[PATCH] D95987: [llvm-cov] Reset ExecutionCount only if first line segment is start of a new region.

[Zequan Wu via Phabricator via llvm-commits]

zequanwu added a comment.

In D95987#2541091 <https://reviews.llvm.org/D95987#2541091>, @pirama wrote:

> Thanks for the patch Zequan.  This fixes the C++ test case that prompted this issue.
>
> The underlying Rust issue is not fixed though:
>
>    1|       |fn notatest() -> Result<(), std::env::VarError> {
>    2|      1|        let result = std::env::var("MEOW")
>    3|      0|                    .or(Ok("a".to_string()))?;
>    4|      1|            println!("{}", result);
>    5|      1|                Ok(())
>    6|      1|}
>    7|       | 
>    8|      1|fn main() {
>    9|      1|        println!("Hello, world! {:?}", notatest());
>   10|      1|}
>
> Here're the regions from `-debug-only=coverage-mapping`:
>
>   Segment at 2:13 (count = 1), RegionEntry
>   Segment at 2:19 (count = 0), Skipped                                                          
>   Segment at 2:22 (count = 1), RegionEntry                                                      
>   Segment at 3:45 (count = 0), RegionEntry
>   Segment at 3:46 (count = 0), Skipped
>   Segment at 4:13 (count = 1), RegionEntry
>   Segment at 5:23 (count = 0), Skipped
>   Segment at 6:1 (count = 1), RegionEntry
>   Segment at 6:2 (count = 0), Skipped
>   Segment at 8:11 (count = 1), RegionEntry
>   Segment at 10:2 (count = 0), Skipped
>
> The ternary operator case in the C++ test case has a gap region corresponding to the `?`.  For the Rust `?` operator, there is no similar gap region.  In fact, the rust frontend associates the counter corresponding to the error path of the `?` operator to the `?` token .  Is there a better sequence of Segments that the Rust frontend can emit to handle this case?

Sorry, I don't know about Rust. I just read the `?` operator doc on https://doc.rust-lang.org/edition-guide/rust-2018/error-handling-and-panics/the-question-mark-operator-for-easier-error-handling.html.

For the example you provided, I am not sure if it's a bug or not. `Segment at 3:45 (count = 0), RegionEntry` corresponds to the error path of the `?` operator, which means error path is not executed. So, the 0 count looks reasonable. If we mark line 3 as executed regardless of which path is chosen, we may lost the information of path choices.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D95987/new/

https://reviews.llvm.org/D95987