HANG (any 64-bit app) debug build with no client

[derekbruening]

From [email protected] on March 18, 2009 15:10:57

I'm a little confused as to why I didn't see this before as I run builds
with no client all the time, but it seems to exist all the way back to at
least 9602 (1.2.0).

When I run (on FC10) debug build with no client we assert reading
/proc/maps and hang with a lock issue:

(gdb) bt

#0 0x00000000711a17fc in syscall_ready ()

from /work/dr/tot/exports/lib64/debug/libdynamorio.so

#1 0x00000000000044fc in ?? ()

#2 0x00000000711b8dc5 in get_thread_id () at
/work/dr/tot/opensource/core/linux/os.c:1199
#3 0x00000000710cb78f in deadlock_avoidance_lock (lock=0x7145dac0,
acquired=0, ownable=1)
at /work/dr/tot/opensource/core/utils.c:653

#4 0x00000000710cb98c in mutex_trylock (lock=0x27)

at /work/dr/tot/opensource/core/utils.c:903                           

#5 0x00000000711ba0c4 in mutex_wait_contended_lock (lock=0x7145dac0)

at /work/dr/tot/opensource/core/linux/os.c:5157                       

#6 0x00000000710cba63 in mutex_lock (lock=0x7145dac0)

at /work/dr/tot/opensource/core/utils.c:876                           

#7 0x00000000710ee6a8 in report_dynamorio_problem
(dcontext=0xffffffffffffffff,
dumpcore_flag=8, exception_addr=0x0, report_ebp=0x0,

fmt=0x711f1950 "Internal DynamoRIO Error: %s:%d %s\n(Error occurred @%d

frags)")
at /work/dr/tot/opensource/core/utils.c:1988

#8 0x00000000710c9eec in internal_error (

file=0x711f1850 "/work/dr/tot/opensource/core/utils.c", line=653,     

expr=0x711f21b8 "lock->owner != get_thread_id() && \"deadlock on

recursive mutex_lock"") at /work/dr/tot/opensource/core/utils.c:179

#9 0x00000000710cb7ab in deadlock_avoidance_lock (lock=0x7145b360,

acquired=<value optimized out>, ownable=1) at

/work/dr/tot/opensource/core/utils.c:653
#10 0x00000000710cb98c in mutex_trylock (lock=0x27)

at /work/dr/tot/opensource/core/utils.c:903                           

#11 0x00000000710cba20 in mutex_lock (lock=0x7145b360)

at /work/dr/tot/opensource/core/utils.c:845                           

#12 0x00000000711b846e in maps_iterator_start (iter=0x7fff1d3053a0,
may_alloc=0)
at /work/dr/tot/opensource/core/linux/os.c:4122

#13 0x00000000711b936d in get_memory_info_from_os (pc=0x7fff1d305000
"��\023�>",
base_pc=0x0, size=0x1, prot=0x711a17fc)

at /work/dr/tot/opensource/core/linux/os.c:5068                       

#14 0x00000000711bca33 in is_readable_without_exception_internal (

pc=<value optimized out>, size=<value optimized out>, query_os=1)     

at /work/dr/tot/opensource/core/linux/os.c:2257                       

#15 0x00000000711bca92 in is_readable_without_exception_query_os (

pc=0x27 \<Address 0x27 out of bounds>, size=0)                         

at /work/dr/tot/opensource/core/linux/os.c:2286                       

#16 0x00000000710ee89c in report_dynamorio_problem
(dcontext=0xffffffffffffffff,
dumpcore_flag=128, exception_addr=0x0, report_ebp=0x7fff1d3055d0
"�W0\035�\177",
---Type to continue, or q to quit---

fmt=<value optimized out>) at /work/dr/tot/opensource/core/utils.c:2021

#17 0x00000000711b96d0 in get_library_bounds (name=0x71227822
"libdynamorio.so",
start=0x7fff1d305810, end=0x7fff1d305808, fullpath=0x71472c20 "",
path_size=260)
at /work/dr/tot/opensource/core/linux/os.c:4367

#18 0x00000000711b9b3e in get_dynamo_library_bounds ()

at /work/dr/tot/opensource/core/linux/os.c:4475                       

#19 0x00000000711b9cdc in get_dynamorio_dll_end ()

at /work/dr/tot/opensource/core/linux/os.c:4535                       

#20 0x000000007111c922 in vmm_heap_init () at
/work/dr/tot/opensource/core/heap.c:1253
#21 0x000000007107b784 in dynamorio_app_init ()

at /work/dr/tot/opensource/core/dynamo.c:439

#22 0x0000000000179668 in _init () at
/work/dr/tot/opensource/core/linux/preload.c:144
#23 0x0000003ea3a0eaf0 in _dl_init_internal () from /lib64/ld-linux-x86-64.so.2
#24 0x0000003ea3a00b2a in _dl_start_user () from /lib64/ld-linux-x86-64.so.2
#25 0x0000000000000001 in ?? ()
#26 0x00007fff1d3071a6 in ?? ()
#27 0x0000000000000000 in ?? ()

Original issue: http://code.google.com/p/dynamorio/issues/detail?id=86

[derekbruening]

From [email protected] on March 18, 2009 13:23:41

this is an assert due to an erroneous check in get_library_bounds() that should be
checking for the non-HAVE_PROC_MAPS condition. fixed in r93 .

Status: Verified