858
[12:55:36] <Reinhilde> oh, it's step support
[12:57:35] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:58:10] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:58:43] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:59:09] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:59:52] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[13:00:41] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[13:01:05] <Reinhilde> i'm getting dizzy andyf
[13:01:59] <Reinhilde> but the code does LOOK good. i don't have an acc on the gerrit though
[13:05:01] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[13:06:57] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[13:06:58] *** amrfrsh <amrfrsh!~Thunderbi@185.212.171.68> has quit IRC (Ping timeout: 245 seconds)
[13:07:29] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[13:09:04] <tsoome> account you can create, that should not be too hard:D
[13:14:59] <andyf> That's reviewboard, and i'm pretty sure that both reviewboard and Gerrit use the same credentials as redmine (the issue tracking system).
[13:24:11] <tsoome> yes they do
[13:27:20] <Reinhilde> i made it hang on "in initialize"
[13:27:40] <tsoome> it?
[13:28:17] <Reinhilde> cron
[13:28:26] <tsoome> aa:)
[13:28:35] <tsoome> that can happen:)
[13:28:55] <tsoome> nowall the fun is there:)
[13:31:21] <Reinhilde> did i just kernel panic my system?
[13:31:44] <tsoome> by running cron?
[13:32:02] <Reinhilde> yes.
[13:32:31] <tsoome> see the panic message...
[13:32:39] <Reinhilde> there doesn't seem to have been a panic message?
[13:32:59] <tsoome> did it reboot?
[13:33:01] <Reinhilde> yes
[13:33:04] <Reinhilde> i think it did a clean reboot
[13:33:17] <Reinhilde> should've skipped the freaking bit after the @ lol
[13:33:23] <andyf> is it possible it somehow ended up running `reboot` ?
[13:33:29] <Reinhilde> andyf: yes, it did
[13:33:51] <Reinhilde> system/filesystem/local:default failed fatally
[13:34:00] <tsoome> run savecore and check /var/adm/messages
[13:34:03] <Reinhilde> great, now i'm in runlevel S
[13:35:00] <andyf> Run `zfs mount -a` to see if that explains where the problem is
[13:35:13] <Reinhilde> it's something related to my jacked zones configuration
[13:35:44] <tsoome> filesystem/local fails when it can not mount something.
[13:37:06] <Reinhilde> aye
[13:37:08] <Reinhilde> found that.
[13:54:05] *** mutin-s <mutin-s!~s-mutin@85.234.114.134> has quit IRC (Remote host closed the connection)
[13:54:30] *** mutin-s <mutin-s!~s-mutin@85.234.114.134> has joined #illumos
[14:03:12] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:03:52] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:04:16] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:04:52] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:05:14] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:07:15] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:07:55] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:08:20] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:08:49] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:09:20] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:09:51] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:10:45] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:11:16] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:11:20] <Reinhilde> andyf: if something comes in a one big commit format, what should i do?
[14:12:23] <andyf> I don't understand what you mean
[14:12:25] <Reinhilde> like all of the modifications i will be doing will come as one big patch
[14:12:32] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:12:52] <andyf> Ok...
[14:12:59] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:13:01] <Reinhilde> does that screw with the process?
[14:13:32] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:13:39] <andyf> Do you mean that there is more than one change but all in one patch?
[14:13:49] <andyf> (like @reboot, Sunday as 7, etc.)
[14:13:52] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:14:04] <Reinhilde> andyf: sunday as 7 has not been added. however, there is more than just @reboot
[14:14:22] <andyf> It's up to you then.
[14:14:30] <andyf> It is fine to have multiple changes in one patch
[14:14:38] <andyf> or you could separate them out
[14:14:55] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:14:57] <tsoome> keep it small, keep it simple. commits are cheap (or so they say;)
[14:15:00] <andyf> The thing that determines what is in a single commit is the testing
[14:15:32] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:15:55] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:16:22] <andyf> I personally prefer one change per commit, so agree with tsoome.
[14:16:28] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:16:56] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:16:59] <andyf> (The Sunday as 7 thing looks like a three-line change, so that might be a good one to do first and take through the review, test and integration process)
[14:17:33] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:18:35] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:18:39] <andyf> (well, three one-line changes, but same difference)
[14:18:56] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:19:15] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:19:34] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:19:37] <andyf> tsoome, yeah, commits are cheap; reviews are unfortunately not!
[14:19:49] <tsoome> we require reviews. you can not assume every dev knows every bit by heart - if you keep small functional changes, it is easier to follow the change.
[14:21:15] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:24:05] <Reinhilde> so if I'm going to be rolling several things into one patch, I should document everything that has been added
[14:26:08] *** MarcelT <MarcelT!~marcel@tyhog-gua01a-216.cronon.net> has quit IRC (Remote host closed the connection)
[14:40:12] <andyf> If you wanted to submit a single patch that did more than one thing, you would need to create an issue for each "thing" in redmine and put all of the issue numbers and descriptions in the commit message
[14:40:36] <andyf> but usually people would do each change separately, unless there is a reason why they need to be all tested together
[14:40:50] <andyf> You can always work on a big patch, then split it out later when you come to the review stage
[14:52:12] <Reinhilde> andyf: are separate @-commands separate things
[14:52:35] <andyf> Probably not.. but it is up to you
[15:12:13] <andyf> Do you mean @reboot, @annually, @daily etc.?
[15:13:00] <Reinhilde> yes
[15:13:13] <Reinhilde> I call them "vixie style cron commands"
[15:13:51] *** hugoAgogo <hugoAgogo!~hugoAgogo@148.252.250.12> has joined #illumos
[15:14:59] <andyf> Yeah, that would be nice, better compatibility with FreeBSD etc.
[15:15:19] <andyf> That has things like @300 too
[15:16:26] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[15:18:13] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has quit IRC (Ping timeout: 245 seconds)
[15:18:50] <Reinhilde> andyf: what would be @300
[15:19:35] <andyf> See https://www.freebsd.org/cgi/man.cgi?query=crontab&sektion=5
[15:23:33] *** hemi770 <hemi770!~hemi666@unaffiliated/hemi770> has quit IRC (Remote host closed the connection)
[15:23:51] <wiedi> what I'd really love cron to have would be a way to define fuzzyness. So if you have 10 cronjobs each in 50 zones they don't go off at the exact same moment. There are a few systems that can do this already in some form or another (freebsd cron, systemd timers, and I think even solaris). I should probably write this up as an IPD
[15:23:57] <andyf> But, to be honest, what we really need is a periodic SMF restarter..
[15:26:14] <tsoome> which is what oracle solaris did add:D
[15:26:35] <andyf> yes, I quite like it, and being able to provide cron fragments along with packages
[15:26:37] <yuripv> what does it do?
[15:31:12] *** hemi770 <hemi770!~hemi666@unaffiliated/hemi770> has joined #illumos
[15:38:22] <andyf> It's another SMF restarter, so you can have bits in SMF manifests that run jobs on a schedule
[15:38:37] <andyf> with the usual privilege bits, so as a user, with additional or fewer privileges
[15:44:39] <Reinhilde> cron as part of smf... tastes like a giant ball of oil
[15:44:40] <yuripv> sounds nice
[15:48:35] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[15:57:26] *** MarcelT <MarcelT!~marcel@tyhog-gua01a-216.cronon.net> has joined #illumos
[15:58:20] <Reinhilde> how would one format an FMRI for a third party service?
[16:03:24] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[16:04:58] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[16:05:16] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[16:08:50] *** jemershaw <jemershaw!~jemershaw@c-76-99-32-145.hsd1.pa.comcast.net> has joined #illumos
[16:16:57] *** vaxsquid <vaxsquid!~vaxsquid@znc.banana.fish> has quit IRC (Ping timeout: 240 seconds)
[16:21:48] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[16:25:56] *** vaxsquid <vaxsquid!~vaxsquid@znc.banana.fish> has joined #illumos
[16:40:42] <Reinhilde> what license are files generated by inetconv(1M) under?
[16:43:22] *** alanc <alanc!~alanc@inet-hqmc06-o.oracle.com> has quit IRC (Remote host closed the connection)
[16:43:49] *** alanc <alanc!~alanc@inet-hqmc06-o.oracle.com> has joined #illumos
[16:51:58] *** MarcelT <MarcelT!~marcel@tyhog-gua01a-216.cronon.net> has quit IRC (Remote host closed the connection)
[16:57:23] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[17:06:07] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[17:18:08] *** vaxsquid <vaxsquid!~vaxsquid@znc.banana.fish> has quit IRC (Ping timeout: 268 seconds)
[17:18:52] *** vaxsquid <vaxsquid!~vaxsquid@2a06:a006:f3:22::a67c> has joined #illumos
[17:25:22] *** mgerdts <mgerdts!~textual@208.184.5.170> has joined #illumos
[17:30:25] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[17:34:03] *** vaxsquid <vaxsquid!~vaxsquid@2a06:a006:f3:22::a67c> has quit IRC (Ping timeout: 245 seconds)
[17:37:35] *** mgerdts <mgerdts!~textual@208.184.5.170> has joined #illumos
[17:41:31] <KungFuJesus> yikes https://www.illumos.org/issues/11856
[17:47:28] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Quit: Leaving)
[17:50:02] *** freakazoid0223 <freakazoid0223!~matt@pool-96-227-98-169.phlapa.fios.verizon.net> has left #illumos ("Leaving")
[17:50:32] <KungFuJesus> the ZoL bug report of it has some more worrisome symptoms, like losing the ability to import or mount the pool
[17:50:37] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 265 seconds)
[17:57:43] <KungFuJesus> it doesn't seem like it's that hard to trigger with chown's or chmod's, either
[17:57:58] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[18:06:41] *** invoker14 <invoker14!viskulka@nat/redhat/x-yvwsswhevkhjfvgg> has joined #illumos
[18:09:21] <andyf> Ouch...
[18:10:16] <KungFuJesus> Correction, just chown's. Still something I do in production, though. I could be being bit by this and not even know it, with some stack corruption
[18:11:33] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 250 seconds)
[18:11:45] <KungFuJesus> the bug is facepalm-able when you look at the cause, a macro is hiding a count increment for indexing into a stack buffer
[18:12:10] <KungFuJesus> a very easy fix, and a very critical one, hopefully maintainers of the zfs bits get this in _quickly_
[18:13:10] *** BH23 <BH23!~BH23@santoroj.plus.com> has joined #illumos
[18:15:03] <dsockwell> big yikes
[18:15:32] <jimklimov> ...hello table meet the face, it's coming crashing once again...
[18:27:45] <LeftWing> KungFuJesus: If there's a patch you feel needs to go in, but hasn't gone in, you can always become a maintainer yourself :)
[18:34:25] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[18:37:53] *** vaxsquid <vaxsquid!~vaxsquid@znc.banana.fish> has joined #illumos
[18:37:59] *** man_u <man_u!~manu@manu2.gandi.net> has quit IRC (Ping timeout: 250 seconds)
[18:46:02] *** BH23 <BH23!~BH23@santoroj.plus.com> has quit IRC (Ping timeout: 240 seconds)
[18:48:09] *** danmcd <danmcd!~danmcd@208.184.5.170> has joined #illumos
[18:48:30] *** amrfrsh <amrfrsh!~Thunderbi@185.212.171.68> has joined #illumos
[18:59:03] *** jvl <jvl!~jvl@mei.hadret.com> has quit IRC (Quit: ZNC - http://znc.in)
[19:00:24] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has joined #illumos
[19:00:28] *** jvl <jvl!~jvl@mei.hadret.com> has joined #illumos
[19:02:38] *** yuripv <yuripv!~yuripv@184.170.246.167> has quit IRC (Remote host closed the connection)
[19:39:54] <KungFuJesus> LeftWing: This definitely fits in the category of "well duh" patches, that I fully comprehend but am without a doubt not knowledgable enough about ZFS' internals to call myself a maintainer
[19:40:25] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Quit: Leaving.)
[19:40:35] <KungFuJesus> I'd have to get a lot more intiment in that code base before I wanted to be a maintainer
[19:40:44] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[19:41:19] <rmustacc> One can contribute without being a 'maintainer', which I feels implies something more than necessairily has to be the case.
[19:41:27] <KungFuJesus> now, to be a contributor, I'm pretty sure that the ZoL patch can be merged into illumos-gate as is without any complication
[19:41:59] <KungFuJesus> see here: https://github.com/zfsonlinux/zfs/pull/8601/commits/3e3afa2517ec64ae95a37cb738eea05c12d9d304
[19:42:05] <LeftWing> Well you can definitely bring the patch over, maintain the original attribution, and put it up for review
[19:42:37] <KungFuJesus> I suppose. I'm surprised Jerry, who's on that ticket isn't doing that very thing right now
[19:42:49] <KungFuJesus> how do I do this? RB?
[19:43:42] <KungFuJesus> illumos-devel mailing list post pleading someone to do it for me?
[19:44:50] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Ping timeout: 240 seconds)
[19:46:15] <despair86> reinhilde: use manifold(1M), grab from PyPI
[19:46:37] <rmustacc> KungFuJesus: You can reach out to Jerry and see what's going on there. He may be working on it already.
[19:46:43] <Reinhilde> wark?
[19:47:08] <Reinhilde> despair86: what?
[19:47:36] <Reinhilde> despair86: i know inetconv is not an ideal way to get a service file
[19:51:38] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 240 seconds)
[19:53:22] <KungFuJesus> rmustacc: I feel a bit weird privately pinging him, can you just walk over to his desk and poke him?
[19:54:19] <rmustacc> I cannot.
[19:54:26] <KungFuJesus> it's a pretty serious bug with even more serious side effects (depending on the stack size)
[19:54:42] <rmustacc> I don't work with Jerry.
[19:55:20] *** mgerdts <mgerdts!~textual@208.184.5.170> has joined #illumos
[19:56:01] <KungFuJesus> You left Joyent, too?
[19:56:03] <rmustacc> I'm not really in a place to drive this bug and follow on in.
[19:56:30] <LeftWing> KungFuJesus: It's totally fine to reach out politely over mail to Jerry
[19:56:47] <LeftWing> Jerry's pretty awesome
[20:00:59] *** yuripv <yuripv!~yuripv@184.170.246.167> has joined #illumos
[20:01:28] <KungFuJesus> LeftWing: here goes
[20:03:40] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[20:11:07] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[20:54:48] <despair86> reinhilde: was referring to > how would one format an FMRI for a third party service?
[21:03:43] <rmustacc> Does anyone happen to have the AMD Zen 2 chips? The Ryzen 3XXX at all?
[21:36:30] *** pmooney <pmooney!~pmooney@67-4-175-230.mpls.qwest.net> has joined #illumos
[21:47:27] <copec> I have a lab with a few Ryzen 5 2400G CPUs for the kids
[21:49:23] <copec> Zen 1 based it looks like
[21:49:36] <copec> hence your 3XXX designation
[21:50:31] *** mgerdts <mgerdts!~textual@208.184.5.170> has joined #illumos
[21:51:12] * copec is of no help
[21:57:32] <rmustacc> No worries. Just wanted to test some stuff there.
[21:57:36] <rmustacc> thanks for taking a look.
[22:04:25] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Remote host closed the connection)
[22:05:19] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has joined #illumos
[22:15:14] <gitomat> [illumos-gate] 9531 Want netstat -u to show PIDs associated with sockets -- Andy Fiddaman <omnios at citrus-it dot co.uk>
[22:15:22] <LeftWing> andyf: Hooray!
[22:15:24] <rmustacc> Thanks for doing that andyf!
[22:15:35] <andyf> Yeah :) 2 years off and on
[22:16:07] <Smithx10> hahaha andyf thank god for that!
[22:16:19] <rmustacc> I guess next up is hyperv?
[22:16:33] <andyf> I do need to get back to that, yes.. I might have found a problem too
[22:16:43] <rmustacc> Uh oh, what's that?
[22:19:55] <andyf> It creates a set of message task queues, one per cpu
[22:20:00] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Remote host closed the connection)
[22:20:07] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[22:20:31] <andyf> and I think it expects the task to eventually run on the same CPU that called ddi_taskq_despatch
[22:20:59] <rmustacc> This is something that it uses to talk to hyperv?
[22:21:03] <rmustacc> Erm, the hypervisor?
[22:21:14] <andyf> it processes messages that were received
[22:21:18] <andyf> (from the hypervisor)
[22:21:30] <rmustacc> Ah, so a message is received from the hypervisor on a per-CPU core?
[22:21:34] <rmustacc> Erm, per-cpu queue
[22:23:39] <andyf> The drivers gets an interrupt, receives a message and schedules a task to deal with it
[22:23:54] <andyf> I have not untangled how it all hangs together yet
[22:24:02] <andyf> but on freebsd each taskq is bound to a specific CPU
[22:24:39] <andyf> and on illumos, there are just ncpus task queues
[22:24:40] <rmustacc> I think the thing we want to understand is if the reply CPU needs to be the same as the inbound CPU.
[22:24:53] <rmustacc> Which is what it sounds like?
[22:25:09] <andyf> exactly, and I don't yet know if it matters
[22:25:09] <rmustacc> Which hyperv driver is this in on FreeBSD?
[22:25:40] <andyf> however, the task uses wrmsr()
[22:25:47] <andyf> it's the vmbus one
[22:25:50] <rmustacc> Oh.
[22:26:36] *** invoker14 <invoker14!viskulka@nat/redhat/x-yvwsswhevkhjfvgg> has quit IRC (Ping timeout: 265 seconds)
[22:27:12] <andyf> so that's one more thing I need to work through before getting back to the other few things left in the storage driver
[22:27:34] <rmustacc> So... I expect it would. Since the wrmsr is exiting on a specific CPU.
[22:27:41] <andyf> yes
[22:27:58] <andyf> HEre's the task queue creation on freebsd
[22:27:59] <andyf> https://github.com/freebsd/freebsd/blob/master/sys/dev/hyperv/vmbus/vmbus.c#L932
[22:28:13] <andyf> I don't know if we have anything similar in terms of binding them to CPUs
[22:28:48] <rmustacc> I'm sure I can find something for you. ;)
[22:30:29] <rmustacc> OK, yes. So looking at them and a few others, we do need to do that.
[22:30:37] <rmustacc> Can you point me to where our code for this is?
[22:31:22] <andyf> https://github.com/omniosorg/illumos-omnios/blob/master/usr/src/uts/intel/io/hyperv/vmbus/vmbus.c#L846
[22:34:25] <andyf> and, coming back to netstat, thanks for all your help with the advice and reviews (@rmustacc)
[22:34:29] *** idx442 <idx442!~kvirc@142.204.28.26> has joined #illumos
[22:36:05] *** invoker14 <invoker14!~viskulka@50.225.197.140> has joined #illumos
[22:37:54] <idx442> hello, is anyone online?
[22:38:45] <alanc> https://lists.freedesktop.org/archives/mesa-announce/2019-October/000551.html - "notably a bunch of work to get Solaris and illumos working with mesa"
[22:40:28] <rmustacc> Are there things we can do to make that eaiser, alanc?
[22:40:52] <rmustacc> andyf: So we can do things here, it's just a not as simple as I'd like.
[22:41:59] <alanc> that was several years of accumulated portability fixes I dumped on them all at once - I bet pkgsrc & other package builders have been carrying similar patches around and not gotten them upstream yet
[22:42:17] <rmustacc> I see. Thanks for doing so.
[22:42:43] <rmustacc> andyf: We don't have an exact equivalent in the taskq API. Though we could probably add a taskq that had implicit CPU binding.
[22:42:53] <rmustacc> I can at least describe how I'd approach it.
[22:43:07] <alanc> and one was even a case where both you & us had added linux-compatible pthread_setname_np, but mesa had to be told that we had it, since it wasn't using configure or meson to check, but #ifdef's
[22:43:33] <rmustacc> Ah, well, thanks for including us in that.
[22:46:01] *** tjg <tjg!tjg@gateway/vpn/privateinternetaccess/tjg> has joined #illumos
[22:46:30] <jperkin> yeh we have 19.2.1 in pkgsrc with a few patches
[22:47:55] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[22:48:22] <alanc> it was mostly dcbaker from Intel who did the illumos side - both in meson & mesa - my part mostly just covered anywhere that #ifdef __sun or uname == SunOS got us both
[22:51:16] *** clapont <clapont!~clapont@unaffiliated/clapont> has quit IRC (Ping timeout: 240 seconds)
[22:52:07] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has quit IRC (Remote host closed the connection)
[22:53:10] <rmustacc> Gotcha.
[22:56:28] *** mgerdts <mgerdts!~textual@208.184.5.170> has joined #illumos
[22:59:30] *** Yogurt <Yogurt!~Yogurt@104-7-67-228.lightspeed.sntcca.sbcglobal.net> has joined #illumos
[23:00:10] *** invoker14 <invoker14!~viskulka@50.225.197.140> has quit IRC (Remote host closed the connection)
[23:00:28] *** invoker14 <invoker14!~viskulka@50.225.197.140> has joined #illumos
[23:02:35] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 246 seconds)
[23:07:06] *** idx442 <idx442!~kvirc@142.204.28.26> has quit IRC (Quit: KVIrc 5.0.0 Aria http://www.kvirc.net/)
[23:23:43] *** clapont <clapont!~clapont@unaffiliated/clapont> has joined #illumos
[23:24:14] <gitomat> [illumos-gate] 11826 Buffer overflow and memory leak in "libbsm" -- Matthias Scheler <matthias.scheler at wdc dot com>
[23:25:02] <Smithx10> Illumos should have an online store so I can order a t-shirt.
[23:25:14] *** bahamas10 <bahamas10!~dave@cpe-72-231-187-25.nycap.res.rr.com> has joined #illumos
[23:25:16] <Smithx10> lol
[23:30:00] *** danmcd <danmcd!~danmcd@208.184.5.170> has quit IRC ()
[23:36:49] <alanc> there's always https://www.cafepress.com/joyentsmartos
[23:39:40] <_mjg> how can i monitor interconnect traffic on illumos? (intel)
[23:39:46] <_mjg> something similar to pcm-numa.x
[23:44:03] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-141-249.rochester.res.rr.com> has joined #illumos
[23:44:33] <jlevon> huh that actually has the joyent division shirt
[23:52:23] *** mgerdts <mgerdts!~textual@208.184.5.170> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[23:54:06] <jbk> heh
top

   October 24, 2019  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >