Memory leak in 1.8.6-p287

[Mike Perham]

We've noticed a pretty severe memory leak since upgrading from p114 to
p287. I ran valgrind on my test suite and got this error:

==24626== 352 (48 direct, 304 indirect) bytes in 2 blocks are definitely
lost in loss record 13 of 56
==24626== at 0x4A19DDB: malloc (vg_replace_malloc.c:207)
==24626== by 0x42CFE3: ruby_xmalloc (gc.c:110)
==24626== by 0x47062B: st_init_table_with_size (st.c:159)
==24626== by 0x48BCC8: rb_class_boot (class.c:31)
==24626== by 0x418649: rb_call0 (eval.c:5870)
==24626== by 0x419298: rb_call (eval.c:6117)
==24626== by 0x419DC6: vafuncall (eval.c:6194)
==24626== by 0x41A032: rb_funcall (eval.c:6211)
==24626== by 0x44198E: rb_obj_alloc (object.c:1534)
==24626== by 0x442BAC: rb_obj_dup (object.c:260)
==24626== by 0x418649: rb_call0 (eval.c:5870)
==24626== by 0x419298: rb_call (eval.c:6117)
==24626== by 0x41504D: rb_eval (eval.c:3490)
==24626== by 0x416F45: rb_eval (eval.c:3700)
==24626== by 0x414F57: rb_eval (eval.c:3484)

Sorry I can't decipher the output very well. Can anyone else figure out
the stack and what this means? I would be happy to get more detail,
just let me know what you need.

mike

 

[Roger Pack]

Do you have any test code to re-create it?

We've noticed a pretty severe memory leak since upgrading from p114 to
p287. I ran valgrind on my test suite and got this error:

==24626== 352 (48 direct, 304 indirect) bytes in 2 blocks are definitely
lost in loss record 13 of 56
==24626== at 0x4A19DDB: malloc (vg_replace_malloc.c:207)
==24626== by 0x42CFE3: ruby_xmalloc (gc.c:110)
==24626== by 0x47062B: st_init_table_with_size (st.c:159)