Switch to DuckDuckGo Search
   April 23, 2020  
< | 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 | >


NOTICE: This channel is no longer actively logged.

Toggle Join/Part | bottom
[00:03:16] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has joined #illumos
[00:10:16] <richlowe> LeftWing: do the illumos.org email addresses in the gerrit notification mails actually work?
[00:21:29] *** andy_js <andy_js!~andy@90.218.209.121> has quit IRC (Quit: andy_js)
[00:21:50] *** ldepandis <ldepandis!~ldepandis@unaffiliated/ldepandis> has quit IRC (Quit: My MacBook has gone to sleep. ZZZzzz…)
[00:40:42] *** tsoome <tsoome!~tsoome@91.209.240.229> has quit IRC (Read error: Connection reset by peer)
[00:41:28] *** tsoome <tsoome!~tsoome@91.209.240.229> has joined #illumos
[00:43:34] <LeftWing> richlowe: If you mean "can I reply to the email" the answer is currently "no"
[00:45:51] <LeftWing> If you meant something else, can you clarify the question?
[01:05:18] <richlowe> LeftWing: I was surprised to see it writing a bunch of us into reply-to with @illumos.org addresses
[01:05:57] <LeftWing> Oh noi
[01:05:59] <LeftWing> no
[01:06:08] <LeftWing> That is ... not right
[01:06:26] <LeftWing> Good lord
[01:22:25] *** joltman <joltman!~znc@unaffiliated/joltman> has quit IRC (Ping timeout: 264 seconds)
[01:23:59] *** joltman <joltman!znc@gateway/vpn/privateinternetaccess/joltman> has joined #illumos
[02:13:01] *** lblume <lblume!~lblume@greenviolet/laoyijiehe/lblume> has quit IRC (Ping timeout: 272 seconds)
[02:18:14] *** neuroserve <neuroserve!~toens@ip-95-222-233-106.hsi15.unitymediagroup.de> has quit IRC (Ping timeout: 240 seconds)
[03:18:57] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-hrqklbeiopcoyhlk> has quit IRC (Read error: Connection reset by peer)
[03:24:22] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-nycsibqimjprfkan> has joined #illumos
[03:38:54] <jbk> heh.. i suspect ::kgrep is probably single threaded..
[03:40:36] *** roryrjb <roryrjb!~rory@209.250.250.222> has quit IRC (Ping timeout: 258 seconds)
[04:03:41] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[04:05:32] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-nycsibqimjprfkan> has quit IRC (Ping timeout: 265 seconds)
[04:10:04] *** liv3010m <liv3010m!~liv3010m@77-72-245-190.fibertel.com.ar> has quit IRC (Read error: Connection reset by peer)
[04:10:45] *** liv3010m <liv3010m!~liv3010m@77-72-245-190.fibertel.com.ar> has joined #illumos
[04:29:58] *** teutat3s <teutat3s!jhotmatrix@gateway/shell/matrix.org/x-iteefeefirgmkamb> has joined #illumos
[04:37:05] *** heroux <heroux!~heroux@507082E8.static.ziggozakelijk.nl> has joined #illumos
[04:42:18] *** heroux <heroux!~heroux@507082E8.static.ziggozakelijk.nl> has quit IRC (Ping timeout: 256 seconds)
[04:43:13] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-lwbluebzyidwsqoj> has joined #illumos
[04:47:58] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has quit IRC (Ping timeout: 256 seconds)
[05:05:58] <jbk> rmustacc: would you happen to know offhand if ixgbe chains a bunch of mblk_ts together (via b_cont) at some point in the tx path ?
[05:06:07] <jbk> (i know that's a bit of a strange question)
[05:27:06] <am11> fyi: i have upstreamed the libm-workaround PR: https://github.com/dotnet/runtime/pull/35325
[05:37:56] <Riastradh> am11: Why do you say that it's a bug in atan2, and not in the compiler's code generation of computing -x and passing the result as an argument atan2?
[05:42:18] <am11> Riastradh: what i understood was that it is either the bug in libm, or the compiler codegen. the situation got interesting/confusing when i tested with clang, which produces hard:0, soft:0 with and without `-O2`.
[05:42:56] <Riastradh> am11: How could it be a bug in libm, if the correct arguments (0,-0) are being passed in both cases?
[05:45:33] <am11> Riastradh: going by the test suite, only atan2 had this problem, and suite had many/all math functions covered.
[05:45:50] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has joined #illumos
[05:48:20] <Riastradh> am11: This looks to me like a code generation bug in the compiler -- have you examined the generated code, and/or confirmed that the same arguments are actually getting passed into libm as seen by the libm side?
[05:51:28] <am11> i posted the asm diffs yesterday (few hours ago), in this channel. i am not 100% sure whether it is a compiler bug or belongs to libm, as folks over at #gcc also indicated that it is not a compiler bug. but sure, it needs a bit more investigation to find the culprit.
[05:51:35] <Riastradh> If it is indeed a code generation bug, simply replacing the implementation of atan2 is unlikely to help -- unless perhaps the bug applies only to the compiler's _builtin_ atan2, in which case replacing the builtin atan2 by _any_ non-builtin atan2 (libm's, coreclr's, whatever) might be what works around it.
[05:52:08] <Riastradh> Got the link handy/
[05:52:09] <Riastradh> ?
[05:54:03] <am11> i don't have old chat log, and restarted my host system. i am collecting the diffs again, a moment.
[06:00:26] <am11> Riastradh: https://paste2.org/AJ7CUfJ4
[06:01:29] <Riastradh> Hm. In `hard', there is no call to atan2 at all!
[06:02:23] <Riastradh> Have you tried `volatile double x = -0.; ... atan2(0, x)'?
[06:04:16] <Riastradh> I see now how this is consistent with the hypothesis of a busted atan2 in libm: for atan2(0, -0.), the compiler correctly constant-folded the atan2 call so it doesn't query libm at all; for atan2(0, -x), it does actually call libm.
[06:05:54] <am11> https://paste2.org/bgAHhvcp
[06:06:10] <am11> (with printf call to keep the values from getting optimized out)
[06:08:33] <am11> ah, so would the fix be to improve compiler optimization (to cover the unary operator case) or fix libm function (or both)?
[06:08:53] <Riastradh> Right, so hard basically just does printf("%g\n", 3.14159265358979), while soft does printf("%g\n", atan2(0, -x)).
[06:09:05] <am11> this was from gcc7 on smartos 2020 (~14 days old build).
[06:09:23] <Riastradh> Try compiling hard with -fno-builtin=atan2?
[06:10:05] <am11> i can get results from clang9 on OI 2020 as well, which produces 0,0 for both cases: -O2 and without.
[06:10:14] <am11> trying wiht -fno-builtin=atan2.
[06:12:15] <Riastradh> (The constant .LC0 = .long 1413754136 .long 1074340347, or 0x400921fb54442d18, is 3.1415926535897931.)
[06:12:47] <am11> with `-fno-builtin`, gcc7 on smartos produces 0 for both hard and soft (same as noticed by clang9 on OI).
[06:12:55] <Riastradh> Right. OK, so that is indeed a libm bug!
[06:13:47] <Riastradh> To make the configure.cmake test more robust, so you don't have to patch it again later, you might say `volatile double x = 0.0' instead of just `double x = 0.0'.
[06:14:07] *** arnoldoree <arnoldoree!~arnoldore@113.210.125.66> has joined #illumos
[06:14:12] <am11> do we use a different libm sources than other operating systems?
[06:14:25] <Riastradh> There are lots of libms and libm variants floating around.
[06:14:40] <am11> that sounds familiar. :)
[06:15:40] <Riastradh> The atan2 in NetBSD's libm (which was, a long time ago, derived from Sun fdlibm) does the right thing.
[06:15:59] <am11> (recently experienced an oawk nightmare, where o stands for old aka pre-1985; and n in nawk is new, as in since 1985)
[06:16:05] <Riastradh> heh
[06:18:38] <Riastradh> It looks to me like NetBSD's atan2 has had the relevant logic since fdlibm was imported in 1994. So I dunno why SmartOS 2020 or OI would get it wrong!
[06:19:12] <Riastradh> I kind of assumed they would have been derived from the same source but maybe not.
[06:19:51] <Riastradh> (not sure offhand where the source for SmartOS libm or OI libm live)
[06:20:33] <jbk> http://src.illumos.org/source/xref/illumos-gate/usr/src/lib/libm/common/
[06:24:31] <Riastradh> If it's <https://github.com/illumos/illumos-gate/blob/master/usr/src/lib/libm/common/C/atan2.c>, that sure does look pretty different from fdlibm, and I don't see any obvious logic to handle signed-zero inputs in it.
[06:25:27] <Riastradh> (but there's a lot of bit-twiddling going on there so it's not obviously wrong)
[06:33:39] <am11> Riastradh: i pinged you on pr about `volatile`, I hope it is ok. :)
[06:40:16] <Riastradh> Sure, that's fine!
[07:08:25] *** BOKALDO <BOKALDO!~BOKALDO@81.198.156.19> has joined #illumos
[07:46:15] <jbk> rmustacc: to answer my own question.. yes (lso)
[08:03:24] <EisNerd> very good mail from josh, looks a bit opensolaris has forgotten what was already true, back in times of Sun xVM
[08:03:50] <mnrmnaugh> can somebody make a habit to ping my ass in here..
[08:04:39] <EisNerd> also process lockstepping is sth even system which where designed for has dropped, see NonStop
[08:04:45] <mnrmnaugh> oof gcc
[08:05:25] <mnrmnaugh> oh oawk, you say. i should retry that. nawk is my favorite thing ever
[08:05:36] <EisNerd> if illumos really likes some process level inter node HA, then think about a checkpoint API
[08:06:19] <EisNerd> and IPC based on named processes
[08:06:47] <mnrmnaugh> no, i think that wont go well
[08:07:10] <mnrmnaugh> to allegate to android
[08:07:12] <mnrmnaugh> lol
[08:11:27] <LeftWing> EisNerd: To the extent that xVM was Xen-based I think it was a bit of a dead end as well unfortunately
[08:25:16] *** pwinder <pwinder!~pwinder@86.11.191.8> has joined #illumos
[08:32:25] <tsoome> please review:) https://code.illumos.org/c/illumos-gate/+/573
[08:35:32] *** neuroserve <neuroserve!~toens@ip-95-222-233-106.hsi15.unitymediagroup.de> has joined #illumos
[08:41:10] *** andy_js <andy_js!~andy@90.218.209.121> has joined #illumos
[08:42:59] *** ldepandis <ldepandis!~ldepandis@unaffiliated/ldepandis> has joined #illumos
[08:53:39] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has quit IRC (Remote host closed the connection)
[08:53:52] <am11> oawk has a different dialect than nawk, so "modern" (1985 onwards) scripts do not work with it. nawk was like a python3 in awk world; syntactically incompatible. in the current OpenIndiane, awk is pointing to oawak in the default configuration (which is being fixed, many thanks!), so stuff like `sub()` doesn't work, basically nothing worked from
[08:53:52] <am11> awk as we know it. interestingly, `man awk` in that setup is newer than the actual (o)awk in PATH. :)
[08:54:06] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has joined #illumos
[08:54:22] <am11> OpenIndiana*
[08:57:01] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has quit IRC (Remote host closed the connection)
[08:58:14] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has joined #illumos
[09:02:29] <EisNerd> LeftWing: does kvm not offer live migration?
[09:02:47] <LeftWing> I don't believe the illumos port of KVM does, no
[09:03:02] <LeftWing> I mean, maybe somebody has patched it in, but if they have I am not aware of it
[09:03:04] <EisNerd> LeftWing: I had xVM in production, was very good
[09:03:14] <LeftWing> I'm sure it worked fine
[09:03:24] <LeftWing> But Xen is burdensome from a maintenance perspective
[09:03:35] <LeftWing> It's very big and it has tendrils everywhere
[09:03:45] *** arnoldoree <arnoldoree!~arnoldore@113.210.125.66> has quit IRC (Ping timeout: 265 seconds)
[09:04:19] <LeftWing> bhyve for instance is substantially less software to look after, but because we're now in the era where a lot of virtualisation is done for you by the silicon, it's feasible
[09:07:31] *** jimklimov <jimklimov!~jimklimov@ip-86-49-254-26.net.upcbroadband.cz> has quit IRC (Quit: Leaving.)
[09:10:17] *** lblume <lblume!~lblume@greenviolet/laoyijiehe/lblume> has joined #illumos
[09:13:44] *** lgtaube <lgtaube!~lgt@213-67-21-71-no84.tbcn.telia.com> has quit IRC (Ping timeout: 256 seconds)
[09:15:36] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has quit IRC (Ping timeout: 240 seconds)
[09:25:12] <wilbury> to compile sbcl (in oi/hipster) you need sbcl.
[09:28:16] *** basvdlei <basvdlei!~basvdlei@2001:980:a4c3:1:e6b9:7aff:fe9f:3993> has quit IRC (Quit: basvdlei)
[09:28:22] *** lgtaube <lgtaube!~lgt@213-67-21-71-no84.tbcn.telia.com> has joined #illumos
[09:28:28] *** basvdlei <basvdlei!~basvdlei@2001:980:a4c3:1:e6b9:7aff:fe9f:3993> has joined #illumos
[09:32:01] *** jimklimov <jimklimov!~jimklimov@78.80.224.132> has joined #illumos
[09:39:23] *** roryrjb <roryrjb!~rory@209.250.250.222> has joined #illumos
[09:40:20] *** jamtorus <jamtorus!~quassel@s91904423.blix.com> has joined #illumos
[09:43:46] *** jellydonut <jellydonut!~quassel@s91904423.blix.com> has quit IRC (Ping timeout: 256 seconds)
[09:52:45] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has quit IRC (Quit: ZNC 1.7.5 - https://znc.in)
[09:53:02] *** neirac <neirac!~neirac@pc-4-149-45-190.cm.vtr.net> has joined #illumos
[09:57:14] *** leoric <leoric!~leoric@46.147.99.67> has joined #illumos
[09:59:51] *** jamtorus is now known as jellydonut
[10:08:06] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has joined #illumos
[10:27:02] <sjorge> jbk: do you want to rereview https://code.illumos.org/c/illumos-gate/+/572 (pam_list) ? there were some changes based on feedback, I hope to send the RTI mail this weekend or maybe friday evening
[10:27:46] <sjorge> LeftWing I guess the CPU still needs to match and no passthru used
[10:28:10] <sjorge> Well I guess for the former you could just mask all the stuff that is not shared on all your CPU's in the cluster
[10:29:33] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has quit IRC (Ping timeout: 250 seconds)
[10:37:46] *** phyre <phyre!~phyre___@78.30.23.107> has joined #illumos
[11:47:15] *** amrfrsh <amrfrsh!~Thunderbi@134.19.189.92> has joined #illumos
[12:22:09] *** arnoldoree <arnoldoree!~arnoldore@2001:d08:2100:62cf:574c:2307:6b59:748b> has joined #illumos
[12:22:10] *** ypankov <ypankov!~ypankov@85.174.206.243> has quit IRC (Quit: leaving)
[12:45:10] *** ldepandis <ldepandis!~ldepandis@unaffiliated/ldepandis> has quit IRC (Read error: Connection reset by peer)
[13:45:29] *** phyre <phyre!~phyre___@78.30.23.107> has quit IRC (Remote host closed the connection)
[14:10:49] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has joined #illumos
[14:24:28] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has joined #illumos
[14:39:57] *** basvdlei <basvdlei!~basvdlei@2001:980:a4c3:1:e6b9:7aff:fe9f:3993> has quit IRC (Remote host closed the connection)
[15:07:04] *** am11 <am11!54f985dd@dsl-hkibng22-54f985-221.dhcp.inet.fi> has joined #illumos
[15:19:37] *** arnoldoree <arnoldoree!~arnoldore@2001:d08:2100:62cf:574c:2307:6b59:748b> has quit IRC (Ping timeout: 272 seconds)
[15:28:36] *** arnoldoree <arnoldoree!~arnoldore@2001:d08:2100:62cf:574c:2307:6b59:748b> has joined #illumos
[15:39:05] *** tsoome <tsoome!~tsoome@91.209.240.229> has quit IRC (Read error: Connection reset by peer)
[15:41:26] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has joined #illumos
[15:45:26] *** man_u <man_u!~manu@fob.gandi.net> has joined #illumos
[15:46:54] *** patdk-lap <patdk-lap!~patrickdk@208.94.189.75> has quit IRC (Remote host closed the connection)
[15:50:16] *** patdk-lap <patdk-lap!~patrickdk@208.94.189.75> has joined #illumos
[16:05:23] *** psarria <psarria!~psarria@37.red-79-146-98.dynamicip.rima-tde.net> has quit IRC (Ping timeout: 250 seconds)
[16:15:06] *** psarria <psarria!~psarria@37.red-79-146-98.dynamicip.rima-tde.net> has joined #illumos
[17:00:51] *** tsoome__ <tsoome__!~tsoome@91.209.240.229> has joined #illumos
[17:03:01] *** tsoome <tsoome!~tsoome@148-52-235-80.sta.estpak.ee> has quit IRC (Ping timeout: 250 seconds)
[17:03:01] *** tsoome__ is now known as tsoome
[17:03:47] *** Kurlon_ <Kurlon_!~Kurlon@cpe-67-253-136-97.rochester.res.rr.com> has quit IRC (Remote host closed the connection)
[17:18:40] *** jimklimov <jimklimov!~jimklimov@78.80.224.132> has quit IRC (Quit: Leaving.)
[17:36:15] *** leoric <leoric!~leoric@46.147.99.67> has quit IRC (Quit: Konversation terminated!)
[17:39:08] <rmustacc> jbk: So, ixgbe can certainly receive a packet from mac with a lot of entries set in b_cont. There is some minor manipulation of b_cont in the lso path, but otherwise, it doesn't itself normally create chains.
[17:47:39] *** Wraithh <Wraithh!ilkka@kapsi.fi> has joined #illumos
[17:48:40] *** jimklimov <jimklimov!~jimklimov@ip-86-49-254-26.net.upcbroadband.cz> has joined #illumos
[18:01:29] *** Kurlon <Kurlon!~Kurlon@167-192.gwi.net> has joined #illumos
[18:06:38] *** Kurlon <Kurlon!~Kurlon@167-192.gwi.net> has quit IRC (Remote host closed the connection)
[18:07:17] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[18:13:55] *** psarria <psarria!~psarria@37.red-79-146-98.dynamicip.rima-tde.net> has quit IRC (Ping timeout: 256 seconds)
[18:15:41] *** jcea <jcea!~Thunderbi@2001:bc8:2ecd:caed:7670:6e00:7670:6e00> has quit IRC (Quit: jcea)
[18:16:52] <jbk> that manipulation is a bit confusing :) took a bit last night
[18:17:07] <jbk> (I'm chasing down a loaned viona buffer that isn't being given back)
[18:17:25] <am11> jbk: is there a more recent revision of https://us-east.manta.joyent.com/jbk/public/lldb.tar.gz somewhere in `git`? i will be soon stepping into the zone where lldb is required for libsosplugin, however it is possible to analyze stack dumps with gdb, in retrospect (so, not a dead-end), but slows down the speed quite a bit.
[18:17:41] <jbk> that's old :)
[18:17:58] <jbk> or wait
[18:18:03] <am11> yup, it is still something :)
[18:18:06] <jbk> thinking of something else
[18:18:13] <jbk> that was just the patches i started on
[18:18:18] <jbk> woodstock is working on that
[18:18:25] <am11> saw it here https://github.com/joyent/pkgsrc-joyent/issues/35#issuecomment-387619834
[18:18:26] <jbk> since lldb has decided to not document anything
[18:18:46] <jbk> that doesn't help
[18:19:05] <jbk> and that's woefully incomplete
[18:19:24] <jbk> like the whole target interface in lldb isn't documented
[18:19:45] <jbk> so the best bet is to just wait until woodstock is able to finish his work
[18:20:06] <sjorge> jbk: I replied to your feedback on cr 572 with a question about size_t/long and calloc
[18:20:21] <toastersonerson1> LeftWing (IRC): I linked a patch for illumo-kvm live migration in the same mail thread .)
[18:22:34] *** psarria <psarria!~psarria@37.red-79-146-98.dynamicip.rima-tde.net> has joined #illumos
[18:48:12] *** lystra <lystra!~lystra@d53-64-11-169.nap.wideopenwest.com> has joined #illumos
[19:06:17] *** arnoldoree <arnoldoree!~arnoldore@2001:d08:2100:62cf:574c:2307:6b59:748b> has quit IRC (Quit: Leaving)
[19:06:37] *** pwinder <pwinder!~pwinder@86.11.191.8> has quit IRC (Quit: This computer has gone to sleep)
[19:08:10] *** man_u <man_u!~manu@fob.gandi.net> has quit IRC (Ping timeout: 256 seconds)
[19:18:25] *** Kurlon_ <Kurlon_!~Kurlon@167-192.gwi.net> has joined #illumos
[19:20:28] <tsoome> I have module file (not loaded into kernel), can I get information from it with mdb? like ::dis etc
[19:21:16] *** Kurlon_ <Kurlon_!~Kurlon@167-192.gwi.net> has quit IRC (Remote host closed the connection)
[19:21:42] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has quit IRC (Ping timeout: 256 seconds)
[19:23:01] <jbk> maybe? for that one at least, there is a dis command as well
[19:31:40] <tsoome> does not really work:/
[19:31:52] <rzezeski> tsoome: you can point mdb at the module (mdb <path>)
[19:32:37] <tsoome> gobjdump is not mapping the function names ffor call instruction, so it’s a bit annoying
[19:33:02] <tsoome> root@openindiana:/mnt/kernel/drv/sparcv9# mdb stmf_sbd
[19:33:03] <tsoome> > ::dis
[19:33:03] <tsoome> mdb: failed to read instruction at 0: no mapping for address
[19:33:19] <jlevon> you need to pass a function name at least
[19:34:01] <tsoome> > _init::dis
[19:34:02] <tsoome> mdb: failed to read instruction at 0x18: no mapping for address
[19:35:15] <tsoome> of course I could just boot -k and get there after panic:)
[19:36:07] <jlevon> wfm? is sparc unahppy somehow?
[19:36:22] <rzezeski> yea...that works for me just fine
[19:37:15] *** nde <nde!uid414739@gateway/web/irccloud.com/x-ngjcoljxhpocrdle> has joined #illumos
[19:37:22] <tsoome> its my fate:) touch something, and found another bug:D
[19:37:32] <jlevon> we all feel like that some days
[19:38:17] <tsoome> thjats the issue: https://paste.ec/paste/5FxeZfLi#P75fv6XD7qj7EG7hfZza82lsO+StmnD5VNvOH3++a3u
[19:38:52] <jlevon> alignment trololol
[19:39:28] <tsoome> probably need to check after recent commits there
[19:39:35] <tsoome> yep
[19:40:00] <rmustacc> When I can work on somehing and not get another bug, that feels like a great day.
[19:40:04] <jlevon> interestingly, ::nm is broken on my stmf_sbd module
[19:41:47] <tsoome> hm…. I wonder, gcc9 is able to report unaligned access issues, maybe it is time to kick the gcc9 build running
[19:42:08] <rmustacc> Ironically, that was something lint did do (though most folks gagged it)
[19:42:20] <jlevon> even ::nm -u is empty
[19:42:22] <jlevon> weird
[19:43:16] <rmustacc> That seems odd.
[19:43:45] <jlevon> guess I'll take a look tomorrow
[19:43:50] <tsoome> nxge: variable may be used uninitialized, also iwn
[19:44:13] <tsoome> but those may have been added after lint was disabled
[19:45:34] <rmustacc> Maybe used uninit is rather different from alignment.
[19:49:32] *** jemershaw <jemershaw!~jemershaw@c-68-83-252-28.hsd1.pa.comcast.net> has quit IRC (Ping timeout: 256 seconds)
[19:50:41] <tsoome> huh. http://src.illumos.org/source/xref/illumos-gate/usr/src/uts/common/io/comstar/stmf/stmf.c#2203
[19:51:05] *** jemershaw <jemershaw!~jemershaw@c-68-83-252-28.hsd1.pa.comcast.net> has joined #illumos
[19:51:05] <tsoome> makes me to wonder...
[19:51:26] <rmustacc> That should just make sure you have an aligned allocation size.
[19:51:45] *** BOKALDO <BOKALDO!~BOKALDO@81.198.156.19> has quit IRC (Quit: Leaving)
[19:53:36] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has quit IRC (Remote host closed the connection)
[20:00:38] *** alanc <alanc!~alanc@inet-hqmc01-o.oracle.com> has joined #illumos
[20:04:50] *** igork <igork!~igork@91.204.56.74> has quit IRC (Ping timeout: 256 seconds)
[20:07:04] *** wacki <wacki!~wacki@i577B8ABE.versanet.de> has joined #illumos
[20:07:56] *** igork <igork!~igork@91.204.56.74> has joined #illumos
[20:31:16] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-136-97.rochester.res.rr.com> has joined #illumos
[20:38:28] <LeftWing> toastersonerson1: Ah, fair enough! Sorry, I missed it when I read through.
[20:49:23] *** jlevon <jlevon!jlevon@2a01:7e00::f03c:92ff:fefb:3ad2> has quit IRC (Quit: Coyote finally caught me)
[20:50:56] *** jlevon <jlevon!jlevon@2a01:7e00::f03c:92ff:fefb:3ad2> has joined #illumos
[21:31:59] *** Kurlon <Kurlon!~Kurlon@cpe-67-253-136-97.rochester.res.rr.com> has quit IRC (Remote host closed the connection)
[21:32:33] *** Kurlon <Kurlon!~Kurlon@bidd-pub-03.gwi.net> has joined #illumos
[21:39:49] <sjorge> LeftWing illumos mail (gerrit) are being blocked again, we're on RBL_SPAMCOP RBL_NIXSPAM
[21:39:56] <sjorge> I assume it's... sendgrid again
[21:40:06] <LeftWing> Ugh I'm sorry
[21:40:51] <LeftWing> sjorge: Do you know which IP address it is this time?
[21:41:24] <sjorge> 167.89.100.130
[21:43:45] <LeftWing> I guess maybe we should just try and manage our own SMTP server :/
[21:59:23] <tsoome> so, my panic I see is caused by 11820 upstream Nexenta iSCSI and COMSTAR fixes
[21:59:39] <LeftWing> On x86?
[21:59:48] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has quit IRC (Ping timeout: 256 seconds)
[22:00:07] <tsoome> sparc.
[22:00:11] <LeftWing> Ah OK
[22:00:32] <tsoome> this one. https://paste.ec/paste/5FxeZfLi#P75fv6XD7qj7EG7hfZza82lsO+StmnD5VNvOH3++a3u
[22:00:58] <LeftWing> Ahh alignment :(
[22:02:12] <tsoome> yes. I did comment out the block in _init to check HardwareAcceleratedLocking, HardwareAcceleratedMove and HardwareAcceleratedInit and the panic is gone.
[22:02:14] <LeftWing> It has ... an unusual looking cast
[22:03:30] <LeftWing> Those appear to be uint8_t globals... I wonder if those are not working correctly on SPARC?
[22:04:28] <tsoome> for very simple reason
[22:04:34] <tsoome> extern int HardwareAcceleratedLocking;
[22:04:51] <tsoome> it is quite justified we get this thing blowing up
[22:05:54] <tsoome> all 3 are actually uint8_t.
[22:06:46] <LeftWing> Ahh!
[22:06:47] <tsoome> looks like we need gcc 10 to diagnose those:D
[22:06:49] <LeftWing> That is irritating
[22:06:57] <LeftWing> This is why I try to put extern things in _headers_
[22:13:26] <tsoome> yep, that did fix it.
[22:13:37] *** mnowak_ <mnowak_!~mnowak_@94.142.238.232> has joined #illumos
[22:21:45] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has quit IRC (Quit: Leaving)
[22:28:03] *** ngchk1 <ngchk1!~ngchk1@b2b-92-50-91-166.unitymedia.biz> has joined #illumos
[22:43:35] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-lwbluebzyidwsqoj> has quit IRC (Read error: Connection reset by peer)
[22:44:20] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-qivqygmcrqxtoshr> has joined #illumos
[22:44:52] <am11> opened an issue for h/w exception handling case: https://github.com/dotnet/runtime/issues/35362. note that the (years old) issue i have referenced was from NetBSD folks, for the exact same test with slightly different location. something is corrupting the frame when sigsegv happens in our case, it seems.
[22:46:50] *** KeiraT <KeiraT!~k4ra@gateway/tor-sasl/k4ra> has quit IRC (Write error: Broken pipe)
[22:46:50] *** cantstanya <cantstanya!~chatting@gateway/tor-sasl/cantstanya> has quit IRC (Remote host closed the connection)
[22:48:36] *** KeiraT <KeiraT!~k4ra@gateway/tor-sasl/k4ra> has joined #illumos
[22:49:40] *** cantstanya <cantstanya!~chatting@gateway/tor-sasl/cantstanya> has joined #illumos
[22:49:45] <am11> this is the last failure in PAL tests (provided thread issue is fixed upstream, for non-elevated user), once we are through this, then the 'real' runtime testing (other parts like VM, codegen etc.) will start. 🙈
[22:51:29] *** heroux_ <heroux_!~heroux@507082E8.static.ziggozakelijk.nl> has joined #illumos
[22:55:20] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-qivqygmcrqxtoshr> has quit IRC (Read error: Connection reset by peer)
[22:55:22] *** heroux_ is now known as heroux
[22:59:34] *** heroux <heroux!~heroux@507082E8.static.ziggozakelijk.nl> has quit IRC (Ping timeout: 240 seconds)
[23:00:42] *** heroux <heroux!sandroco@gateway/shell/insomnia247/x-exbgkgzwabkhhmzs> has joined #illumos
[23:11:18] *** KungFuJesus <KungFuJesus!~adam@72.49.160.91> has joined #illumos
[23:12:07] <KungFuJesus> hmm, gkrellmd is segfaulting for me for some odd reason
[23:14:51] <KungFuJesus> looks to be a null pointer deref for a function
[23:15:42] <KungFuJesus> despite building separately with debug info (outside of the oi-userland provided package), the stack trace looks completely senseless
[23:24:05] *** gh34 <gh34!~textual@cpe-184-58-181-106.wi.res.rr.com> has quit IRC (Quit: Textual IRC Client: www.textualapp.com)
[23:25:42] *** ptribble <ptribble!~ptribble@cpc92716-cmbg20-2-0-cust138.5-4.cable.virginm.net> has quit IRC (Remote host closed the connection)
[23:26:36] <richlowe> forgot to keep the frame pointer?
[23:33:28] <KungFuJesus> looks like it only compiled with -O2
[23:33:42] <KungFuJesus> I don't see any fomit-frame-pointer's
[23:34:24] *** phyre <phyre!~phyre___@78.30.23.107> has joined #illumos
[23:36:19] *** phyre <phyre!~phyre___@78.30.23.107> has quit IRC (Remote host closed the connection)
[23:36:21] <sjorge> LeftWing: perhaps, postfix + rspamd from package source work great... assuming we don’t need imap
[23:37:44] <LeftWing> We just need a relay
[23:38:06] <LeftWing> And it's fairly constrained in terms of what will go through it -- just Gerrit and the bug tracker
[23:38:19] <LeftWing> So the need for outbound spam filtering is probably fairly minimal
[23:38:28] <sjorge> So rspamd can be dropped too then
[23:38:51] <sjorge> Then again, rspam also does the dkim stuff
[23:38:56] <LeftWing> We actually run Postfix already locally on the Gerrit and Redmine machines, it manages the forwarding through sendgrid today
[23:38:57] <sjorge> For outbound mail
[23:39:08] <LeftWing> I could probably set up a central mail IP instead...
[23:39:27] <LeftWing> I wonder if I will be able to use Digital Ocean for that or if their IP blocks are all pre blacklisted
[23:39:57] <LeftWing> I'll put it on my TODO list anyway.
[23:43:18] <sjorge> I’m at sys (cheap ovh tier) and it took about a while of me having the up to clear it from blacklist and het it on a few whitelist. Not been an issue since, so I think digital ocean should be fine
[23:43:33] <sjorge> Also... who pays for the illumos.org hosting?
[23:43:45] *** KungFuJesus <KungFuJesus!~adam@72.49.160.91> has quit IRC (Quit: Lost terminal)
[23:54:33] <LeftWing> sjorge: That's a complicated question at the moment, but if you have money I won't refuse it haha
[23:55:01] <LeftWing> I've been working on getting the project attached to an existing foundation
[23:55:29] <LeftWing> But they're all pretty busy and it's slow going, especially with recent events
[23:57:38] *** KungFuJesus <KungFuJesus!~adam@72.49.160.91> has joined #illumos
[23:57:58] <KungFuJesus> Yeah, there's no build flags to omit the frame pointer here, it just seems my stack is a mystery to both mdb and gdb
[23:58:45] <KungFuJesus> I can't tell you what commit hash my previous boot environment was on, but I can say for sure it didn't segfault. This is strange
top

   April 23, 2020  
< | 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 | >