LLVM's code coverage mapping format is used to provide code coverage
analysis using LLVM's and Clang's instrumenation based profiling
(Clang's -fprofile-instr-generate option).
This document is aimed at those who use LLVM's code coverage mapping to provide code coverage analysis for their own programs, and for those who would like to know how it works under the hood. A prior knowledge of how Clang's profile guided optimization works is useful, but not required.
We start by showing how to use LLVM and Clang for code coverage analysis, then we briefly describe LLVM's code coverage mapping format and the way that Clang and LLVM's code coverage tool work with this format. After the basics are down, more advanced features of the coverage mapping format are discussed - such as the data structures, LLVM IR representation and the binary encoding.
Here's a short story that describes how to generate code coverage overview for a sample source file called test.c.
First, compile an instrumented version of your program using Clang's
-fprofile-instr-generateoption with the additional-fcoverage-mappingoption:clang -o test -fprofile-instr-generate -fcoverage-mapping test.cThen, run the instrumented binary. The runtime will produce a file called default.profraw containing the raw profile instrumentation data:
./testAfter that, merge the profile data using the llvm-profdata tool:
llvm-profdata merge -o test.profdata default.profrawFinally, run LLVM's code coverage tool (llvm-cov) to produce the code coverage overview for the sample source file:
llvm-cov show ./test -instr-profile=test.profdata test.c
LLVM's code coverage mapping format is designed to be a self contained data format, that can be embedded into the LLVM IR and object files. It's described in this document as a mapping format because its goal is to store the data that is required for a code coverage tool to map between the specific source ranges in a file and the execution counts obtained after running the instrumented version of the program.
The mapping data is used in two places in the code coverage process:
- When clang compiles a source file with
-fcoverage-mapping, it generates the mapping information that describes the mapping between the source ranges and the profiling instrumentation counters. This information gets embedded into the LLVM IR and conveniently ends up in the final executable file when the program is linked. - It is also used by llvm-cov - the mapping information is extracted from an object file and is used to associate the execution counts (the values of the profile instrumentation counters), and the source ranges in a file. After that, the tool is able to generate various code coverage reports for the program.
The coverage mapping format aims to be a "universal format" that would be suitable for usage by any frontend, and not just by Clang. It also aims to provide the frontend the possibility of generating the minimal coverage mapping data in order to reduce the size of the IR and object files - for example, instead of emitting mapping information for each statement in a function, the frontend is allowed to group the statements with the same execution count into regions of code, and emit the mapping information only for those regions.
The remainder of this guide is meant to give you insight into the way the coverage mapping format works.
The coverage mapping format operates on a per-function level as the profile instrumentation counters are associated with a specific function. For each function that requires code coverage, the frontend has to create coverage mapping data that can map between the source code ranges and the profile instrumentation counters for that function.
The function's coverage mapping data contains an array of mapping regions. A mapping region stores the source code range that is covered by this region, the file id, the coverage mapping counter and the region's kind. There are several kinds of mapping regions:
Code regions associate portions of source code and coverage mapping counters. They make up the majority of the mapping regions. They are used by the code coverage tool to compute the execution counts for lines, highlight the regions of code that were never executed, and to obtain the various code coverage statistics for a function. For example:
int main(int argc, const char *argv[]) { // Code Region from 1:40 to 9:2 if (argc > 1) { // Code Region from 3:17 to 5:4 printf("%s\n", argv[1]); } else { // Code Region from 5:10 to 7:4 printf("\n"); } return 0; }
Skipped regions are used to represent source ranges that were skipped by Clang's preprocessor. They don't associate with coverage mapping counters, as the frontend knows that they are never executed. They are used by the code coverage tool to mark the skipped lines inside a function as non-code lines that don't have execution counts. For example:
int main() { // Code Region from 1:12 to 6:2 #ifdef DEBUG // Skipped Region from 2:1 to 4:2 printf("Hello world"); #endif return 0; }
Expansion regions are used to represent Clang's macro expansions. They have an additional property - expanded file id. This property can be used by the code coverage tool to find the mapping regions that are created as a result of this macro expansion, by checking if their file id matches the expanded file id. They don't associate with coverage mapping counters, as the code coverage tool can determine the execution count for this region by looking up the execution count of the first region with a corresponding file id. For example:
int func(int x) { #define MAX(x,y) ((x) > (y)? (x) : (y)) return MAX(x, 42); // Expansion Region from 3:10 to 3:13 }
The source range record contains the starting and ending location of a certain mapping region. Both locations include the line and the column numbers.
The file id an integer value that tells us in which source file or macro expansion is this region located. It enables Clang to produce mapping information for the code defined inside macros, like this example demonstrates:
void func(const char *str) { // Code Region from 1:28 to 6:2 with file id 0 #define PUT printf("%s\n", str) // 2 Code Regions from 2:15 to 2:34 with file ids 1 and 2 if(*str) PUT; // Expansion Region from 4:5 to 4:8 with file id 0 that expands a macro with file id 1 PUT; // Expansion Region from 5:3 to 5:6 with file id 0 that expands a macro with file id 2 }
A coverage mapping counter can represents a reference to the profile instrumentation counter. The execution count for a region with such counter is determined by looking up the value of the corresponding profile instrumentation counter.
It can also represent a binary arithmetical expression that operates on coverage mapping counters or other expressions. The execution count for a region with an expression counter is determined by evaluating the expression's arguments and then adding them together or subtracting them from one another. In the example below, a subtraction expression is used to compute the execution count for the compound statement that follows the else keyword:
int main(int argc, const char *argv[]) { // Region's counter is a reference to the profile counter #0 if (argc > 1) { // Region's counter is a reference to the profile counter #1 printf("%s\n", argv[1]); } else { // Region's counter is an expression (reference to the profile counter #0 - reference to the profile counter #1) printf("\n"); } return 0; }
Finally, a coverage mapping counter can also represent an execution count of of zero. The zero counter is used to provide coverage mapping for unreachable statements and expressions, like in the example below:
int main() { return 0; printf("Hello world!\n"); // Unreachable region's counter is zero }
The zero counters allow the code coverage tool to display proper line execution counts for the unreachable lines and highlight the unreachable code. Without them, the tool would think that those lines and regions were still executed, as it doesn't possess the frontend's knowledge.
The coverage mapping data is stored in the LLVM IR using a single global constant structure variable called __llvm_coverage_mapping with the __llvm_covmap section specifier.
For example, let’s consider a C file and how it gets compiled to LLVM:
int foo() {
return 42;
}
int bar() {
return 13;
}The coverage mapping variable generated by Clang has 3 fields:
- Coverage mapping header.
- An array of function records.
- Coverage mapping data which is an array of bytes. Zero paddings are added at the end to force 8 byte alignment.
@__llvm_coverage_mapping = internal constant { { i32, i32, i32, i32 }, [2 x { i64, i32, i64 }], [40 x i8] }
{
{ i32, i32, i32, i32 } ; Coverage map header
{
i32 2, ; The number of function records
i32 20, ; The length of the string that contains the encoded translation unit filenames
i32 20, ; The length of the string that contains the encoded coverage mapping data
i32 1, ; Coverage mapping format version
},
[2 x { i64, i32, i64 }] [ ; Function records
{ i64, i32, i64 } {
i64 0x5cf8c24cdb18bdac, ; Function's name MD5
i32 9, ; Function's encoded coverage mapping data string length
i64 0 ; Function's structural hash
},
{ i64, i32, i64 } {
i64 0xe413754a191db537, ; Function's name MD5
i32 9, ; Function's encoded coverage mapping data string length
i64 0 ; Function's structural hash
}],
[40 x i8] c"..." ; Encoded data (dissected later)
}, section "__llvm_covmap", align 8The function record layout has evolved since version 1. In version 1, the function record for foo is defined as follows:
{ i8*, i32, i32, i64 } { i8* getelementptr inbounds ([3 x i8]* @__profn_foo, i32 0, i32 0), ; Function's name
i32 3, ; Function's name length
i32 9, ; Function's encoded coverage mapping data string length
i64 0 ; Function's structural hash
}The coverage mapping header has the following fields:
- The number of function records.
- The length of the string in the third field of __llvm_coverage_mapping that contains the encoded translation unit filenames.
- The length of the string in the third field of __llvm_coverage_mapping that contains the encoded coverage mapping data.
- The format version. The current version is 2 (encoded as a 1).
A function record is a structure of the following type:
{ i64, i32, i64 }It contains function name's MD5, the length of the encoded mapping data for that function, and function's structural hash value.
The encoded data is stored in a single string that contains the encoded filenames used by this translation unit and the encoded coverage mapping data for each function in this translation unit.
The encoded data has the following structure:
[filenames, coverageMappingDataForFunctionRecord0, coverageMappingDataForFunctionRecord1, ..., padding]
If necessary, the encoded data is padded with zeroes so that the size of the data string is rounded up to the nearest multiple of 8 bytes.
Here's an overview of the encoded data that was stored in the IR for the coverage mapping sample that was shown earlier:
The IR contains the following string constant that represents the encoded coverage mapping data for the sample translation unit:
c"\01\12/Users/alex/test.c\01\00\00\01\01\01\0C\02\02\01\00\00\01\01\04\0C\02\02\00\00"
The string contains values that are encoded in the LEB128 format, which is used throughout for storing integers. It also contains a string value.
The length of the substring that contains the encoded translation unit filenames is the value of the second field in the __llvm_coverage_mapping structure, which is 20, thus the filenames are encoded in this string:
c"\01\12/Users/alex/test.c"
This string contains the following data:
- Its first byte has a value of
0x01. It stores the number of filenames contained in this string. - Its second byte stores the length of the first filename in this string.
- The remaining 18 bytes are used to store the first filename.
- Its first byte has a value of
The length of the substring that contains the encoded coverage mapping data for the first function is the value of the third field in the first structure in an array of function records stored in the third field of the __llvm_coverage_mapping structure, which is the 9. Therefore, the coverage mapping for the first function record is encoded in this string:
c"\01\00\00\01\01\01\0C\02\02"
This string consists of the following bytes:
0x01The number of file ids used by this function. There is only one file id used by the mapping data in this function.
0x00An index into the filenames array which corresponds to the file "/Users/alex/test.c".
0x00The number of counter expressions used by this function. This function doesn't use any expressions.
0x01The number of mapping regions that are stored in an array for the function's file id #0.
0x01The coverage mapping counter for the first region in this function. The value of 1 tells us that it's a coverage mapping counter that is a reference to the profile instrumentation counter with an index of 0.
0x01The starting line of the first mapping region in this function.
0x0CThe starting column of the first mapping region in this function.
0x02The ending line of the first mapping region in this function.
0x02The ending column of the first mapping region in this function.
The length of the substring that contains the encoded coverage mapping data for the second function record is also 9. It's structured like the mapping data for the first function record.
The two trailing bytes are zeroes and are used to pad the coverage mapping data to give it the 8 byte alignment.
The per-function coverage mapping data is encoded as a stream of bytes, with a simple structure. The structure consists of the encoding types like variable-length unsigned integers, that are used to encode File ID Mapping, Counter Expressions and the Mapping Regions.
The format of the structure follows:
[file id mapping, counter expressions, mapping regions]
The translation unit filenames are encoded using the same encoding types as the per-function coverage mapping data, with the following structure:
[numFilenames : LEB128, filename0 : string, filename1 : string, ...]
This section describes the basic types that are used by the encoding format
and can appear after : in the [foo : type] description.
LEB128 is an unsigned integer value that is encoded using DWARF's LEB128 encoding, optimizing for the case where values are small (1 byte for values less than 128).
[length : LEB128, characters...]
String values are encoded with a LEB value for the length of the string and a sequence of bytes for its characters.
[numIndices : LEB128, filenameIndex0 : LEB128, filenameIndex1 : LEB128, ...]
File id mapping in a function's coverage mapping stream contains the indices into the translation unit's filenames array.
[value : LEB128]
A coverage mapping counter is stored in a single LEB value. It is composed of two things --- the tag which is stored in the lowest 2 bits, and the counter data which is stored in the remaining bits.
The counter's tag encodes the counter's kind and, if the counter is an expression, the expression's kind. The possible tag values are:
- 0 - The counter is zero.
- 1 - The counter is a reference to the profile instrumentation counter.
- 2 - The counter is a subtraction expression.
- 3 - The counter is an addition expression.
The counter's data is interpreted in the following manner:
- When the counter is a reference to the profile instrumentation counter, then the counter's data is the id of the profile counter.
- When the counter is an expression, then the counter's data is the index into the array of counter expressions.
[numExpressions : LEB128, expr0LHS : LEB128, expr0RHS : LEB128, expr1LHS : LEB128, expr1RHS : LEB128, ...]
Counter expressions consist of two counters as they represent binary arithmetic operations. The expression's kind is determined from the tag of the counter that references this expression.
[numRegionArrays : LEB128, regionsForFile0, regionsForFile1, ...]
The mapping regions are stored in an array of sub-arrays where every region in a particular sub-array has the same file id.
The file id for a sub-array of regions is the index of that sub-array in the main array e.g. The first sub-array will have the file id of 0.
[numRegions : LEB128, region0, region1, ...]
The mapping regions for a specific file id are stored in an array that is sorted in an ascending order by the region's starting location.
[header, source range]
The mapping region record contains two sub-records --- the header, which stores the counter and/or the region's kind, and the source range that contains the starting and ending location of this region.
[counter]
or
[pseudo-counter]
The header encodes the region's counter and the region's kind.
The value of the counter's tag distinguishes between the counters and pseudo-counters --- if the tag is zero, than this header contains a pseudo-counter, otherwise this header contains an ordinary counter.
A mapping region whose header has a counter with a non-zero tag is a code region.
[value : LEB128]
A pseudo-counter is stored in a single LEB value, just like the ordinary counter. It has the following interpretation:
bits 0-1: tag, which is always 0.
bit 2: expansionRegionTag. If this bit is set, then this mapping region is an expansion region.
remaining bits: data. If this region is an expansion region, then the data contains the expanded file id of that region.
Otherwise, the data contains the region's kind. The possible region kind values are:
- 0 - This mapping region is a code region with a counter of zero.
- 2 - This mapping region is a skipped region.
[deltaLineStart : LEB128, columnStart : LEB128, numLines : LEB128, columnEnd : LEB128]
The source range record contains the following fields:
deltaLineStart: The difference between the starting line of the current mapping region and the starting line of the previous mapping region.
If the current mapping region is the first region in the current sub-array, then it stores the starting line of that region.
columnStart: The starting column of the mapping region.
numLines: The difference between the ending line and the starting line of the current mapping region.
columnEnd: The ending column of the mapping region.