bpo-1635741: Fix potential refleaks in binascii module#18613
Uh oh!
There was an error while loading. Please reload this page.
Conversation
Codecov Report
@@ Coverage Diff @@## master #18613 +/- ## ========================================== + Coverage 82.06% 82.13% +0.06% ========================================== Files 1955 1955 Lines 584080 584621 +541 Branches 44458 44484 +26 ========================================== + Hits 479327 480176 +849 + Misses 95124 94794 -330 - Partials 9629 9651 +22
Continue to review full report at Codecov.
|
Modules/binascii.c Outdated
| binascii_clear(PyObject*m) | ||
| { | ||
| binascii_state*state=get_binascii_state(m); | ||
| if (state==NULL){ |
There was a problem hiding this comment.
I don't think that state can be NULL. If it's NULL, it's a bug in Python. No? Same remark in binascii_traverse().
There was a problem hiding this comment.
I do some debug in my vm:
Breakpoint 1, binascii_traverse (module=0x7ffff7e803d0, visit=0x4630b6 <bad_traverse_test>, arg=0x0) at /temp/shihai/cpython/Modules/binascii.c:1654 1654{(gdb) n 1655 binascii_state *state = get_binascii_state(module); (gdb) n 1656 if (state == NULL){(gdb) n 1657 return -1; (gdb) p state $5 = (binascii_state *) 0x0 (gdb) c Continuing. Breakpoint 3, binascii_exec (module=0x7ffff7e803d0) at /temp/shihai/cpython/Modules/binascii.c:1615 1615 binascii_exec(PyObject *module){What's the reason?
the malloc operation of md_state will be called after first calling binascii_traverse.
https://github.com/python/cpython/blob/master/Objects/moduleobject.c#L399.
Modules/binascii.c Outdated
| }; | ||
| staticint | ||
| binascii_traverse(PyObject*m, visitprocvisit, void*arg) |
There was a problem hiding this comment.
nitpick: I would prefer to use longer variable name than just "m". Rename "m" to "mod" or "module". Same remark if following functions.
There was a problem hiding this comment.
ok, no probleam, i will update it.
Modules/binascii.c Outdated
| binascii_traverse(PyObject*m, visitprocvisit, void*arg) | ||
| { | ||
| binascii_state*state=get_binascii_state(m); | ||
| if (state==NULL){ |
There was a problem hiding this comment.
I will update this behavior soon.
vstinner commented Mar 2, 2020
I created https://bugs.python.org/issue39824 to change the md_state==NULL case. |
4 participants
https://bugs.python.org/issue1635741