-
Notifications
You must be signed in to change notification settings - Fork 13k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
coverage: Store coverage source regions as Span
until codegen
#133418
Merged
Merged
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit
Hold shift + click to select a range
619a272
coverage: Ignore functions that end up having no mappings
Zalathar 87fe7de
coverage: Rename some FFI fields from `span` to `cov_span`
Zalathar b9fb1a6
coverage: Store coverage source regions as `Span` until codegen
Zalathar 2748009
coverage: Identify source files by ID, not by interned filename
Zalathar File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
coverage: Store coverage source regions as
Span
until codegen
- Loading branch information
commit b9fb1a69d2fe1c146cbdf181cf9e0eaf15935799
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
124 changes: 124 additions & 0 deletions
124
compiler/rustc_codegen_llvm/src/coverageinfo/mapgen/spans.rs
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,124 @@ | ||
use rustc_middle::mir::coverage::FunctionCoverageInfo; | ||
use rustc_span::source_map::SourceMap; | ||
use rustc_span::{BytePos, Pos, SourceFile, Span}; | ||
use tracing::debug; | ||
|
||
use crate::coverageinfo::ffi; | ||
use crate::coverageinfo::mapgen::LocalFileId; | ||
|
||
/// Converts the span into its start line and column, and end line and column. | ||
/// | ||
/// Line numbers and column numbers are 1-based. Unlike most column numbers emitted by | ||
/// the compiler, these column numbers are denoted in **bytes**, because that's what | ||
/// LLVM's `llvm-cov` tool expects to see in coverage maps. | ||
/// | ||
/// Returns `None` if the conversion failed for some reason. This shouldn't happen, | ||
/// but it's hard to rule out entirely (especially in the presence of complex macros | ||
/// or other expansions), and if it does happen then skipping a span or function is | ||
/// better than an ICE or `llvm-cov` failure that the user might have no way to avoid. | ||
pub(crate) fn make_coverage_span( | ||
file_id: LocalFileId, | ||
source_map: &SourceMap, | ||
fn_cov_info: &FunctionCoverageInfo, | ||
file: &SourceFile, | ||
span: Span, | ||
) -> Option<ffi::CoverageSpan> { | ||
let span = ensure_non_empty_span(source_map, fn_cov_info, span)?; | ||
|
||
let lo = span.lo(); | ||
let hi = span.hi(); | ||
|
||
// Column numbers need to be in bytes, so we can't use the more convenient | ||
// `SourceMap` methods for looking up file coordinates. | ||
let line_and_byte_column = |pos: BytePos| -> Option<(usize, usize)> { | ||
let rpos = file.relative_position(pos); | ||
let line_index = file.lookup_line(rpos)?; | ||
let line_start = file.lines()[line_index]; | ||
// Line numbers and column numbers are 1-based, so add 1 to each. | ||
Some((line_index + 1, (rpos - line_start).to_usize() + 1)) | ||
}; | ||
|
||
let (mut start_line, start_col) = line_and_byte_column(lo)?; | ||
let (mut end_line, end_col) = line_and_byte_column(hi)?; | ||
|
||
// Apply an offset so that code in doctests has correct line numbers. | ||
// FIXME(#79417): Currently we have no way to offset doctest _columns_. | ||
start_line = source_map.doctest_offset_line(&file.name, start_line); | ||
end_line = source_map.doctest_offset_line(&file.name, end_line); | ||
|
||
check_coverage_span(ffi::CoverageSpan { | ||
file_id: file_id.as_u32(), | ||
start_line: start_line as u32, | ||
start_col: start_col as u32, | ||
end_line: end_line as u32, | ||
end_col: end_col as u32, | ||
}) | ||
} | ||
|
||
fn ensure_non_empty_span( | ||
source_map: &SourceMap, | ||
fn_cov_info: &FunctionCoverageInfo, | ||
span: Span, | ||
) -> Option<Span> { | ||
if !span.is_empty() { | ||
return Some(span); | ||
} | ||
|
||
let lo = span.lo(); | ||
let hi = span.hi(); | ||
|
||
// The span is empty, so try to expand it to cover an adjacent '{' or '}', | ||
// but only within the bounds of the body span. | ||
let try_next = hi < fn_cov_info.body_span.hi(); | ||
let try_prev = fn_cov_info.body_span.lo() < lo; | ||
if !(try_next || try_prev) { | ||
return None; | ||
} | ||
|
||
source_map | ||
.span_to_source(span, |src, start, end| try { | ||
// We're only checking for specific ASCII characters, so we don't | ||
// have to worry about multi-byte code points. | ||
if try_next && src.as_bytes()[end] == b'{' { | ||
Some(span.with_hi(hi + BytePos(1))) | ||
} else if try_prev && src.as_bytes()[start - 1] == b'}' { | ||
Some(span.with_lo(lo - BytePos(1))) | ||
} else { | ||
None | ||
} | ||
}) | ||
.ok()? | ||
} | ||
|
||
/// If `llvm-cov` sees a source region that is improperly ordered (end < start), | ||
/// it will immediately exit with a fatal error. To prevent that from happening, | ||
/// discard regions that are improperly ordered, or might be interpreted in a | ||
/// way that makes them improperly ordered. | ||
fn check_coverage_span(cov_span: ffi::CoverageSpan) -> Option<ffi::CoverageSpan> { | ||
let ffi::CoverageSpan { file_id: _, start_line, start_col, end_line, end_col } = cov_span; | ||
|
||
// Line/column coordinates are supposed to be 1-based. If we ever emit | ||
// coordinates of 0, `llvm-cov` might misinterpret them. | ||
let all_nonzero = [start_line, start_col, end_line, end_col].into_iter().all(|x| x != 0); | ||
// Coverage mappings use the high bit of `end_col` to indicate that a | ||
// region is actually a "gap" region, so make sure it's unset. | ||
let end_col_has_high_bit_unset = (end_col & (1 << 31)) == 0; | ||
// If a region is improperly ordered (end < start), `llvm-cov` will exit | ||
// with a fatal error, which is inconvenient for users and hard to debug. | ||
let is_ordered = (start_line, start_col) <= (end_line, end_col); | ||
|
||
if all_nonzero && end_col_has_high_bit_unset && is_ordered { | ||
Some(cov_span) | ||
} else { | ||
debug!( | ||
?cov_span, | ||
?all_nonzero, | ||
?end_col_has_high_bit_unset, | ||
?is_ordered, | ||
"Skipping source region that would be misinterpreted or rejected by LLVM" | ||
); | ||
// If this happens in a debug build, ICE to make it easier to notice. | ||
debug_assert!(false, "Improper source region: {cov_span:?}"); | ||
None | ||
} | ||
} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Remark:
:ferrisAware:
but this is fine.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, I'm very much aware of why
BytePos(1)
is bad news in general; in the past I have fixedBytePos(1)
bugs in the previous incarnation of this code.In this very specific set of circumstances, we have confirmed in advance that the code point we're skipping over is a single byte (ASCII
{
or}
), so adjusting by 1 to skip that character really is OK.