Seq IndexDefect leaks memeory in `arc` and `orc`

[doongjohn]

Description

Seq IndexDefect leaks memeory in arc and orc.

try:
  var a = @[0]
  a[1] = 0 # <-- IndexDefect this leaks

  # raise newException(IndexDefect, "") # <-- this does not leak
  # echo 10 div 0  # <-- this does not leak
  # assert(false)  # <-- this does not leak
except Defect as err:
  echo "defect: ", err.msg

Wrapping it in a proc does not help

proc a =
  try:
    var a = @[0]
    a[1] = 0
  except Defect as err:
    echo "defect: ", err.msg
a()

Nim Version

Nim Compiler Version 2.1.1 [Linux: amd64]
Compiled at 2024-06-12
Copyright (c) 2006-2024 by Andreas Rumpf

git hash: 0b5a938
active boot switches: -d:release

Current Output

nim c --mm:arc -d:useMalloc main.nim && valgrind --leak-check=full --show-leak-kinds=all ./main

==5609== Memcheck, a memory error detector
==5609== Copyright (C) 2002-2024, and GNU GPL'd, by Julian Seward et al.
==5609== Using Valgrind-3.23.0 and LibVEX; rerun with -h for copyright info
==5609== Command: ./main
==5609== 
defect: index 1 not in 0 .. 0
==5609== 
==5609== HEAP SUMMARY:
==5609==     in use at exit: 16 bytes in 1 blocks
==5609==   total heap usage: 7 allocs, 6 frees, 1,218 bytes allocated
==5609== 
==5609== 16 bytes in 1 blocks are still reachable in loss record 1 of 1
==5609==    at 0x484CA2F: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==5609==    by 0x405C88: alloc0Impl__system_u1732 (in /home/..../src/main)
==5609==    by 0x405CAA: allocShared0Impl__system_u1745 (in /home/..../src/main)
==5609==    by 0x405CDF: alignedAlloc0__system_u1910 (in /home/..../src/main)
==5609==    by 0x406686: newSeqPayload (in /home/..../src/main)
==5609==    by 0x407211: NimMainModule (in /home/..../src/main)
==5609==    by 0x407153: NimMainInner (in /home/..../src/main)
==5609==    by 0x407164: NimMain (in /home/..../src/main)
==5609==    by 0x40719E: main (in /home/..../src/main)
==5609== 
==5609== LEAK SUMMARY:
==5609==    definitely lost: 0 bytes in 0 blocks
==5609==    indirectly lost: 0 bytes in 0 blocks
==5609==      possibly lost: 0 bytes in 0 blocks
==5609==    still reachable: 16 bytes in 1 blocks
==5609==         suppressed: 0 bytes in 0 blocks
==5609== 
==5609== For lists of detected and suppressed errors, rerun with: -s
==5609== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

Expected Output

no leak

Possible Solution

No response

Additional Information

refc does not leak.

[alex65536]

Isn't it the same as #23714?

[doongjohn]

@alex65536 #23714 does not "leak" even if I do multiple try except. However this issue actually leaks:

# with one try except
try:
  var a = @[0]
  a[1] = 0 # <-- IndexDefect
except Defect as err:
  echo "defect: ", err.msg

==527== Memcheck, a memory error detector
==527== Copyright (C) 2002-2024, and GNU GPL'd, by Julian Seward et al.
==527== Using Valgrind-3.23.0 and LibVEX; rerun with -h for copyright info
==527== Command: ./app
==527== 
defect: index 1 not in 0 .. 0
==527== 
==527== HEAP SUMMARY:
==527==     in use at exit: 16,400 bytes in 2 blocks
==527==   total heap usage: 8 allocs, 6 frees, 17,610 bytes allocated
==527== 
==527== 16 bytes in 1 blocks are still reachable in loss record 1 of 2
==527==    at 0x484CA2F: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==527==    by 0x4082F7: alloc0Impl__system_u1732 (in /home/..../app)
==527==    by 0x408319: allocShared0Impl__system_u1745 (in /home/..../app)
==527==    by 0x40834E: alignedAlloc0__system_u1916 (in /home/..../app)
==527==    by 0x408CF5: newSeqPayload (in /home/..../app)
==527==    by 0x40988A: NimMainModule (in /home/..../app)
==527==    by 0x4097CC: NimMainInner (in /home/..../app)
==527==    by 0x4097DD: NimMain (in /home/..../app)
==527==    by 0x409817: main (in /home/..../app)
==527== 
==527== 16,384 bytes in 1 blocks are still reachable in loss record 2 of 2
==527==    at 0x4845794: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==527==    by 0x4025C2: allocImpl__system_u1730 (in /home/..../app)
==527==    by 0x4025E4: allocSharedImpl (in /home/..../app)
==527==    by 0x40290E: init__system_u3292 (in /home/..../app)
==527==    by 0x404959: registerCycle__system_u3415 (in /home/..../app)
==527==    by 0x404B16: rememberCycle__system_u3455 (in /home/..../app)
==527==    by 0x405F71: nimDecRefIsLastCyclicDyn (in /home/..../app)
==527==    by 0x408F5A: eqdestroy___main_u133 (in /home/..../app)
==527==    by 0x4099BE: NimMainModule (in /home/..../app)
==527==    by 0x4097CC: NimMainInner (in /home/..../app)
==527==    by 0x4097DD: NimMain (in /home/..../app)
==527==    by 0x409817: main (in /home/..../app)
==527== 
==527== LEAK SUMMARY:
==527==    definitely lost: 0 bytes in 0 blocks
==527==    indirectly lost: 0 bytes in 0 blocks
==527==      possibly lost: 0 bytes in 0 blocks
==527==    still reachable: 16,400 bytes in 2 blocks
==527==         suppressed: 0 bytes in 0 blocks

# with two try except
try:
  var a = @[0]
  a[1] = 0 # <-- IndexDefect
except Defect as err:
  echo "defect: ", err.msg

try:
  var a = @[0]
  a[1] = 0 # <-- IndexDefect
except Defect as err:
  echo "defect: ", err.msg

==567== Memcheck, a memory error detector
==567== Copyright (C) 2002-2024, and GNU GPL'd, by Julian Seward et al.
==567== Using Valgrind-3.23.0 and LibVEX; rerun with -h for copyright info
==567== Command: ./app
==567== 
defect: index 1 not in 0 .. 0
defect: index 1 not in 0 .. 0
==567== 
==567== HEAP SUMMARY:
==567==     in use at exit: 16,416 bytes in 3 blocks
==567==   total heap usage: 14 allocs, 11 frees, 17,812 bytes allocated
==567== 
==567== 16 bytes in 1 blocks are still reachable in loss record 1 of 3
==567==    at 0x484CA2F: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==567==    by 0x4082F7: alloc0Impl__system_u1732 (in /home/..../app)
==567==    by 0x408319: allocShared0Impl__system_u1745 (in /home/..../app)
==567==    by 0x40834E: alignedAlloc0__system_u1916 (in /home/..../app)
==567==    by 0x408CF5: newSeqPayload (in /home/..../app)
==567==    by 0x40988A: NimMainModule (in /home/..../app)
==567==    by 0x4097CC: NimMainInner (in /home/..../app)
==567==    by 0x4097DD: NimMain (in /home/..../app)
==567==    by 0x409817: main (in /home/..../app)
==567== 
==567== 16 bytes in 1 blocks are still reachable in loss record 2 of 3
==567==    at 0x484CA2F: calloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==567==    by 0x4082F7: alloc0Impl__system_u1732 (in /home/..../app)
==567==    by 0x408319: allocShared0Impl__system_u1745 (in /home/..../app)
==567==    by 0x40834E: alignedAlloc0__system_u1916 (in /home/..../app)
==567==    by 0x408CF5: newSeqPayload (in /home/..../app)
==567==    by 0x409A0F: NimMainModule (in /home/..../app)
==567==    by 0x4097CC: NimMainInner (in /home/..../app)
==567==    by 0x4097DD: NimMain (in /home/..../app)
==567==    by 0x409817: main (in /home/..../app)
==567== 
==567== 16,384 bytes in 1 blocks are still reachable in loss record 3 of 3
==567==    at 0x4845794: malloc (in /usr/libexec/valgrind/vgpreload_memcheck-amd64-linux.so)
==567==    by 0x4025C2: allocImpl__system_u1730 (in /home/..../app)
==567==    by 0x4025E4: allocSharedImpl (in /home/..../app)
==567==    by 0x40290E: init__system_u3292 (in /home/..../app)
==567==    by 0x404959: registerCycle__system_u3415 (in /home/..../app)
==567==    by 0x404B16: rememberCycle__system_u3455 (in /home/..../app)
==567==    by 0x405F71: nimDecRefIsLastCyclicDyn (in /home/..../app)
==567==    by 0x408F5A: eqdestroy___main_u139 (in /home/..../app)
==567==    by 0x4099BE: NimMainModule (in /home/..../app)
==567==    by 0x4097CC: NimMainInner (in /home/..../app)
==567==    by 0x4097DD: NimMain (in /home/..../app)
==567==    by 0x409817: main (in /home/..../app)
==567== 
==567== LEAK SUMMARY:
==567==    definitely lost: 0 bytes in 0 blocks
==567==    indirectly lost: 0 bytes in 0 blocks
==567==      possibly lost: 0 bytes in 0 blocks
==567==    still reachable: 16,416 bytes in 3 blocks
==567==         suppressed: 0 bytes in 0 blocks
==567== 
==567== For lists of detected and suppressed errors, rerun with: -s
==567== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

[alex65536]

Oh, this is a valid issue indeed :)

Even though catching IndexDefect (and any other Defect) IMO rarely seems to be a good idea, it still looks like a bug.

[beef331]

rarely seems to be a good idea

--panics:on means it's just not a good idea unless you're testing.