1FileCheck - Flexible pattern matching file verifier 2=================================================== 3 4 5SYNOPSIS 6-------- 7 8 9**FileCheck** *match-filename* [*--check-prefix=XXX*] [*--strict-whitespace*] 10 11 12DESCRIPTION 13----------- 14 15 16**FileCheck** reads two files (one from standard input, and one specified on the 17command line) and uses one to verify the other. This behavior is particularly 18useful for the testsuite, which wants to verify that the output of some tool 19(e.g. llc) contains the expected information (for example, a movsd from esp or 20whatever is interesting). This is similar to using grep, but it is optimized 21for matching multiple different inputs in one file in a specific order. 22 23The *match-filename* file specifies the file that contains the patterns to 24match. The file to verify is always read from standard input. 25 26 27OPTIONS 28------- 29 30 31 32**-help** 33 34 Print a summary of command line options. 35 36 37 38**--check-prefix** *prefix* 39 40 FileCheck searches the contents of *match-filename* for patterns to match. By 41 default, these patterns are prefixed with "CHECK:". If you'd like to use a 42 different prefix (e.g. because the same input file is checking multiple 43 different tool or options), the **--check-prefix** argument allows you to specify 44 a specific prefix to match. 45 46 47 48**--input-file** *filename* 49 50 File to check (defaults to stdin). 51 52 53**--strict-whitespace** 54 55 By default, FileCheck canonicalizes input horizontal whitespace (spaces and 56 tabs) which causes it to ignore these differences (a space will match a tab). 57 The --strict-whitespace argument disables this behavior. 58 59 60 61**-version** 62 63 Show the version number of this program. 64 65 66 67 68EXIT STATUS 69----------- 70 71 72If **FileCheck** verifies that the file matches the expected contents, it exits 73with 0. Otherwise, if not, or if an error occurs, it will exit with a non-zero 74value. 75 76 77TUTORIAL 78-------- 79 80 81FileCheck is typically used from LLVM regression tests, being invoked on the RUN 82line of the test. A simple example of using FileCheck from a RUN line looks 83like this: 84 85 86.. code-block:: llvm 87 88 ; RUN: llvm-as < %s | llc -march=x86-64 | FileCheck %s 89 90 91This syntax says to pipe the current file ("%s") into llvm-as, pipe that into 92llc, then pipe the output of llc into FileCheck. This means that FileCheck will 93be verifying its standard input (the llc output) against the filename argument 94specified (the original .ll file specified by "%s"). To see how this works, 95let's look at the rest of the .ll file (after the RUN line): 96 97 98.. code-block:: llvm 99 100 define void @sub1(i32* %p, i32 %v) { 101 entry: 102 ; CHECK: sub1: 103 ; CHECK: subl 104 %0 = tail call i32 @llvm.atomic.load.sub.i32.p0i32(i32* %p, i32 %v) 105 ret void 106 } 107 108 define void @inc4(i64* %p) { 109 entry: 110 ; CHECK: inc4: 111 ; CHECK: incq 112 %0 = tail call i64 @llvm.atomic.load.add.i64.p0i64(i64* %p, i64 1) 113 ret void 114 } 115 116 117Here you can see some "CHECK:" lines specified in comments. Now you can see 118how the file is piped into llvm-as, then llc, and the machine code output is 119what we are verifying. FileCheck checks the machine code output to verify that 120it matches what the "CHECK:" lines specify. 121 122The syntax of the CHECK: lines is very simple: they are fixed strings that 123must occur in order. FileCheck defaults to ignoring horizontal whitespace 124differences (e.g. a space is allowed to match a tab) but otherwise, the contents 125of the CHECK: line is required to match some thing in the test file exactly. 126 127One nice thing about FileCheck (compared to grep) is that it allows merging 128test cases together into logical groups. For example, because the test above 129is checking for the "sub1:" and "inc4:" labels, it will not match unless there 130is a "subl" in between those labels. If it existed somewhere else in the file, 131that would not count: "grep subl" matches if subl exists anywhere in the 132file. 133 134The FileCheck -check-prefix option 135~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 136 137 138The FileCheck -check-prefix option allows multiple test configurations to be 139driven from one .ll file. This is useful in many circumstances, for example, 140testing different architectural variants with llc. Here's a simple example: 141 142 143.. code-block:: llvm 144 145 ; RUN: llvm-as < %s | llc -mtriple=i686-apple-darwin9 -mattr=sse41 \ 146 ; RUN: | FileCheck %s -check-prefix=X32 147 ; RUN: llvm-as < %s | llc -mtriple=x86_64-apple-darwin9 -mattr=sse41 \ 148 ; RUN: | FileCheck %s -check-prefix=X64 149 150 define <4 x i32> @pinsrd_1(i32 %s, <4 x i32> %tmp) nounwind { 151 %tmp1 = insertelement <4 x i32>; %tmp, i32 %s, i32 1 152 ret <4 x i32> %tmp1 153 ; X32: pinsrd_1: 154 ; X32: pinsrd $1, 4(%esp), %xmm0 155 156 ; X64: pinsrd_1: 157 ; X64: pinsrd $1, %edi, %xmm0 158 } 159 160 161In this case, we're testing that we get the expected code generation with 162both 32-bit and 64-bit code generation. 163 164 165The "CHECK-NEXT:" directive 166~~~~~~~~~~~~~~~~~~~~~~~~~~~ 167 168 169Sometimes you want to match lines and would like to verify that matches 170happen on exactly consecutive lines with no other lines in between them. In 171this case, you can use CHECK: and CHECK-NEXT: directives to specify this. If 172you specified a custom check prefix, just use "<PREFIX>-NEXT:". For 173example, something like this works as you'd expect: 174 175 176.. code-block:: llvm 177 178 define void @t2(<2 x double>* %r, <2 x double>* %A, double %B) { 179 %tmp3 = load <2 x double>* %A, align 16 180 %tmp7 = insertelement <2 x double> undef, double %B, i32 0 181 %tmp9 = shufflevector <2 x double> %tmp3, 182 <2 x double> %tmp7, 183 <2 x i32> < i32 0, i32 2 > 184 store <2 x double> %tmp9, <2 x double>* %r, align 16 185 ret void 186 187 ; CHECK: t2: 188 ; CHECK: movl 8(%esp), %eax 189 ; CHECK-NEXT: movapd (%eax), %xmm0 190 ; CHECK-NEXT: movhpd 12(%esp), %xmm0 191 ; CHECK-NEXT: movl 4(%esp), %eax 192 ; CHECK-NEXT: movapd %xmm0, (%eax) 193 ; CHECK-NEXT: ret 194 } 195 196 197CHECK-NEXT: directives reject the input unless there is exactly one newline 198between it an the previous directive. A CHECK-NEXT cannot be the first 199directive in a file. 200 201 202The "CHECK-NOT:" directive 203~~~~~~~~~~~~~~~~~~~~~~~~~~ 204 205 206The CHECK-NOT: directive is used to verify that a string doesn't occur 207between two matches (or before the first match, or after the last match). For 208example, to verify that a load is removed by a transformation, a test like this 209can be used: 210 211 212.. code-block:: llvm 213 214 define i8 @coerce_offset0(i32 %V, i32* %P) { 215 store i32 %V, i32* %P 216 217 %P2 = bitcast i32* %P to i8* 218 %P3 = getelementptr i8* %P2, i32 2 219 220 %A = load i8* %P3 221 ret i8 %A 222 ; CHECK: @coerce_offset0 223 ; CHECK-NOT: load 224 ; CHECK: ret i8 225 } 226 227 228 229FileCheck Pattern Matching Syntax 230~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 231 232 233The CHECK: and CHECK-NOT: directives both take a pattern to match. For most 234uses of FileCheck, fixed string matching is perfectly sufficient. For some 235things, a more flexible form of matching is desired. To support this, FileCheck 236allows you to specify regular expressions in matching strings, surrounded by 237double braces: **{{yourregex}}**. Because we want to use fixed string 238matching for a majority of what we do, FileCheck has been designed to support 239mixing and matching fixed string matching with regular expressions. This allows 240you to write things like this: 241 242 243.. code-block:: llvm 244 245 ; CHECK: movhpd {{[0-9]+}}(%esp), {{%xmm[0-7]}} 246 247 248In this case, any offset from the ESP register will be allowed, and any xmm 249register will be allowed. 250 251Because regular expressions are enclosed with double braces, they are 252visually distinct, and you don't need to use escape characters within the double 253braces like you would in C. In the rare case that you want to match double 254braces explicitly from the input, you can use something ugly like 255**{{[{][{]}}** as your pattern. 256 257 258FileCheck Variables 259~~~~~~~~~~~~~~~~~~~ 260 261 262It is often useful to match a pattern and then verify that it occurs again 263later in the file. For codegen tests, this can be useful to allow any register, 264but verify that that register is used consistently later. To do this, FileCheck 265allows named variables to be defined and substituted into patterns. Here is a 266simple example: 267 268 269.. code-block:: llvm 270 271 ; CHECK: test5: 272 ; CHECK: notw [[REGISTER:%[a-z]+]] 273 ; CHECK: andw {{.*}}[[REGISTER]] 274 275 276The first check line matches a regex (**%[a-z]+**) and captures it into 277the variable "REGISTER". The second line verifies that whatever is in REGISTER 278occurs later in the file after an "andw". FileCheck variable references are 279always contained in **[[ ]]** pairs, and their names can be formed with the 280regex **[a-zA-Z][a-zA-Z0-9]***. If a colon follows the name, then it is a 281definition of the variable; otherwise, it is a use. 282 283FileCheck variables can be defined multiple times, and uses always get the 284latest value. Note that variables are all read at the start of a "CHECK" line 285and are all defined at the end. This means that if you have something like 286"**CHECK: [[XYZ:.\\*]]x[[XYZ]]**", the check line will read the previous 287value of the XYZ variable and define a new one after the match is performed. If 288you need to do something like this you can probably take advantage of the fact 289that FileCheck is not actually line-oriented when it matches, this allows you to 290define two separate CHECK lines that match on the same line. 291