Show enters and exits. Hide enters and exits.
| 00:29:25 | devinus | http://bench.rubini.us/ -- wait so rubinius is only 10% as fast as MRI? |
| 00:29:39 | devinus | 10-15%? |
| 06:34:42 | boyscout | Set the proper type for a MethodTableBucket - 5c6ba6e - Evan Phoenix |
| 06:34:42 | boyscout | Reenable accessor inlining - eb153b9 - Evan Phoenix |
| 06:34:42 | boyscout | Remove unused Kernel#breakpoint - 2ccf2ce - Evan Phoenix |
| 06:42:24 | boyscout | CI: 2ccf2ce success. 2733 files, 10853 examples, 33893 expectations, 0 failures, 0 errors |
| 10:00:51 | boyscout | Hash performance improvements. - eb3e7f5 - Brian Ford |
| 10:00:51 | boyscout | Update Hash specs. - 6e7b26d - Brian Ford |
| 10:00:51 | boyscout | Updated 1.8.7 Hash methods. - 636e37b - Brian Ford |
| 10:03:05 | boyscout | CI: 636e37b success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 17:24:25 | evan | woop. reader and writer accessor inlining |
| 17:35:39 | boyscout | Inline accessor writers as well - fd67de0 - Evan Phoenix |
| 17:38:40 | boyscout | CI: fd67de0 success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 17:38:46 | evan | danku boyscout |
| 17:59:58 | cremes | anyone having success running this stuff on snow leopard? |
| 18:00:17 | cremes | my build hasn't worked for at least a week; it doesn't even finish compiling: http://pastie.org/526481 |
| 18:06:25 | evan | not that I know of |
| 18:06:30 | dbussink | cremes: it's a known issue with apple's gcc 4.2.1 |
| 18:06:56 | cremes | dbussink: i see... any known workarounds or wait for an updated seed? |
| 18:07:12 | dbussink | cremes: but it fails differently for me on my snow leopard test machine |
| 18:07:20 | dbussink | cremes: and there are no known workarounds afaik |
| 18:07:43 | dbussink | but it's on an old macbook which isn't 64 bit |
| 18:07:45 | evan | cremes: you need to force release libudis |
| 18:07:55 | evan | er. |
| 18:07:57 | evan | force rebuild |
| 18:07:58 | dbussink | that probably explains the difference in issues |
| 18:08:09 | cremes | how do i force it? is there a rake task? |
| 18:08:15 | evan | um. |
| 18:08:16 | dbussink | cremes: if you get passed by this, it will fail later on ;) |
| 18:08:17 | evan | should be.. |
| 18:08:43 | evan | cremes: otherwise, just rm libudis86.a in assembler |
| 18:08:47 | dbussink | evan: did you even look into the gcc 4.2.1 issue? |
| 18:08:49 | dbussink | ever |
| 18:08:57 | evan | and go into assembler/udis86-1.7 and make clean |
| 18:09:07 | evan | dbussink: no, i need to |
| 18:09:12 | evan | rue was, but he's not really in here anymore i guess. |
| 18:09:19 | evan | i don't think he had time anyway |
| 18:09:33 | evan | the error is seeing a nil or something, yels? |
| 18:09:43 | cremes | rebuilding now... |
| 18:10:05 | dbussink | evan: you, it goes wrong when compiling readline |
| 18:10:19 | cremes | dbussink: snow leopard also has gcc 4.0; have you tried compiling with it instead of 4.2.1? |
| 18:10:21 | evan | could ya pastie the error again? |
| 18:10:26 | evan | i've got a little time |
| 18:10:31 | evan | i'll see if I can't slay it. |
| 18:11:40 | dbussink | http://gist.github.com/137049 |
| 18:12:45 | dbussink | evan: probably some bug in the compiler, there's no issue if i use gcc 4.2.4 from macports |
| 18:12:54 | dbussink | or the 4.3 or 4.4 series |
| 18:12:54 | evan | really? |
| 18:12:56 | evan | ha |
| 18:13:01 | evan | well, i guess thats "good" |
| 18:13:31 | dbussink | evan: i thought it was also reproducable by using 4.2 on regular leopard |
| 18:13:42 | evan | it might be |
| 18:13:49 | evan | but i need to get validation of the error from you |
| 18:13:55 | evan | so i can know if i'm reproing the same thing here :D |
| 18:14:02 | dbussink | hehe, that's true yeah |
| 18:15:23 | evan | years of debugging has taught me to ask stupid questions up front |
| 18:15:26 | evan | so i don't chase my tail for days |
| 18:16:18 | dbussink | yeah, i agree, that's definitely true |
| 18:17:15 | evan | yay! i got methods that just return a number inlining |
| 18:17:19 | evan | how silly! |
| 18:17:22 | cremes | dbussink: yeah, i fail on readline too though mine segfaults instead of giving any kind of backtrace |
| 18:20:26 | dbussink | evan: all great things start small :) |
| 18:20:33 | evan | heh |
| 18:22:47 | dbussink | evan: ok, compiling with 4.0 on snow leopard works |
| 18:22:52 | evan | yay |
| 18:23:15 | dbussink | evan: and compiling on regular leopard gives the same error with the available 4.2 as on snow leopard |
| 18:23:21 | evan | ok, i'm adding inlining for methods that return symbols |
| 18:23:23 | evan | then i'll check this out. |
| 18:23:28 | dbussink | np :) |
| 18:23:40 | dbussink | evan: does it also inline when you have returns / blocks etc.? |
| 18:23:57 | evan | atm, i'm doing form detection |
| 18:24:02 | evan | it's not inlining any method yet |
| 18:24:04 | evan | soon though :) |
| 18:24:13 | evan | i mean, it's not inlining just any method |
| 18:24:25 | evan | i added support for it to inline accessors (attr_reader, attr_writer) |
| 18:24:36 | evan | now i'm adding support for detecting trivial methods and inlining them |
| 18:24:43 | dbussink | ah, cool |
| 18:24:47 | evan | next is small but non-trivial methods |
| 18:24:48 | dbussink | for what definition of trivial? |
| 18:24:55 | evan | small == no blocks, no exception handlers |
| 18:25:09 | evan | trivial is pretty much on expression on one line |
| 18:25:11 | evan | def foo; 1; end |
| 18:25:15 | evan | def foo; :blah; end |
| 18:25:16 | evan | etc. |
| 18:25:28 | evan | s/on expr/an expr/ |
| 18:26:16 | evan | hehe |
| 18:26:16 | evan | word |
| 18:26:19 | evan | symbols work now. |
| 18:26:20 | evan | :D |
| 18:28:27 | dbussink | w00t :) |
| 18:29:05 | evan | ok, now on to 4.2.1 |
| 18:29:12 | boyscout | Inline methods that just return a number - e697744 - Evan Phoenix |
| 18:29:12 | boyscout | Add inlining of methods just returning symbols - c9abaaa - Evan Phoenix |
| 18:29:39 | cremes | dbussink: how did you recompile with gcc-4.0? i tried "GCC=gcc-4.0 rake build" but it didn't pick up the change |
| 18:29:46 | evan | cremes: CC= |
| 18:29:48 | evan | not GCC |
| 18:29:51 | cremes | oops |
| 18:31:24 | boyscout | CI: c9abaaa success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 18:32:15 | dbussink | evan: does the ruby core have a lot of trivial methods? |
| 18:32:38 | evan | no, not alot. |
| 18:32:56 | evan | these are the low hanging fruit of inlining though |
| 18:32:59 | evan | the accessors make a difference |
| 18:33:01 | evan | so they're heavily used. |
| 18:33:07 | evan | s/so/since/ |
| 18:33:35 | evan | and we're at the stage where every little bit helps |
| 18:36:50 | evan | oh, i've got the blow up here. |
| 18:37:18 | evan | lets poke. |
| 18:39:07 | evan | hm hm, ok. |
| 18:39:33 | evan | that masgn, for some reason, is get confused. |
| 18:40:56 | evan | yep! |
| 18:40:56 | evan | ok |
| 18:41:03 | evan | rather than [], [], nil, nil, [] |
| 18:41:07 | evan | that rhs is coming out as |
| 18:41:12 | evan | [], nil, nil, [], [] |
| 18:43:57 | evan | heh |
| 18:44:04 | evan | a, b, c, d, e = :a, :b, :c, :d, :e |
| 18:44:04 | evan | p [a, b, c, d, e] |
| 18:44:12 | evan | what do I get under gcc 4.2 though? |
| 18:44:16 | evan | [:b, :d, :c, :a, :e] |
| 18:44:22 | evan | hah |
| 18:44:26 | evan | it's completely jumbled |
| 18:44:46 | evan | so weird. |
| 18:45:16 | evan | i'm on it though. |
| 18:48:47 | cremes | hmmm, i see that there is now a snow leopard seed update available through software update; maybe that includes a fixed compiler :) |
| 18:49:07 | evan | i've isolated it to the rotate bytecode |
| 18:49:11 | evan | i'm working on it now. |
| 18:51:17 | evan | should be easy |
| 18:51:25 | evan | now that i've got a minimal test case |
| 18:52:00 | cremes | evan, as usual you rock! |
| 18:52:13 | dbussink | cremes: it doesn't :( |
| 18:52:20 | dbussink | cremes: already installed it and no help |
| 18:52:53 | cremes | too bad; i should open a radar ticket on this; a rubinius compilation failure should be a showstopper bug for snow leopard!! |
| 18:53:27 | evan | well |
| 18:53:29 | dbussink | hold the presses, no release before it compiles! |
| 18:53:30 | dbussink | :P |
| 18:53:43 | evan | it's possible the code in rotate is using an undefined behavior |
| 18:53:46 | evan | i'm simplifying it |
| 18:53:57 | evan | to see if i can't figure out where the undefined behavior is |
| 18:54:10 | cremes | evan: what file are you working on (and line no); I want to follow along if i may |
| 18:54:33 | evan | sure |
| 18:54:37 | evan | it's easy |
| 18:54:38 | evan | here |
| 18:54:53 | evan | http://gist.github.com/137060 |
| 18:54:57 | evan | run |
| 18:55:12 | evan | ruby lib/compiler/mri_compile.rb scratch/masgn.rb scratch/masgn.rbc |
| 18:55:13 | evan | then |
| 18:55:19 | evan | bin/rbx scratch/masgn.rb |
| 18:55:32 | evan | you should see that the locals don't have the correct values |
| 18:55:41 | cremes | cool; will do |
| 18:57:48 | cremes | of course, i can't do that last part because i don't have bin/rbx successfully compiled on this box |
| 18:58:26 | evan | it should be compiling |
| 18:59:54 | dbussink | cremes: just curious, is it creating a 32 or 64 bit binary for you? |
| 19:00:16 | cremes | dbussink: i'm recompiling now; when done i'll check |
| 19:00:34 | cremes | i'm pretty sure it is outputting mach-o x86_64 but I |
| 19:00:36 | cremes | ll confirm |
| 19:02:33 | cremes | dbussink: 64-bit |
| 19:04:01 | cremes | i get a segfault when running "bin/rbx scratch/masgn.rb" |
| 19:04:44 | dbussink | cremes: hmm, that might be a 64 bit related issue |
| 19:05:04 | cremes | yeah? is there any easy way to force 32-bit compilation? |
| 19:05:27 | dbussink | export RC_ARCHS="i386" |
| 19:05:31 | dbussink | that might help |
| 19:05:36 | dbussink | dunno for sure though |
| 19:05:42 | dbussink | evan: i see the weird order too yeah |
| 19:06:08 | evan | ok |
| 19:06:09 | evan | good. |
| 19:17:50 | dbussink | evan: already able to determine whether it's a compiler bug or not? |
| 19:19:28 | evan | working on that now |
| 19:19:32 | evan | i'm moving the logic into a .c file |
| 19:19:41 | evan | and trying to observe it in isolation |
| 19:26:09 | dbussink | cremes: did that work btw? |
| 19:26:17 | mernen | cremes: rbx segfaults on my (ubuntu) 64-bit builds too |
| 19:26:20 | cremes | nope |
| 19:26:29 | dbussink | mernen: ah, probably the same issue then |
| 19:27:19 | cremes | maybe i'll try running it under the debugger to see where it dies |
| 19:27:49 | mernen | I've got two or three different problem points |
| 19:28:01 | mernen | but it's always related to stack_push |
| 19:28:38 | mernen | tried to figure it out yesterday, but so far that seems to be beyond my abilities |
| 19:29:38 | evan | mernen: if you've got some backtraces |
| 19:29:42 | evan | feel free to put them into a github issue |
| 19:29:45 | evan | as points of data |
| 19:29:47 | mernen | it's been like that since 9bdf2fa, btw |
| 19:30:04 | mernen | the VMMethod::interpreter refactor |
| 19:30:19 | mernen | evan: I have a few core dumps here |
| 19:31:20 | dbussink | mernen: you git bisected it? |
| 19:31:25 | mernen | yep |
| 19:32:02 | mernen | bisected until `rbx -rFAIL -e ''` didn't crash |
| 19:33:10 | mernen | I mean, didn't segfault |
| 19:34:19 | evan | hm. ok. |
| 19:38:56 | evan | ha! |
| 19:39:03 | evan | i think i got a version that gcc likes! |
| 19:39:10 | evan | bingo! |
| 19:41:58 | dbussink | evan: can you show me the c repro? |
| 19:42:23 | dbussink | evan: or do you already have a workaround? |
| 19:43:11 | evan | i never got a C repro |
| 19:43:38 | evan | i think because the interpreter has different alias semantics that a trivial C program |
| 19:43:43 | evan | so it's hard to get the bug to appear |
| 19:43:54 | evan | i'm retesting the new rotate code |
| 19:44:21 | dbussink | evan: but is it an obvious gcc bug or is there undefined behavior involved too? |
| 19:45:32 | evan | i can't see an undefined behavior in the current one |
| 19:45:47 | evan | it's just moving data in-place in an array |
| 19:46:08 | evan | but i've redo it using pointer access and pointer math |
| 19:46:14 | evan | which seems to solve it |
| 19:48:09 | evan | running the specs now (which run!) |
| 19:53:10 | evan | interesting! |
| 19:53:27 | evan | gcc 4.2 results in in what appears to be a bit faster code |
| 19:53:31 | evan | 43.8s for a spec run |
| 19:53:38 | evan | i'm going to clean and recompile under gcc-4.0 |
| 19:54:03 | dbussink | evan: it's inside the rotate instruction? |
| 19:56:35 | evan | the bug is, yes. |
| 19:56:39 | evan | it's fixed |
| 19:56:46 | evan | i'm testing on gcc 4.0 now |
| 20:01:25 | evan | weird! |
| 20:01:32 | evan | this change makes the interpreter faster |
| 20:01:43 | evan | maybe the old code was confounding gcc's optimizer |
| 20:01:53 | dbussink | yeah, just wanted to say that too :P |
| 20:02:37 | brixen | evan: btw, with the new hash code, I'm getting just a hair over 40sec full spec runs |
| 20:03:02 | evan | hey! why are you online! |
| 20:03:03 | evan | :D |
| 20:03:06 | brixen | I'm playing with an optimizer pass in the bytecode compiler |
| 20:03:07 | brixen | :) |
| 20:03:29 | brixen | I'm just sitting here in a coffee shop and *felt* all the activity here :D |
| 20:03:37 | brixen | cute place http://thumpcoffee.com/ |
| 20:03:43 | evan | hah |
| 20:04:13 | brixen | but seriously, I'll have sub 40sec full CI runs shortly I think |
| 20:04:18 | evan | yay! |
| 20:04:42 | evan | and to think, i was all excited about the first sub 50s time not long ago |
| 20:04:48 | evan | and now I just saw 43s! |
| 20:04:54 | brixen | http://gist.github.com/137080 |
| 20:05:02 | brixen | I'm lazily poking at Array#| |
| 20:05:11 | brixen | it's the biggest user of Hash in the spec runs |
| 20:05:16 | evan | yeah, so |
| 20:05:25 | evan | i'm going to have to come up with a solution for Class#new |
| 20:05:34 | evan | i had to take the check_serial code out when I changed the ICs |
| 20:05:45 | evan | i think i'm going to have new to ba a meta opcode like _call |
| 20:05:47 | brixen | btw, this is the only place in Bend that serves stumptown coffee |
| 20:05:56 | evan | that can check if new is Class#new |
| 20:06:05 | brixen | that would be cool |
| 20:06:05 | evan | and do the allocation and call initialize |
| 20:06:18 | evan | without boxing the arguments |
| 20:06:27 | evan | the only thing about that, it's got crappy IC semantics |
| 20:06:41 | evan | at the call site, i really want to get an IC for allocate and initilaize |
| 20:06:45 | evan | seperately |
| 20:06:50 | evan | so I need to think about this |
| 20:07:12 | brixen | hmm |
| 20:08:38 | boyscout | Rework the rotate instruction (gcc 4.2 fix) - cc26edb - Evan Phoenix |
| 20:08:41 | evan | ok, gcc 4.2 people |
| 20:08:43 | evan | give that a shot. |
| 20:11:24 | evan | so, something i was thinking about |
| 20:11:30 | evan | for Kernel methods |
| 20:11:41 | cremes | well, it *might* be better but i still get segfaults probably due to it being 64-bit |
| 20:11:51 | evan | the inliner could inline them without a guard |
| 20:11:53 | cremes | so i can't actually make sure this other problem is fiexed |
| 20:12:14 | cremes | dbussink: work for you? |
| 20:12:19 | evan | and the deoptimizer just needs to find out if adding a new method causes masking of a Kernel method |
| 20:13:01 | dbussink | cremes: compiling |
| 20:13:12 | boyscout | CI: cc26edb success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 20:13:31 | brixen | evan: like at method def time, do a method lookup and see if you hit a Kernel method, if so, set a flag for those inlined methods somehow? |
| 20:14:06 | slava | hi |
| 20:14:29 | evan | brixen: yes, do the lookup and if you find a Kernel method |
| 20:14:34 | brixen | gutten tag herr slava |
| 20:14:44 | evan | then you tell all inliners of that Kernel method to deoptimize |
| 20:14:49 | brixen | makes sense |
| 20:14:50 | evan | that cuts deep |
| 20:15:04 | evan | so if we can reason that certain inliners won't see the newly defined method |
| 20:15:07 | evan | we don't need to deoptimize them |
| 20:15:20 | evan | could do that with hierarchy analysis |
| 20:17:24 | mernen | hah! |
| 20:17:30 | evan | "tell all inliners of a method to go back to using the interpreter" == dynamic deoptimization |
| 20:17:36 | evan | don't let jargon get in your way! |
| 20:17:44 | mernen | evan: I see sometimes UnwindInfo::stack_depth has a value of 0xffffffff (-1) |
| 20:17:48 | mernen | is that right? |
| 20:17:57 | evan | likely no. |
| 20:18:08 | evan | wel |
| 20:18:15 | evan | lets see... |
| 20:18:20 | mernen | I mean, is that expected? |
| 20:18:28 | evan | yeah, i'm checking |
| 20:18:32 | evan | at one point, I believe it was. |
| 20:18:33 | mernen | anyway, trouble is stack_depth is a uint32_t |
| 20:18:42 | evan | oh. |
| 20:18:43 | evan | hehe |
| 20:18:44 | mernen | on a 32-bit machine, that value is just -1 |
| 20:18:54 | mernen | (for all that matters) |
| 20:18:58 | evan | but on 64bit |
| 20:18:59 | evan | thats 4b! |
| 20:19:01 | mernen | but on 64-bit it throws the stack pointer waaay off |
| 20:19:04 | evan | probably not the intended value. |
| 20:19:16 | evan | ok, let me check-a-poo |
| 20:19:32 | evan | ok, yes |
| 20:19:56 | evan | stack_calculate_sp() can return -1 |
| 20:19:59 | evan | thats valid |
| 20:20:08 | evan | so stack_depth should be just an int |
| 20:20:31 | evan | the reason -1 is valid, btw |
| 20:20:32 | mernen | compiling now, let's see if that solves it |
| 20:20:47 | evan | is because the stack pointer starts live pointing to one BELOW the stack |
| 20:20:58 | evan | because the stack pointer is expect to be pointing at the current stack top |
| 20:21:02 | evan | and if there is nothing on the stack |
| 20:21:06 | evan | it points one below the stack |
| 20:21:11 | brixen | the stack is topless! |
| 20:21:17 | evan | SPRING BREAK! |
| 20:21:23 | brixen | heh |
| 20:21:32 | mernen | that explains why `def f() end` with no body segfaults, I guess |
| 20:21:52 | evan | ah! probably. |
| 20:22:50 | mernen | looking good, so far |
| 20:23:20 | mernen | all tests pass |
| 20:23:24 | evan | yay! |
| 20:23:33 | evan | mernen: good job sherlock! |
| 20:31:37 | boyscout | Inline methods which return self - af009db - Evan Phoenix |
| 20:32:40 | evan | another trivial case |
| 20:32:51 | evan | and, happily, that only require 3 lines of code to do |
| 20:33:44 | boyscout | CI: af009db success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 20:36:21 | brixen | Finished in 40.635683 seconds |
| 20:36:25 | brixen | soo close! |
| 20:38:01 | evan | oooh |
| 20:38:33 | dbussink | cremes: fixed it for me (32 bit snow leopard) |
| 20:39:04 | evan | yay |
| 20:39:20 | boyscout | Change type of UnwindInfo::stack_depth to int - 570217f - Daniel Luz |
| 20:40:23 | mernen | cremes: see if it still segfaults on 64-bit |
| 20:43:33 | boyscout | CI: 570217f success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 20:44:53 | boyscout | Push down entries on rehash/redistribute. - 4bfc246 - Brian Ford |
| 20:47:06 | boyscout | CI: 4bfc246 success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 20:52:49 | dbussink | mernen: hmm, i'm still seeing a segfault on 64 bit when compiling readline |
| 20:53:25 | mernen | weird, it works fine here |
| 20:53:43 | mernen | readline and all |
| 20:54:27 | mernen | did you try a clean build? |
| 21:06:42 | dbussink | mernen: hmm, did a rake clean and still the same issue |
| 21:07:51 | dbussink | mernen: are you compiling with RBX_LLVM=1 ? |
| 21:10:38 | radarek | dbussink: I have the same problem |
| 21:16:26 | mernen | nope, no LLVM here |
| 21:20:07 | mernen | I'll try LLVM now, but that'll take a while |
| 21:21:42 | dbussink | evan: looks like there's a segfault in the compile thread when using llvm or something like that |
| 21:21:47 | dbussink | evan: for 64 bit systems that it |
| 21:22:23 | mernen | I just have to place LLVM 2.5 source into vm/external_libs/llvm and build with RBX_LLVM=1, right? |
| 21:22:32 | dbussink | evan: http://gist.github.com/137107 |
| 21:23:04 | mernen | dbussink: even with a segfault on readline, the rbx binary should be there |
| 21:23:18 | mernen | can you test for other segfaults? |
| 21:23:40 | mernen | in my case, stuff like rbx -rINVALID_FILE and rbx -e 'def f; end' were sure ways to segfault |
| 21:28:17 | slava | evan: can you tell me more about your type feeback? |
| 21:31:41 | radarek | radarek@ruby:~/opt/src/rubinius(master)$ ./bin/rbx -e "def f; end" |
| 21:31:41 | radarek | Segmentation fault |
| 21:31:41 | radarek | radarek@ruby:~/opt/src/rubinius(master)$ ./bin/rbx -rFOO |
| 21:31:41 | radarek | Segmentation fault |
| 21:32:11 | mernen | radarek: what about def f; 1 end? |
| 21:32:27 | radarek | mernen: works |
| 21:32:31 | mernen | seems to be the exact same problem, then |
| 21:32:33 | radarek | mernen: -e "puts 1" also works |
| 21:33:12 | dbussink | mernen: ok, i should pull before complaning :P |
| 21:33:28 | dbussink | having a rubinius clone on 3 machines is bound to confuse oneself :) |
| 21:34:07 | radarek | ok, I'll pull and rebuild it |
| 21:36:13 | radarek | mernen: after pulling and rebuilding -e "def f; end" works fine but it still can't compile readline |
| 21:38:36 | mernen | are you using LLVM too? |
| 21:40:00 | radarek | mernen: yes |
| 21:48:18 | dbussink | mernen: did you try running with llvm too? |
| 21:48:30 | mernen | I'm compiling LLVM now |
| 21:48:45 | dbussink | going to try without llvm and see whether it goes away |
| 21:49:25 | mernen | vm/llvm/jit.hpp:15:42: error: llvm/CodeGen/MachineCodeInfo.h: No such file or directory |
| 21:49:58 | dbussink | mernen: did you get llvm from svn trunk? |
| 21:50:11 | dbussink | 2.5 is too old |
| 21:50:11 | mernen | no, 2.5 release |
| 21:50:14 | mernen | oh |
| 21:54:42 | dbussink | mernen: ok, without llvm it compiles succesfully, but there are a bunch of test failures |
| 22:03:33 | boyscout | Use the correct number of digit bits on 64 bit - 5c7d56a - Dirkjan Bussink |
| 22:09:58 | dbussink | mernen: that fixes the failures i saw on 64 bit without llvm |
| 22:10:33 | boyscout | CI: 5c7d56a success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 22:13:07 | boyscout | Fix compile time warning in syck on 64 bit - 5da02ef - Dirkjan Bussink |
| 22:15:20 | boyscout | CI: 5da02ef success. 2723 files, 10829 examples, 33841 expectations, 0 failures, 0 errors |
| 22:18:19 | mernen | hmm |
| 22:18:46 | mernen | dbussink: now several C++ tests are failing |
| 22:21:28 | dbussink | mernen: they were for me, but they were fixed with my last commits |
| 22:21:38 | dbussink | evan: you there? |
| 22:21:57 | mernen | weird, I was in fact getting one spec failure before |
| 22:22:12 | mernen | but rubinius' tests themselves were all okay |
| 22:22:12 | dbussink | mernen: that's what i'm seeing now |
| 22:22:23 | dbussink | mernen: did you rebuild libtommath? |
| 22:22:35 | mernen | I rake-clean'd… I think |
| 22:22:51 | dbussink | mernen: you can do a make clean and then make in vm/external_libs/libtommath |
| 22:23:00 | mernen | now, on the llvm copy |
| 22:23:20 | mernen | vm/external_libs/llvm/include/llvm/ExecutionEngine/JIT.h:33: undefined reference to `LLVMLinkInJIT' |
| 22:23:51 | mernen | I'm getting hundreds of references to LLVMLinkInJIT |
| 22:24:48 | mernen | any ideas? |
| 22:27:06 | mernen | command not found: vm/external_libs/llvm/Release/bin/llvm-config --cflags |
| 22:27:10 | mernen | that might have something to do with it! |
| 22:30:07 | mernen | dbussink: ok, rebuilt libtommath, tests now pass |
| 22:37:33 | dbussink | mernen: hmm, you did a svn checkout of llvm? |
| 22:38:01 | mernen | yeah, but apparently rake didn't finish building LLVM |
| 22:38:14 | mernen | I'm running make manually now |
| 22:46:31 | mernen | ok, llvm done, readline now segfaults here too |
| 22:48:51 | mernen | A problem internal to GDB has been detected, further debugging may prove unreliable. |
| 22:50:46 | dbussink | mernen: auch, that's a nasty on |
| 22:50:47 | dbussink | one |