gh-148390: fix undefined behavior of memoryview(...).cast("?")#148454
gh-148390: fix undefined behavior of memoryview(...).cast("?")#148454picnixz wants to merge 2 commits intopython:mainfrom
memoryview(...).cast("?")#148454Conversation
picnixz
requested review from
encukou,
serhiy-storchaka,
skirpichev and
vstinner
StanFromIreland
left a comment
StanFromIreland
left a comment
There was a problem hiding this comment.
If the UB is fixed, please remove it from the suppressions list:
cpython/Tools/ubsan/suppressions.txt
Lines 12 to 13 in 3ab94d6
skirpichev
left a comment
skirpichev
left a comment
There was a problem hiding this comment.
Looks good.
Though, maybe move macro closer to unpack_single() definition?
The problem is that it's being used in two places so... I preferred moving it where we had "helpers". |
| } | ||
|
|
||
| // Return 0 if PTR represents "false", and 1 otherwise. | ||
| #define UNPACK_TO_BOOL(PTR) (memcmp((PTR), &(_Bool){0}, sizeof(_Bool)) != 0) |
There was a problem hiding this comment.
What version of C supports &(_Bool){0}? It was not supported prior to C99, but does C99 support this?
There was a problem hiding this comment.
It's supported since C99: https://en.cppreference.com/w/c/language/compound_literal.html.
There was a problem hiding this comment.
Great. But it seems that the unnamed _Bool object has automatic storage duration. I.e. it is created on the stack. Would not it be more efficient to use constant static object initialized only once?
4 participants
memoryview.cast("?")has undocumented undefined behavior #148390