Bug #6006: Fix calculation of HEAP_OBJ_LIMIT and HEAP_BITMAP_LIMIT

• Status: Closed
• Assignee: authorNari (Narihiro Nakamura)
• Target version: 2.0.0
• ruby -v: ruby 2.0.0dev (2012-02-11 trunk 34555) [i686-linux]

[funny_falcon (Yura Sokolov)]

https://github.com/ruby/ruby/pull/92

[funny_falcon (Yura Sokolov)]

Sorry for issue without description.
Details:

HEAP_OBJ_LIMIT used sizeof(struct heaps_slot) while heap is currently allocated with struct heaps_header
HEAP_BITMAP_LIMIT were calculated from HEAP_OBJ_LIMIT/sizeof(uintptr_t) - one Byte for each object, not one Bit.

[authorNari (Narihiro Nakamura)]

• Assignee set to authorNari (Narihiro Nakamura)

[authorNari (Narihiro Nakamura)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r34581.
Yura, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• gc.c (HEAP_OBJ_LIMIT, HEAP_BITMAP_LIMIT): HEAP_OBJ_LIMIT used
  sizeof(struct heaps_slot) while heap is currently allocated
  with struct heaps_header.
  HEAP_BITMAP_LIMIT were calculated from
  HEAP_OBJ_LIMIT/sizeof(uintptr_t) - one Byte for each object,
  not one Bit. [Bug #6006]
  patched by Sokolov Yura. https://github.com/ruby/ruby/pull/92

[naruse (Yui NARUSE)]

• Status changed from Closed to Assigned

Narihiro Nakamura wrote:

This issue was solved with changeset r34581.

• gc.c (HEAP_OBJ_LIMIT, HEAP_BITMAP_LIMIT): HEAP_OBJ_LIMIT used
  sizeof(struct heaps_slot) while heap is currently allocated
  with struct heaps_header.
  HEAP_BITMAP_LIMIT were calculated from
  HEAP_OBJ_LIMIT/sizeof(uintptr_t) - one Byte for each object,
  not one Bit. [Bug #6006]
  patched by Sokolov Yura. https://github.com/ruby/ruby/pull/92

This seems to break CentOS 5.6 64bit.
http://c5664.rubyci.org/~chkbuild/ruby-trunk/log/20120213T130301Z.diff.html.gz

But it doesn't always happen...
http://c5664.rubyci.org/~chkbuild/ruby-trunk/recent.html

[funny_falcon (Yura Sokolov)]

I think it will fix https://github.com/funny-falcon/ruby/commit/5945932de2d682b4fe9b5de3d32b07c2d9b5179d (updated https://github.com/ruby/ruby/pull/92 )

It happens cause on 64-bit platform sometime there should be objs-=2 instead of objs--.
But it easier to compute objs exactly.

[funny_falcon (Yura Sokolov)]

Another one update: calculate p and objs without conditions at all.
https://github.com/ruby/ruby/pull/92.diff

[authorNari (Narihiro Nakamura)]

• Status changed from Assigned to Closed

This issue seems to be solved at r34597.
http://c5664.rubyci.org/~chkbuild/ruby-trunk/log/20120214T110302Z.log.html.gz

Thanks :)

[naruse (Yui NARUSE)]

• Status changed from Closed to Assigned

It still happens: http://c5664.rubyci.org/~chkbuild/ruby-trunk/log/20120214T170302Z.diff.html.gz
I note that I locally applied Yura's revised patch on [ruby-core:42605] to my env (FreeBSD 9.0 64bit),
but it happens segv on miniruby.

compiling ../../ruby/dmyext.c
linking miniruby
../../ruby/tool/mkconfig.rb: [BUG] Segmentation fault
ruby 2.0.0dev (2012-02-14 trunk 34599) [x86_64-freebsd9.0]

-- Control frame information -----------------------------------------------
c:0001 p:0000 s:0002 b:0002 l:0006c8 d:0006c8 TOP

-- C level backtrace information -------------------------------------------
0x44a807 <rb_warning+674> at /home/naruse/obj/ruby/miniruby ../../ruby/error.c:266
0x44a922 <rb_bug+228> at /home/naruse/obj/ruby/miniruby ../../ruby/error.c:285
0x5126f1 <ruby_posix_signal+514> at /home/naruse/obj/ruby/miniruby ../../ruby/signal.c:603
0x800c9d723 <_pthread_sigmask+707> at /lib/libthr.so.3
0x800c9d897 <_pthread_sigmask+1079> at /lib/libthr.so.3
0x7ffffffff003
0x4c1b9d <rb_parser_compile_file+15395> at /home/naruse/obj/ruby/miniruby ../../ruby/parse.y:7236
0x4c60b9 <rb_parser_compile_file+33087> at /home/naruse/obj/ruby/miniruby ../../ruby/parse.y:8515
0x4b0c6e <rb_uv_to_utf8+5753> at /home/naruse/obj/ruby/miniruby parse.c:5182
0x4bd926 <rb_uv_to_utf8+58161> at /home/naruse/obj/ruby/miniruby ../../ruby/parse.y:5798
0x59bd1d <rb_remove_event_hook+1866> at /home/naruse/obj/ruby/miniruby ../../ruby/thread.c:4510
0x59bbc9 <rb_remove_event_hook+1526> at /home/naruse/obj/ruby/miniruby ../../ruby/thread.c:4486
0x4bdab3 <rb_uv_to_utf8+58558> at /home/naruse/obj/ruby/miniruby ../../ruby/parse.y:5824
0x4be023 <rb_parser_compile_file+169> at /home/naruse/obj/ruby/miniruby ../../ruby/parse.y:5954
0x5110f0 <ruby_init_loadpath+15872> at /home/naruse/obj/ruby/miniruby ../../ruby/ruby.c:1631
0x4512ba <rb_ensure+172> at /home/naruse/obj/ruby/miniruby ../../ruby/eval.c:742
0x511202 <ruby_init_loadpath+16146> at /home/naruse/obj/ruby/miniruby ../../ruby/ruby.c:1668
0x510318 <ruby_init_loadpath+12328> at /home/naruse/obj/ruby/miniruby ../../ruby/ruby.c:1404
0x5117c5 <ruby_process_options+133> at /home/naruse/obj/ruby/miniruby ../../ruby/ruby.c:1808
0x44f75f <ruby_options+199> at /home/naruse/obj/ruby/miniruby ../../ruby/eval.c:73
0x41682d <_start+349> at /home/naruse/obj/ruby/miniruby ../../ruby/main.c:38
0x41675e <_start+142> at /home/naruse/obj/ruby/miniruby

-- Other runtime information -----------------------------------------------

• Loaded script: ../../ruby/tool/mkconfig.rb

• Loaded features:

  0 enumerator.so

[NOTE]
You may have encountered a bug in the Ruby interpreter or extension libraries.
Bug reports are welcome.
For details: http://www.ruby-lang.org/bugreport.html

*** Signal 6

Stop in /home/naruse/obj/ruby.

[authorNari (Narihiro Nakamura)]

• Status changed from Assigned to Closed

This issue was solved with changeset r34618.
Yura, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• gc.c (HEAP_BITMAP_LIMIT): HEAP_BITMAP_LIMIT is computed on the
  basis of HEAP_SIZE because it must covers a whole heap block.
  [ruby-trunk - Bug #6006]