Switch to DuckDuckGo Search
   April 11, 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 | >

Toggle Join/Part | bottom
[00:07:06] *** Kurlon <Kurlon!~Kurlon@98.13.72.207> has joined #illumos
[00:24:41] *** merzo <merzo!~merzo@net-31-156-168-27.cust.vodafonedsl.it> has joined #illumos
[00:30:44] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has quit IRC (Ping timeout: 250 seconds)
[00:36:55] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has joined #illumos
[00:47:40] *** polishdub <polishdub!~polishdub@207.86.38.254> has quit IRC (Quit: leaving)
[00:49:20] *** snuff-work <snuff-work!~snuff-wor@202-161-112-134.tpgi.com.au> has joined #illumos
[00:58:22] *** merzo <merzo!~merzo@net-31-156-168-27.cust.vodafonedsl.it> has quit IRC (Ping timeout: 246 seconds)
[00:59:48] *** merzo <merzo!~merzo@151.95.106.110> has joined #illumos
[01:22:27] *** andy_js <andy_js!~andy@94.6.62.238> has quit IRC (Quit: andy_js)
[02:31:41] *** kev009_ <kev009_!~kev009@ip72-222-200-117.ph.ph.cox.net> has joined #illumos
[02:32:14] *** gitomat <gitomat!~nodebot@165.225.148.18> has quit IRC (Remote host closed the connection)
[02:32:15] *** myrkraverk <myrkraverk!~chatzilla@unaffiliated/myrkraverk> has quit IRC (Read error: Connection reset by peer)
[02:32:15] *** kev009 <kev009!~kev009@ip72-222-200-117.ph.ph.cox.net> has quit IRC (Read error: Connection reset by peer)
[02:32:15] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has quit IRC (Remote host closed the connection)
[02:32:15] *** jessfraz_ <jessfraz_!~jessfraz@unaffiliated/jessfraz> has quit IRC (Quit: ZNC - http://znc.in)
[02:32:15] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has quit IRC (Ping timeout: 245 seconds)
[02:32:15] *** ChrisJames02170 <ChrisJames02170!~py@169.48.236.23.bc.googleusercontent.com> has quit IRC (Ping timeout: 245 seconds)
[02:32:20] *** jessfraz <jessfraz!~jessfraz@unaffiliated/jessfraz> has joined #illumos
[02:32:23] *** jessfraz is now known as jessfraz_
[02:32:28] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #illumos
[02:32:28] *** gitomat <gitomat!~nodebot@165.225.148.18> has joined #illumos
[02:32:32] *** ed209 <ed209!~ed209@165.225.128.67> has quit IRC (Ping timeout: 245 seconds)
[02:32:39] *** zerick <zerick!~zerick@irc.quassel.zerick.io> has quit IRC (Quit: No Ping reply in 180 seconds.)
[02:32:49] *** jocthbr <jocthbr!~salci@138-122-44-143.host.cicloti.com.br> has quit IRC (Quit: No Ping reply in 180 seconds.)
[02:32:57] *** aaro <aaro!aaro@gateway/shell/xshellz/x-lrqvmaucjunsvqog> has quit IRC (Ping timeout: 245 seconds)
[02:33:13] *** myrkraverk_ <myrkraverk_!~chatzilla@180.190.48.7> has joined #illumos
[02:33:46] *** zerick <zerick!~zerick@45.32.164.116> has joined #illumos
[02:34:06] *** jocthbr <jocthbr!~salci@138-122-44-143.host.cicloti.com.br> has joined #illumos
[02:34:17] *** aaro <aaro!aaro@gateway/shell/xshellz/x-znypoongsbqyiubb> has joined #illumos
[02:35:05] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has joined #illumos
[02:36:41] *** rsully <rsully!~rsully@unaffiliated/rsully> has quit IRC (Quit: rsully)
[03:08:16] *** arekinath <arekinath!~arekinath@pdpc/supporter/professional/arekinath> has quit IRC (Remote host closed the connection)
[03:08:16] *** melloc <melloc!~codytimec@timecube.club> has quit IRC (Remote host closed the connection)
[03:20:07] *** ed209 <ed209!~ed209@165.225.128.67> has joined #illumos
[03:27:35] *** arekinath <arekinath!~arekinath@pdpc/supporter/professional/arekinath> has joined #illumos
[03:27:55] *** melloc <melloc!~codytimec@timecube.club> has joined #illumos
[04:02:43] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[04:36:14] *** merzo <merzo!~merzo@151.95.106.110> has quit IRC (Ping timeout: 268 seconds)
[06:03:32] *** freakazoid0223 <freakazoid0223!~matt@pool-108-52-159-210.phlapa.fios.verizon.net> has quit IRC (Remote host closed the connection)
[06:27:05] *** jimklimov1 <jimklimov1!~jimklimov@ip-86-49-245-88.net.upcbroadband.cz> has quit IRC (Quit: Leaving.)
[07:36:13] <tsoome> :)
[07:38:26] <tsoome> I am known (by practical examples) about making mistakes from time to time.
[07:41:26] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has joined #illumos
[07:46:51] *** neuroserve <neuroserve!~toens@195.71.113.124> has joined #illumos
[07:54:08] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Quit: tsoome)
[08:40:35] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[09:12:37] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has quit IRC (Quit: Leaving)
[09:13:04] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has joined #illumos
[09:13:13] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has quit IRC (Remote host closed the connection)
[09:19:54] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has joined #illumos
[09:23:59] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[09:25:57] *** psarria <psarria!~psarria@142.red-79-146-98.dynamicip.rima-tde.net> has joined #illumos
[09:36:56] *** AllanJude <AllanJude!~allan@freebsd/developer/AllanJude> has quit IRC (Read error: Connection reset by peer)
[09:37:31] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has quit IRC (Quit: Bryson)
[09:38:12] *** AllanJude <AllanJude!~allan@freebsd/developer/AllanJude> has joined #illumos
[09:40:07] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has joined #illumos
[09:41:11] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has quit IRC (Client Quit)
[09:59:30] *** jimklimov1 <jimklimov1!~jimklimov@31.7.243.238> has joined #illumos
[09:59:42] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Ping timeout: 246 seconds)
[10:40:45] *** man_u <man_u!~manu@manu2.gandi.net> has joined #illumos
[10:53:44] *** jimklimov1 <jimklimov1!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[10:54:46] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[10:55:24] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[10:55:51] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[10:56:17] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[10:57:06] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[10:57:33] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[10:58:26] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[11:48:44] *** andy_js <andy_js!~andy@94.6.62.238> has joined #illumos
[11:57:40] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[11:58:20] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[11:58:42] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[11:59:35] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[11:59:57] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:00:35] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:01:01] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:01:35] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:07:18] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:07:45] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:08:07] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:09:10] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:09:51] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:10:40] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:12:54] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:13:25] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:13:47] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:14:20] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:14:39] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:15:00] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:15:19] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:16:12] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:16:38] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:18:12] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[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
[12:22:02] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:22:29] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[12:24:14] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[12:24:42] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[13:54:15] *** jcea <jcea!~Thunderbi@2001:41d0:1:8a82:7670:6e00:7670:6e00> has joined #illumos
[13:54:40] *** zeroSignal <zeroSignal!~root@fsf/member/zeroSignal> has quit IRC (Remote host closed the connection)
[13:55:00] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 250 seconds)
[14:11:29] *** psydroid <psydroid!psydroidma@gateway/shell/matrix.org/x-bmsyhrkskbqoucfd> has quit IRC (Ping timeout: 240 seconds)
[14:11:31] *** Ericson2314 <Ericson2314!ericson231@gateway/shell/matrix.org/x-gwbtptuixajhshhb> has quit IRC (Ping timeout: 252 seconds)
[14:12:02] *** jhot[m] <jhot[m]!jhotmatrix@gateway/shell/matrix.org/x-ijrthueuykwkgexx> has quit IRC (Ping timeout: 252 seconds)
[14:14:03] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has quit IRC (Remote host closed the connection)
[14:19:32] *** mno-hime <mno-hime!~mno-hime@94.142.238.232> has joined #illumos
[14:20:42] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[14:37:27] *** chrisBF <chrisBF!519d0514@gateway/web/freenode/ip.81.157.5.20> has joined #illumos
[14:45:56] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:46:26] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:46:48] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:47:26] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:48:23] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Read error: Connection reset by peer)
[14:48:58] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has joined #illumos
[14:48:58] *** freakazoid0223 <freakazoid0223!~matt@pool-108-52-159-210.phlapa.fios.verizon.net> has joined #illumos
[14:55:32] *** dgtlcmo <dgtlcmo!~dgtlcmo@c-24-61-203-74.hsd1.ma.comcast.net> has joined #illumos
[15:20:49] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has joined #illumos
[15:26:02] <jlevon> must stop looking for CTF GCC7 differences. sigh.
[15:27:07] *** gh34 <gh34!~textual@rrcs-162-155-144-114.central.biz.rr.com> has joined #illumos
[15:32:02] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[15:49:27] *** dgtlcmo <dgtlcmo!~dgtlcmo@c-24-61-203-74.hsd1.ma.comcast.net> has quit IRC (Ping timeout: 250 seconds)
[15:58:36] <andyf> I am going to abandon trying to fix nscd issues today... there are too many
[15:58:46] <andyf> jlevon, swap?
[16:00:15] *** polishdub <polishdub!~polishdub@ip98-168-33-198.ph.ph.cox.net> has joined #illumos
[16:00:23] <jlevon> uh no thanks
[16:11:29] *** dgtlcmo <dgtlcmo!~dgtlcmo@c-24-61-203-74.hsd1.ma.comcast.net> has joined #illumos
[16:13:40] <gitomat> [illumos-gate] 10701 Correct lock ASSERTs in vdev_label_read/write -- Olaf Faaland <faaland1 at llnl dot gov>
[16:17:32] <andyf> Does anyone have any idea why the if statement is here? - https://github.com/illumos/illumos-gate/blob/master/usr/src/cmd/nscd/nscd_nswstate.c#L502
[16:18:51] <jlevon> hmm that's a good one.
[16:18:59] <andyf> smatch is unhappy
[16:19:08] <jlevon> in what way?
[16:20:46] <andyf> we previously assumed 'base' could be null
[16:20:52] <andyf> base is used after the assert
[16:21:08] <jlevon> hmm it doesn't figure the assert
[16:21:59] <andyf> nope. I don't understand the if (base == NULL) though
[16:22:25] <jlevon> I wonder if there's a way to teach smatch about assert.
[16:22:46] <jlevon> regardless, that if makes no sense to me at all.
[16:24:49] <andyf> I'll remove it, smatch should be content then - for the wrong reason but at least the code is right
[16:28:49] <gitomat> [illumos-gate] 10686 Debug macros causes smatch issues -- John Levon <john.levon at joyent dot com>
[16:29:20] <andyf> thanks. smatch is really useful
[16:29:48] <jlevon> we're nearly there...
[16:51:17] *** neuroserve <neuroserve!~toens@195.71.113.124> has quit IRC (Ping timeout: 245 seconds)
[17:13:50] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has joined #illumos
[17:15:07] <andyf> I have nscd smatch clean, but it's going to be awful to review and test
[17:26:28] *** jimklimov <jimklimov!~jimklimov@31.7.243.238> has quit IRC (Ping timeout: 264 seconds)
[17:30:46] <rmustacc> andyf: Should I throw in CERRWARN whil we're at it?
[17:31:20] <rmustacc> I'm cleaning up both while I'm at it (usually)
[17:32:08] <andyf> yeah, I've done that too
[17:32:16] <andyf> and CTF
[17:32:25] <andyf> and move to 64-bit, which was the starting point
[17:32:35] <andyf> since we keep having problems with misaligned dereferences in libnss
[17:36:44] <igork> i see CTF update on joyent side for new ctf tools, but not on illumos. have joyent plans upstream updates to illumos or need additional updates?
[17:39:40] <jlevon> yes I'm planning to
[17:39:43] <jlevon> but there's more incoming yet.
[17:49:12] <rmustacc> andyf: If that's a problem, we should add the type checking to it so we can make sure that the types are the same between ILP32/LP64
[17:50:09] <jlevon> wonder if we can add a make rule for that.
[17:50:27] <tsoome> _typeof() :)
[17:51:25] <tsoome> rmustacc: do you happen to have any good ideas about errno value in https://www.illumos.org/rb/r/1661/ ?:)
[17:55:33] <rmustacc> tsoome: Let me circle back on that later today and I'll throw out some suggestions based on the semantics.
[17:55:54] <rmustacc> tsoome: Do you happen to know where we actually have implementations of those?
[17:55:57] <rmustacc> As I didn't find any in the gate.
[18:04:51] <tsoome> yes i do
[18:04:52] <tsoome> sec
[18:05:51] <tsoome> http://src.illumos.org/source/xref/gfx-drm/usr/src/uts/common/io/drm/drm_fb_helper.c#938
[18:06:00] <tsoome> there we get our only callback
[18:06:40] <tsoome> and function itself http://src.illumos.org/source/xref/gfx-drm/usr/src/uts/common/io/drm/drm_fb_helper.c#925
[18:18:50] <igork> what is plans for integrate 10601 Pool allocation classes ?
[18:18:57] <tsoome> but yes, as you can see, the actual error is well masked:)
[18:19:07] <igork> https://illumos.org/rb/r/1662
[18:21:07] <igork> danmcd: will you be superhero with it? :)
[18:22:03] <tsoome> still waiting https://www.illumos.org/rb/r/1226/ + https://www.illumos.org/rb/r/1572/ too,
[18:31:22] *** Bryson <Bryson!~anonymous@ip68-224-118-31.lv.lv.cox.net> has quit IRC (Quit: Bryson)
[18:45:26] <andyf> rmustacc: it isn’t that, it’s to do with stack alignment. Bugs in nss really but I have not had any luck with trying to get this fixed upstream. 64-bit nscd will make the problem go away there.
[18:45:46] <andyf> I can dig out an example when I’m back home if you’re interested.
[18:49:31] <gitomat> [illumos-gate] 10675 sdt: NULL pointer issues -- Toomas Soome <tsoome at me dot com>
[18:50:53] <rmustacc> andyf: Yes, we need to understand this because even if 64-bit nscd is there, that's not sufficient.
[18:51:06] <rmustacc> When nscd is disabled, libc will call into the backend directly to do name service resolution.
[18:51:13] <rmustacc> So you're not going to get out of having 32-bit nss usage.
[18:51:34] <igork> andyf: we are using 64bit nscd on dilos a long time
[18:51:59] <igork> and others tools too
[18:52:26] <igork> you can try take a look updates on tree
[18:52:35] <rmustacc> It doesn't really help the problem here, though.
[18:52:58] <rmustacc> Because nscd can be disabled. So if we have problems, they still exist regardless of the bitness.
[18:53:05] <rmustacc> bitness of nscd.
[18:53:32] <igork> i think we have issue with address alignment
[18:53:45] <igork> i can see strange issues on sparc
[18:54:19] <igork> we have found one issue with useradd, but probably it can be related to more global issue
[18:54:20] <rmustacc> jlevon: I do have a make rule for the type alignment stuff.
[18:54:33] <jlevon> oh.
[19:02:03] *** man_u <man_u!~manu@manu2.gandi.net> has quit IRC (Quit: man_u)
[19:14:50] *** fanta7531 <fanta7531!~fanta7531@p200300F76BC5AB006C935AFF232371E7.dip0.t-ipconnect.de> has joined #illumos
[19:29:47] *** lewellyn <lewellyn!~lewellyn@about/windows/lewellyn> has quit IRC (Read error: Connection reset by peer)
[19:32:22] <richlowe> I think you're talking about different NSSs
[19:32:41] <richlowe> at least, I followed a link in #omnios yesterday and briefly freaked out similarly until I noticed.
[19:34:04] <richlowe> oh dear, or our idea of "nss" calling into their idea of "nss" :\
[19:37:26] <jbk> what's a few duplicate TLAs between friends? :)
[19:39:30] <alanc> doesn't that happen with the ldap backend for nsswitch using the mozilla-based ldap code using the nss library for SSL support?
[19:39:38] <richlowe> alanc: precisely, yes.
[19:40:17] <richlowe> the top N for N ~= 40 frames are mozilla
[19:40:27] <richlowe> I didn't scroll far enough to get scared again.
[19:42:09] <richlowe> though reading bottom up, I'm already pretty scared when you get to:
[19:42:14] <richlowe> fe2dfd20 switcher+0xfd(deadbeed, fe2dfd48, b8, 0, 0, 806e295)
[19:43:20] <alanc> somehow, deadbeed is scarier than deadbeef, since it implies someone already manipulated the beef
[19:55:23] *** lewellyn <lewellyn!~lewellyn@about/windows/lewellyn> has joined #illumos
[19:55:23] *** ChanServ sets mode: +o lewellyn
[20:07:46] <andyf> Sorry, yes, Mozilla NSS. I’m out at present but will post more details later.
[20:09:18] <rmustacc> It doens't matter which NSS it is.
[20:09:29] <rmustacc> My point is that if it's being used by nscd, then libc will use it when nscd is down.
[20:10:21] <rmustacc> Because nscd only basically invokes the name service switch plugins on behalf of a program via a door call. So if nscd is using Mozilla NSS via a name service switch plugin, then libc will do the exact same thing.
[20:10:34] <rmustacc> So a 64-bit nscd doesn't alleviate the need to fix the underlying issue.
[20:11:22] <rmustacc> So the acronym confusion is fine, it, unfortunately, doesn't change things.
[20:14:21] <richlowe> rmustacc: right, when I first saw it I didn't see our nss was below it, I thought we were just all confused. Then I remembered the thing alanc brought up
[20:14:27] <richlowe> and found only I was confused
[20:14:35] <richlowe> but I'm not seeing why andy thinks it's alignment related
[20:16:07] <gitomat> [illumos-gate] 10676 fbt: NULL pointer errors -- Toomas Soome <tsoome at me dot com>
[20:18:09] <gitomat> [illumos-gate] 10677 systrace: NULL pointer errors -- Toomas Soome <tsoome at me dot com>
[20:20:28] <gitomat> [illumos-gate] 10678 dtrace: NULL pointer errors -- Toomas Soome <tsoome at me dot com>
[20:20:51] <richlowe> oh, wow
[20:20:53] <richlowe> someone's an asshole
[20:21:04] <richlowe> #define NAME_SERVICE_DOOR_COOKIE ((void*)(0xdeadbeef^NAME_SERVICE_DOOR_VERSION))
[20:21:18] <rmustacc> That seems.
[20:22:30] <jlevon> wut
[20:22:38] <jlevon> why
[20:23:21] <gitomat> [illumos-gate] 10679 dcpc: NULL pointer errors -- Toomas Soome <tsoome at me dot com>
[20:25:33] <gitomat> [illumos-gate] 10680 fasttrap: NULL pointer errors -- Toomas Soome <tsoome at me dot com>
[20:40:29] <andyf> @richlowe - when compiling libfreebl3.so (part of Mozilla NSS), gcc optimises memcpy() to use SSE instructions.
[20:40:44] <andyf> however, in 32-bit the stack is only 8-byte aligned
[20:41:58] <andyf> I'm just looking for the one of the places that this bit us
[20:42:25] <andyf> Here - https://github.com/nss-dev/nss/blob/471fcb11a6c7b8a7c8e49e609c72e8bc2a297397/lib/freebl/rijndael.c#L764
[20:42:40] <andyf> The newIV variable may not be aligned
[20:43:28] <andyf> movaps is the instruction that is used for the memcpy, and that requires 16-byte alignment
[20:44:05] <rmustacc> andyf: Well, the 8-byte aligned thing sounds off.
[20:44:17] <rmustacc> andyf: gcc/binutils changed the ABI a while back for SSE.
[20:44:24] <rmustacc> So it should be 16-byte aligned.
[20:44:33] <rmustacc> Maybe I missed that in a door callback?
[20:44:45] <rmustacc> But at least, the OS tries to give apps 16-byte alignment.
[20:44:51] <rmustacc> And gcc is definitely assuming that it's getting it.
[20:45:29] <andyf> If I add the force_align_arg_pointer attribute to that function, I can see that gcc adds a prologue to align the stack
[20:46:02] <rmustacc> Do you have a stack trace so we can see what's voilating that alignment?
[20:46:02] *** alanc <alanc!~alanc@129.157.69.49> has quit IRC (Remote host closed the connection)
[20:46:30] *** alanc <alanc!~alanc@129.157.69.49> has joined #illumos
[20:47:10] <rmustacc> This sounds like we missed something in illumos#3912 and illumos#6507.
[20:47:41] <rmustacc> And by extension the old Sun bug 6881217.
[20:47:58] <andyf> I'll see what I have in my chat history, can't find the core just now
[20:48:53] <rmustacc> OK.
[20:49:21] <richlowe> it sounds at least plausible the door path was missed
[20:49:22] <andyf> This first came up in October '17, but it looks we have another user being bitten by it now, just probably not rijndael
[20:49:28] *** polishdub <polishdub!~polishdub@ip98-168-33-198.ph.ph.cox.net> has quit IRC (Remote host closed the connection)
[20:49:28] <richlowe> doors are exciting.
[20:49:35] <richlowe> and sometimes hit you in the ass on your way out
[20:50:07] <rmustacc> I wonder if the door return magic means that we're not setting things up correctly.
[20:51:51] <andyf> From my notes here, the newIV variable in the function I linked to was at FE28A5C8 in the core
[20:52:03] <andyf> The thread's stack pointer is 0xfe28a5c8
[20:52:34] <richlowe> hotdog's stack is somewhwat more aligned than that.
[20:52:47] <richlowe> right?
[20:53:16] <andyf> The gcc6 release notes say: The option -mstackrealign is now automatically activated in 32-bit mode whenever the use of SSE instructions is requested.
[20:53:21] <andyf> but this is still gcc5
[20:53:39] <andyf> richlowe - I have not looked properly yet
[20:53:50] <rmustacc> Past me says in the i386 crt 'The stack needs to be 16-byte aligned with a 4-byte bias'
[20:55:33] <andyf> ok, I found the core
[20:56:08] <andyf> Stack: https://paste.ec/paste/OkF+LuJc#uIw7PHE-IVk3P4grB5Y2DM5Umf1o5dYV9W9sQQOkfts
[20:56:42] <rmustacc> andyf: Do you have the stack of the main thread by chance available?
[20:57:16] <andyf> You may need to guide me...
[20:57:36] <rmustacc> You can just do '1::findstack -v' in mdb.
[20:57:52] <andyf> [ 08047c68 libc.so.1`__pause+0x15() ]
[20:57:52] <andyf> 08047da8 main+0x910(8047d9c, fef5c6e8, 8047dd4, 805891f, 1, 8047de0)
[20:57:52] <andyf> 08047dd4 _start+0x83(1, 8047ebc, 0, 8047ecb, 8047edf, 8047ef7)
[20:57:55] <rmustacc> I'd expect things, if properly aligned, IIRC, to be at offset 8.
[20:57:56] <andyf> only three lines
[20:57:59] <rmustacc> That's fine.
[20:58:12] <richlowe> hotdogs stack is xxxxxxx0
[20:58:21] <rmustacc> So, the door return asm probably needs to make sure we're properly aligned.
[20:58:33] <rmustacc> Which means, it sounds like I should write a test for this and get it fixed in the door's ASM.
[20:58:40] <rmustacc> Or if someone else wants to.
[20:59:00] <rmustacc> But looks like based on my notes, we're not properly aligned when we get to switcher.
[20:59:05] <rmustacc> At which point everything else is bad.
[20:59:15] <rmustacc> Which means, probably door_return screwed up?
[20:59:46] <rmustacc> More specifically the __door_return asm routine.
[21:00:18] <rmustacc> Though I admit, I'd have to spend some time looking at how we create doors threads again.
[21:03:14] <andyf> ah, so door return could be creating a thread with a misaligned stack?
[21:03:25] <andyf> (sorry, still catching up)
[21:03:40] <rmustacc> Possibly.
[21:03:42] <richlowe> or creating a mis-aligned frame itself.
[21:04:09] <andyf> then it goes off into the library functions and eventually hits an SSE instruction that needs 16-byte alignment
[21:04:49] <rmustacc> Well, I'd say more pedantically the __door_return asm stub needs to end up going through and making sure that when it calls into other code it provides the proper alignment.
[21:05:00] <rmustacc> That's generally true of any asm function that does a call.
[21:05:12] <rmustacc> I suspect it's not right now.
[21:06:30] <rmustacc> Basically, at the call instruction it needs to be 16-byte aligned, which it looks like it's not there.
[21:06:36] <andyf> and of course gcc's -mstack-realign fixes it by adding `andl $0xfffffff0,%esp` etc. to each function prologue
[21:07:04] <rmustacc> Yes.
[21:07:22] <richlowe> rmustacc: ... the slop in door_return() sure is a thing.
[21:07:30] <richlowe> (the non-underscore C code)
[21:07:58] <rmustacc> Uh.
[21:08:02] <rmustacc> Well, that's a thing.
[21:09:48] <andyf> richlowe - yeah, hotdog's stack is a misaligned movaps too
[21:30:41] *** polishdub <polishdub!~polishdub@ip98-168-33-198.ph.ph.cox.net> has joined #illumos
[21:43:45] *** chrisBF <chrisBF!519d0514@gateway/web/freenode/ip.81.157.5.20> has quit IRC (Ping timeout: 256 seconds)
[22:03:07] *** dgtlcmo <dgtlcmo!~dgtlcmo@c-24-61-203-74.hsd1.ma.comcast.net> has quit IRC (Ping timeout: 240 seconds)
[22:20:01] *** 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
[22:23:13] *** idodeclare <idodeclare!~textual@cpe-76-185-177-63.satx.res.rr.com> has joined #illumos
[22:24:18] *** sjorge_be <sjorge_be!~sjorge@unaffiliated/sjorge> has joined #illumos
[22:27:09] *** sjorge <sjorge!~sjorge@unaffiliated/sjorge> has quit IRC (Ping timeout: 244 seconds)
[22:27:09] *** sjorge_be is now known as sjorge
[22:42:16] *** fanta7531 <fanta7531!~fanta7531@p200300F76BC5AB006C935AFF232371E7.dip0.t-ipconnect.de> has quit IRC (Quit: fanta7531)
[22:48:23] *** amrfrsh <amrfrsh!~Thunderbi@host-60054ecc6afffe842a0f.ip6.padnet.de> has joined #illumos
[22:49:22] *** gratto <gratto!~themsay@149.254.234.147> has joined #illumos
[22:49:41] *** dgtlcmo <dgtlcmo!~dgtlcmo@c-24-61-203-74.hsd1.ma.comcast.net> has joined #illumos
[22:52:18] *** gratto <gratto!~themsay@149.254.234.147> has quit IRC (Read error: Connection reset by peer)
[22:52:27] *** gratto <gratto!~themsay@149.254.234.147> has joined #illumos
[22:53:37] *** sarvet <sarvet!~sarvet@dslb-092-073-082-169.092.073.pools.vodafone-ip.de> has joined #illumos
[23:05:58] *** gh34 <gh34!~textual@rrcs-162-155-144-114.central.biz.rr.com> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)
[23:26:42] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has quit IRC (Ping timeout: 245 seconds)
[23:29:11] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #illumos
top

   April 11, 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 | >