[EXT_EP-11845] For a float if statement followed by divide, compiler generated code performs divide first Created: 30/Jul/24 Updated: 13/Aug/24 Resolved: 13/Aug/24 |
|
| Status: | Declined |
| Project: | Embedded Software & Tools |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Bug | Priority: | Not Prioritized |
| Reporter: | TI User | Assignee: | TI User |
| Resolution: | Duplicate | Votes: | 0 |
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Product: | Code Generation Tools |
| Internal ID: | CODEGEN-12700 |
| Forum URL: | https://e2e.ti.com/e2eprivate/continental/continental-adcu5/f/1194/t/1392796 |
| Found In Release: | ARMCLANG_4.0.0.LTS ARMCLANG_3.2.2.LTS |
| Affected Platform/Device: | default |
| Decline Reason: | This functionality is implemented as a new feature in the 4.0.0.LTS release of the tools via "-ffp-exception-behavior=strict|maytrap" option which can be used to prevent floating point exceptions by preventing floating point operations from being moved outside of conditional control flow.
Please see the release notes for more information on how to use this option: https://software-dl.ti.com/codegen/esd/cgt_public_sw/ARM_LLVM/4.0.0.LTS/README.html#suppress-floating-point-speculation-with--ffp-exceptions-compiler-option |
| Description |
|
The related test case has these lines ... if(float1 > CP_MIN)
{
f_array[1u] = (1.f / *f_Info);
}
Code generated by the compiler performs the division before the if statement. For values where that divide causes an FPU exception to be raised, the if statement should keep the divide from happening. But because the divide happens no matter what, the FPU exception is incorrectly raised for those values. |