| Issue |
131343
|
| Summary |
[LLVM-IR] long/ulong to floating point conversion is giving wrong results for cortex-a9
|
| Labels |
new issue
|
| Assignees |
|
| Reporter |
MrcSnm
|
```c
extern void sceClibPrintf(const char* fmt, ...);
void test_clang()
{
int64_t a = 1000000;
double b = a;
sceClibPrintf("%f\n", b);
sceClibPrintf("%lld\n", a);
}
```
I discovered that while building for PS Vita.
Build with: `clang -c .\test_clang.c -mcpu=cortex-a9 --target=armv7a-unknown-unknown -mfloat-abi=hard`
When running, b is printing 0.0.
And please, of course I've tested that with GCC, so, the problem is not printing. I've discovered that issue because of having wrong numbers.
If I did instead `double b = 1000000`, it works correctly.
![]()
_______________________________________________
llvm-bugs mailing list
llvm-bugs@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
