On 01/03 03:08, Alexey Suslikov wrote:
> Hi ports@.
>
> Roman Kravchuk and me are working on Ruby on Rails 4.0 port (incl
> Active Record and such).
Offtopic, but:
We did have a previous rails port, which was removed, partly because it
was old and insecure, and partly because the benefit of the port does
not outweigh the cost. Search the ports@ archives for details.
If you are planning on submitting the port for inclusion in the ports
system, you'll have to explain how the benefits outweight the costs.
> Although we got it running and working (at first glance), there
> is one thing which bothers us (hence the kinda preliminary mail).
>
> Ruby thread safe library (which RoR 4.0 depends on) fails to pass
> test suite: most of the time it segfaults at 50 of 68 tests (as
> I said before, RoR 4.0 itself is working at first glance).
>
> Same test suite works under Ubuntu Linux:
>
> Finished tests in 475.533748s, 0.1430 tests/s, 5.0827 assertions/s.
> 68 tests, 2417 assertions, 0 failures, 0 errors, 0 skips
>
> ruby -v: ruby 2.0.0p353 (2013-11-22 revision 43784) [x86_64-linux-gnu]
>
> So at this point we'd love to receive some help from other people
> on (possible) threading issues.
>
> Attached are Ruby thread safe port and accompanying Ruby atomic.
>
> With Ruby thread safe tests, the following will give verbose output:
> cd `make show=WRKSRC` && rake20 --trace test
>
> Tested here on -current amd64 (Ruby atomic tests also pass).
I tested on a not-so-current -current on amd64 and everything passes:
Finished tests in 1466.086436s, 0.0464 tests/s, 1.6486 assertions/s.
68 tests, 2417 assertions, 0 failures, 0 errors, 0 skips
ruby -v: ruby 2.0.0p353 (2013-11-22 revision 43784) [x86_64-openbsd]
$ dmesg |head -n1
OpenBSD 5.4-current (GENERIC.MP) #195: Mon Dec 9 16:34:18 MST 2013
On i386, the first time I ran the tests, they didn't segfault, but
the tests don't complete:
...
TestCacheTorture#test_put_if_absent = 103.66 s = .
TestHash#test_concurrency =
/usr/obj/ports/thread_safe-0.1.3-ruby20/thread_safe-0.1.3/test/test_hash.rb:11:in
`[]=': can't add a new key into hash during iteration (RuntimeError)
from
/usr/obj/ports/thread_safe-0.1.3-ruby20/thread_safe-0.1.3/test/test_hash.rb:11:in
`block (4 levels) in test_concurrency'
from
/usr/obj/ports/thread_safe-0.1.3-ruby20/thread_safe-0.1.3/test/test_hash.rb:10:in
`times'
from
/usr/obj/ports/thread_safe-0.1.3-ruby20/thread_safe-0.1.3/test/test_hash.rb:10:in
`block (3 levels) in test_concurrency'
rake aborted!
I'm not sure why this wasn't just a failure and why it caused the test
suite to abort.
The second time, I did get a segfault, on a test that passed on the
previous run:
TestCacheTorture#test_add_remove_to_zero =
/usr/obj/ports/thread_safe-0.1.3-ruby20/thread_safe-0.1.3/lib/thread_safe/non_concurrent_cache_backend.rb:10:
[BUG] Segmentation fault
GDB backtrace looks like:
(gdb) bt
#0 0x0e6cf68d in kill () at <stdin>:2
#1 0x0e73a416 in raise (s=6) at /usr/src/lib/libc/gen/raise.c:39
#2 0x0e73a33c in abort () at /usr/src/lib/libc/stdlib/abort.c:70
#3 0x0e2dc21d in rb_bug () from /usr/local/lib/libruby20.so.0.1
#4 0x0e39089a in sigsegv () from /usr/local/lib/libruby20.so.0.1
#5 <signal handler called>
#6 0x0e398509 in find_entry () from /usr/local/lib/libruby20.so.0.1
#7 0x0e398626 in st_lookup () from /usr/local/lib/libruby20.so.0.1
#8 0x0e2f8b8a in rb_hash_aref () from /usr/local/lib/libruby20.so.0.1
#9 0x0e3f4adf in vm_exec_core () from /usr/local/lib/libruby20.so.0.1
#10 0x0e3f8866 in vm_exec () from /usr/local/lib/libruby20.so.0.1
Is this the same segfault that you were receiving?
The i386 machine was running:
ruby 2.0.0p353 (2013-11-22 revision 43784) [i386-openbsd]
$ dmesg |head -n1
OpenBSD 5.4-current (GENERIC.MP) #152: Sat Nov 30 00:50:11 MST 2013
Not sure how much this information helps, though.
Thanks,
Jeremy
