[llvm-bugs] [Bug 42811] New: transparent union pointers not understood by static analyzer
via llvm-bugs
llvm-bugs at lists.llvm.org
Mon Jul 29 06:24:31 PDT 2019
https://bugs.llvm.org/show_bug.cgi?id=42811
Bug ID: 42811
Summary: transparent union pointers not understood by static
analyzer
Product: clang
Version: 8.0
Hardware: PC
OS: Linux
Status: NEW
Severity: enhancement
Priority: P
Component: Static Analyzer
Assignee: dcoughlin at apple.com
Reporter: equinox-llvm at diac24.net
CC: dcoughlin at apple.com, llvm-bugs at lists.llvm.org
The static analyzer doesn't seem to understand transparent union arguments,
particularly if the union members are pointers:
struct s1 {
int x;
};
struct s2 {
int y;
};
union arg {
struct s1 *s1;
struct s2 *s2;
} __attribute__((transparent_union));
void set1(union arg arg)
{
arg.s1->x = 0;
}
void set2(union arg arg)
{
arg.s2->y = 0;
}
int test1(void)
{
struct s2 s2;
set1(&s2);
return s2.y ? 1 : 0;
}
int test2(void)
{
struct s2 s2;
set2(&s2);
return s2.y ? 1 : 0;
}
$ scan-build-8 clang-8 -c -o transp_union_sa.o transp_union_sa.c
scan-build: Using '/usr/lib/llvm-8/bin/clang' for static analysis
transp_union_sa.c:27:9: warning: Branch condition evaluates to a garbage value
return s2.y ? 1 : 0;
^~~~
transp_union_sa.c:34:9: warning: Branch condition evaluates to a garbage value
return s2.y ? 1 : 0;
^~~~
2 warnings generated.
Ideally, neither test1 nor test2 should generate a warning; the test1 case is
much harder to get right though since it'd need the SA to understand that the x
and y struct fields for s1 and s2 are in the same place. test2 is much easier
and should definitely not result in a warning.
$ clang-8 --version
clang version 8.0.1-+rc4-1 (tags/RELEASE_801/rc4)
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin
(Debian)
--
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20190729/1b26574d/attachment.html>
More information about the llvm-bugs
mailing list