maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/DebugInfo/COFF/inlining.ll

300 lines
11 KiB
LLVM
Raw Normal View History

[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
; RUN: llc -mcpu=core2 -mtriple=i686-pc-win32 < %s | FileCheck %s --check-prefix=ASM
; RUN: llc -mcpu=core2 -mtriple=i686-pc-win32 < %s -filetype=obj | llvm-readobj -codeview | FileCheck %s --check-prefix=OBJ
; This LL file was generated by running 'clang -O1 -g -gcodeview' on the
; following code:
; 1: extern volatile int x;
; 2: static inline void foo() {
; 3: int y = 1;
; 4: x += (int)&y;
; 5: x += 2;
; 6: x += 3;
; 7: }
; 8: static inline void bar() {
; 9: x += 4;
; 10: foo();
; 11: x += 5;
; 12: }
; 13: void baz() {
; 14: x += 6;
; 15: bar();
; 16: x += 7;
; 17: }
[codeview] Add new directives to record inlined call site line info Summary: Previously we were trying to represent this with the "contains" list of the .cv_inline_linetable directive, which was not enough information. Now we directly represent the chain of inlined call sites, so we know what location to emit when we encounter a .cv_loc directive of an inner inlined call site while emitting the line table of an outer function or inlined call site. Fixes PR29146. Also fixes PR29147, where we would crash when .cv_loc directives crossed sections. Now we write down the section of the first .cv_loc directive, and emit an error if any other .cv_loc directive for that function is in a different section. Also fixes issues with discontiguous inlined source locations, like in this example: volatile int unlikely_cond = 0; extern void __declspec(noreturn) abort(); __forceinline void f() { if (!unlikely_cond) abort(); } int main() { unlikely_cond = 0; f(); unlikely_cond = 0; } Previously our tables gave bad location information for the 'abort' call, and the debugger wouldn't snow the inlined stack frame for 'f'. It is important to emit good line tables for this code pattern, because it comes up whenever an asan bug occurs in an inlined function. The __asan_report* stubs are generally placed after the normal function epilogue, leading to discontiguous regions of inlined code. Reviewers: majnemer, amccarth Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D24014 llvm-svn: 280822
2016-09-08 00:15:31 +08:00
; ASM: .cv_func_id 0
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
; ASM: .cv_loc 0 1 13 0 is_stmt 0 # t.cpp:13:0
; ASM: .cv_loc 0 1 14 5 # t.cpp:14:5
; ASM: addl $6, "?x@@3HC"
[codeview] Add new directives to record inlined call site line info Summary: Previously we were trying to represent this with the "contains" list of the .cv_inline_linetable directive, which was not enough information. Now we directly represent the chain of inlined call sites, so we know what location to emit when we encounter a .cv_loc directive of an inner inlined call site while emitting the line table of an outer function or inlined call site. Fixes PR29146. Also fixes PR29147, where we would crash when .cv_loc directives crossed sections. Now we write down the section of the first .cv_loc directive, and emit an error if any other .cv_loc directive for that function is in a different section. Also fixes issues with discontiguous inlined source locations, like in this example: volatile int unlikely_cond = 0; extern void __declspec(noreturn) abort(); __forceinline void f() { if (!unlikely_cond) abort(); } int main() { unlikely_cond = 0; f(); unlikely_cond = 0; } Previously our tables gave bad location information for the 'abort' call, and the debugger wouldn't snow the inlined stack frame for 'f'. It is important to emit good line tables for this code pattern, because it comes up whenever an asan bug occurs in an inlined function. The __asan_report* stubs are generally placed after the normal function epilogue, leading to discontiguous regions of inlined code. Reviewers: majnemer, amccarth Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D24014 llvm-svn: 280822
2016-09-08 00:15:31 +08:00
; ASM: .cv_inline_site_id 1 within 0 inlined_at 1 15 3
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
; ASM: .cv_loc 1 1 9 5 # t.cpp:9:5
; ASM: addl $4, "?x@@3HC"
[codeview] Add new directives to record inlined call site line info Summary: Previously we were trying to represent this with the "contains" list of the .cv_inline_linetable directive, which was not enough information. Now we directly represent the chain of inlined call sites, so we know what location to emit when we encounter a .cv_loc directive of an inner inlined call site while emitting the line table of an outer function or inlined call site. Fixes PR29146. Also fixes PR29147, where we would crash when .cv_loc directives crossed sections. Now we write down the section of the first .cv_loc directive, and emit an error if any other .cv_loc directive for that function is in a different section. Also fixes issues with discontiguous inlined source locations, like in this example: volatile int unlikely_cond = 0; extern void __declspec(noreturn) abort(); __forceinline void f() { if (!unlikely_cond) abort(); } int main() { unlikely_cond = 0; f(); unlikely_cond = 0; } Previously our tables gave bad location information for the 'abort' call, and the debugger wouldn't snow the inlined stack frame for 'f'. It is important to emit good line tables for this code pattern, because it comes up whenever an asan bug occurs in an inlined function. The __asan_report* stubs are generally placed after the normal function epilogue, leading to discontiguous regions of inlined code. Reviewers: majnemer, amccarth Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D24014 llvm-svn: 280822
2016-09-08 00:15:31 +08:00
; ASM: .cv_inline_site_id 2 within 1 inlined_at 1 10 3
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
; ASM: .cv_loc 2 1 3 7 # t.cpp:3:7
; ASM: .cv_loc 2 1 4 5 # t.cpp:4:5
; ASM: addl {{.*}}, "?x@@3HC"
; ASM: .cv_loc 2 1 5 5 # t.cpp:5:5
; ASM: addl $2, "?x@@3HC"
; ASM: .cv_loc 2 1 6 5 # t.cpp:6:5
; ASM: addl $3, "?x@@3HC"
; ASM: .cv_loc 1 1 11 5 # t.cpp:11:5
; ASM: addl $5, "?x@@3HC"
; ASM: .cv_loc 0 1 16 5 # t.cpp:16:5
; ASM: addl $7, "?x@@3HC"
; ASM: .cv_loc 0 1 17 1 # t.cpp:17:1
[codeview] Translate basic DITypes to CV type records Summary: This is meant to be the tiniest step towards DIType to CV type index translation that I could come up with. Whenever translation fails, we use type index zero, which is the unknown type. Reviewers: aaboud, zturner Subscribers: llvm-commits, amccarth Differential Revision: http://reviews.llvm.org/D20840 llvm-svn: 271408
2016-06-02 01:05:51 +08:00
; ASM: .section .debug$S,"dr"
; ASM: .long 246 # Inlinee lines subsection
; ASM: .long [[inline_end:.*]]-[[inline_beg:.*]] #
; ASM: [[inline_beg]]:
; ASM: .long 0
; ASM: # Inlined function bar starts at t.cpp:8
; ASM: .long 4098 # Type index of inlined function
Re-land "Fix Bug 30978 by emitting cv file checksums." This reverts r313431 and brings back r313374 with a fix to write checksums as binary data and not ASCII hex strings. llvm-svn: 313657
2017-09-20 02:14:45 +08:00
; ASM: .cv_filechecksumoffset 1 # Offset into filechecksum table
[codeview] Translate basic DITypes to CV type records Summary: This is meant to be the tiniest step towards DIType to CV type index translation that I could come up with. Whenever translation fails, we use type index zero, which is the unknown type. Reviewers: aaboud, zturner Subscribers: llvm-commits, amccarth Differential Revision: http://reviews.llvm.org/D20840 llvm-svn: 271408
2016-06-02 01:05:51 +08:00
; ASM: .long 8 # Starting line number
; ASM: # Inlined function foo starts at t.cpp:2
; ASM: .long 4099
Re-land "Fix Bug 30978 by emitting cv file checksums." This reverts r313431 and brings back r313374 with a fix to write checksums as binary data and not ASCII hex strings. llvm-svn: 313657
2017-09-20 02:14:45 +08:00
; ASM: .cv_filechecksumoffset 1 # Offset into filechecksum table
[codeview] Translate basic DITypes to CV type records Summary: This is meant to be the tiniest step towards DIType to CV type index translation that I could come up with. Whenever translation fails, we use type index zero, which is the unknown type. Reviewers: aaboud, zturner Subscribers: llvm-commits, amccarth Differential Revision: http://reviews.llvm.org/D20840 llvm-svn: 271408
2016-06-02 01:05:51 +08:00
; ASM: .long 2
; ASM: [[inline_end]]:
; ASM: .long 241 # Symbol subsection for baz
; ASM: .long {{.*}} # Subsection size
; ASM: .short 4429
; ASM: .long
; ASM: .long
; ASM: .long
[codeview] Add new directives to record inlined call site line info Summary: Previously we were trying to represent this with the "contains" list of the .cv_inline_linetable directive, which was not enough information. Now we directly represent the chain of inlined call sites, so we know what location to emit when we encounter a .cv_loc directive of an inner inlined call site while emitting the line table of an outer function or inlined call site. Fixes PR29146. Also fixes PR29147, where we would crash when .cv_loc directives crossed sections. Now we write down the section of the first .cv_loc directive, and emit an error if any other .cv_loc directive for that function is in a different section. Also fixes issues with discontiguous inlined source locations, like in this example: volatile int unlikely_cond = 0; extern void __declspec(noreturn) abort(); __forceinline void f() { if (!unlikely_cond) abort(); } int main() { unlikely_cond = 0; f(); unlikely_cond = 0; } Previously our tables gave bad location information for the 'abort' call, and the debugger wouldn't snow the inlined stack frame for 'f'. It is important to emit good line tables for this code pattern, because it comes up whenever an asan bug occurs in an inlined function. The __asan_report* stubs are generally placed after the normal function epilogue, leading to discontiguous regions of inlined code. Reviewers: majnemer, amccarth Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D24014 llvm-svn: 280822
2016-09-08 00:15:31 +08:00
; ASM: .cv_inline_linetable 1 1 8 Lfunc_begin0 Lfunc_end0
[codeview] Translate basic DITypes to CV type records Summary: This is meant to be the tiniest step towards DIType to CV type index translation that I could come up with. Whenever translation fails, we use type index zero, which is the unknown type. Reviewers: aaboud, zturner Subscribers: llvm-commits, amccarth Differential Revision: http://reviews.llvm.org/D20840 llvm-svn: 271408
2016-06-02 01:05:51 +08:00
; ASM: .short 4429
; ASM: .long
; ASM: .long
; ASM: .long
; ASM: .cv_inline_linetable 2 1 2 Lfunc_begin0 Lfunc_end0
; ASM: .short 4430
; ASM: .short 4430
Re-land "Fix Bug 30978 by emitting cv file checksums." This reverts r313431 and brings back r313374 with a fix to write checksums as binary data and not ASCII hex strings. llvm-svn: 313657
2017-09-20 02:14:45 +08:00
; ASM: .cv_filechecksums
[codeview] Improve readability of type record assembly Adds the method MCStreamer::EmitBinaryData, which is usually an alias for EmitBytes. In the MCAsmStreamer case, it is overridden to emit hex dump output like this: .byte 0x0e, 0x00, 0x08, 0x10 .byte 0x03, 0x00, 0x00, 0x00 .byte 0x00, 0x00, 0x00, 0x00 .byte 0x00, 0x10, 0x00, 0x00 Also, when verbose asm comments are enabled, this patch prints the dump output for each comment before its record, like this: # ArgList (0x1000) { # TypeLeafKind: LF_ARGLIST (0x1201) # NumArgs: 0 # Arguments [ # ] # } .byte 0x06, 0x00, 0x01, 0x12 .byte 0x00, 0x00, 0x00, 0x00 This should make debugging easier and testing more convenient. Reviewers: aaboud Subscribers: majnemer, zturner, amccarth, aaboud, llvm-commits Differential Revision: http://reviews.llvm.org/D20711 llvm-svn: 271313
2016-06-01 02:45:36 +08:00
; ASM: .section .debug$T,"dr"
; ASM: .long 4 # Debug section magic
; ASM: # ArgList (0x1000) {
; ASM: # TypeLeafKind: LF_ARGLIST (0x1201)
; ASM: # NumArgs: 0
; ASM: # Arguments [
; ASM: # ]
; ASM: # }
; ASM: .byte 0x06, 0x00, 0x01, 0x12
; ASM: .byte 0x00, 0x00, 0x00, 0x00
; ASM: # Procedure (0x1001) {
; ASM: # TypeLeafKind: LF_PROCEDURE (0x1008)
; ASM: # ReturnType: void (0x3)
; ASM: # CallingConvention: NearC (0x0)
; ASM: # FunctionOptions [ (0x0)
; ASM: # ]
; ASM: # NumParameters: 0
; ASM: # ArgListType: () (0x1000)
; ASM: # }
; ASM: .byte 0x0e, 0x00, 0x08, 0x10
; ASM: .byte 0x03, 0x00, 0x00, 0x00
; ASM: .byte 0x00, 0x00, 0x00, 0x00
; ASM: .byte 0x00, 0x10, 0x00, 0x00
; ASM: # FuncId (0x1002) {
; ASM: # TypeLeafKind: LF_FUNC_ID (0x1601)
; ASM: # ParentScope: 0x0
; ASM: # FunctionType: void () (0x1001)
; ASM: # Name: bar
; ASM: # }
; ASM: .byte 0x0e, 0x00, 0x01, 0x16
; ASM: .byte 0x00, 0x00, 0x00, 0x00
; ASM: .byte 0x01, 0x10, 0x00, 0x00
; ASM: .byte 0x62, 0x61, 0x72, 0x00
; ASM: # FuncId (0x1003) {
; ASM: # TypeLeafKind: LF_FUNC_ID (0x1601)
; ASM: # ParentScope: 0x0
; ASM: # FunctionType: void () (0x1001)
; ASM: # Name: foo
; ASM: # }
; ASM: .byte 0x0e, 0x00, 0x01, 0x16
; ASM: .byte 0x00, 0x00, 0x00, 0x00
; ASM: .byte 0x01, 0x10, 0x00, 0x00
; ASM: .byte 0x66, 0x6f, 0x6f, 0x00
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; We should only the LF_FUNC_ID records that we needed to reference.
; OBJ: CodeViewTypes [
[codeview] Translate basic DITypes to CV type records Summary: This is meant to be the tiniest step towards DIType to CV type index translation that I could come up with. Whenever translation fails, we use type index zero, which is the unknown type. Reviewers: aaboud, zturner Subscribers: llvm-commits, amccarth Differential Revision: http://reviews.llvm.org/D20840 llvm-svn: 271408
2016-06-02 01:05:51 +08:00
; OBJ: Section: .debug$T
[codeview] Dump the type index on the first line of each record This will make it easier to write FileCheck tests. llvm-svn: 269444
2016-05-14 01:48:24 +08:00
; OBJ: ArgList (0x1000) {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: TypeLeafKind: LF_ARGLIST (0x1201)
; OBJ: NumArgs: 0
; OBJ: }
[codeview] Align class and print names of types Summary: This way we can get rid of one of the fields in the .def file. Reviewers: llvm-commits Subscribers: zturner Differential Revision: http://reviews.llvm.org/D20251 llvm-svn: 269461
2016-05-14 03:37:07 +08:00
; OBJ: Procedure (0x1001) {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: TypeLeafKind: LF_PROCEDURE (0x1008)
; OBJ: ReturnType: void (0x3)
; OBJ: NumParameters: 0
; OBJ: ArgListType: () (0x1000)
; OBJ: }
[codeview] Dump the type index on the first line of each record This will make it easier to write FileCheck tests. llvm-svn: 269444
2016-05-14 01:48:24 +08:00
; OBJ: FuncId (0x1002) {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: TypeLeafKind: LF_FUNC_ID (0x1601)
; OBJ: ParentScope: 0x0
; OBJ: FunctionType: void () (0x1001)
; OBJ: Name: bar
; OBJ: }
[codeview] Dump the type index on the first line of each record This will make it easier to write FileCheck tests. llvm-svn: 269444
2016-05-14 01:48:24 +08:00
; OBJ: FuncId (0x1003) {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: TypeLeafKind: LF_FUNC_ID (0x1601)
; OBJ: ParentScope: 0x0
; OBJ: FunctionType: void () (0x1001)
; OBJ: Name: foo
; OBJ: }
[CodeView] Implement function-type indices We still need to do something about member functions and calling conventions. Differential Revision: http://reviews.llvm.org/D20900 llvm-svn: 271541
2016-06-03 01:13:53 +08:00
; OBJ: FuncId (0x1004) {
; OBJ: TypeLeafKind: LF_FUNC_ID (0x1601)
; OBJ: ParentScope: 0x0
; OBJ: FunctionType: void () (0x1001)
; OBJ: Name: baz
; OBJ: }
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ-NOT: TypeLeafKind: LF_FUNC_ID
; OBJ: ]
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: Subsection [
; OBJ: SubSectionType: InlineeLines (0xF6)
; OBJ: SubSectionSize: 0x1C
; OBJ: InlineeSourceLine {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: Inlinee: bar (0x1002)
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: FileID: D:\src\llvm\build\t.cpp (0x0)
; OBJ: SourceLineNum: 8
; OBJ: }
; OBJ: InlineeSourceLine {
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: Inlinee: foo (0x1003)
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: FileID: D:\src\llvm\build\t.cpp (0x0)
; OBJ: SourceLineNum: 2
; OBJ: }
; OBJ: ]
; OBJ: Subsection [
; OBJ: SubSectionType: Symbols (0xF1)
[codeview] Change readobj symbol dumping format Avoid duplicating DictScope with hand-written names everywhere. Print the S_-prefixed symbol kind for every record. This should make it easier to search for certain kinds of records when debugging PDB linking. llvm-svn: 307732
2017-07-12 07:41:41 +08:00
; OBJ: {{.*}}Proc{{.*}}Sym {
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: PtrParent: 0x0
; OBJ: PtrEnd: 0x0
; OBJ: PtrNext: 0x0
[X86] Optimization for replacing LEA with MOV at frame index elimination time Summary: Replace a LEA instruction of the form 'lea (%esp), %ebx' --> 'mov %esp, %ebx' MOV is preferable over LEA because usually there are more issue-slots available to execute MOVs than LEAs. Latest processors also support zero-latency MOVs. Fixes pr29022. Reviewers: hfinkel, delena, igorb, myatsina, mkuper Differential Revision: https://reviews.llvm.org/D24705 llvm-svn: 282385
2016-09-26 14:42:07 +08:00
; OBJ: CodeSize: 0x3C
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: DbgStart: 0x0
; OBJ: DbgEnd: 0x0
[CodeView] Implement function-type indices We still need to do something about member functions and calling conventions. Differential Revision: http://reviews.llvm.org/D20900 llvm-svn: 271541
2016-06-03 01:13:53 +08:00
; OBJ: FunctionType: baz (0x1004)
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: CodeOffset: ?baz@@YAXXZ+0x0
; OBJ: Segment: 0x0
; OBJ: Flags [ (0x0)
; OBJ: ]
; OBJ: DisplayName: baz
; OBJ: LinkageName: ?baz@@YAXXZ
; OBJ: }
[codeview] Change readobj symbol dumping format Avoid duplicating DictScope with hand-written names everywhere. Print the S_-prefixed symbol kind for every record. This should make it easier to search for certain kinds of records when debugging PDB linking. llvm-svn: 307732
2017-07-12 07:41:41 +08:00
; OBJ: InlineSiteSym {
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: PtrParent: 0x0
; OBJ: PtrEnd: 0x0
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: Inlinee: bar (0x1002)
[codeview] Shrink inlined call site line info tables For a fully inlined call chain like a -> b -> c -> d, we were emitting line info for 'd' 3 separate times: once for d's actual InlineSite line table, and twice for 'b' and 'c'. This is particularly inefficient when all these functions are in different headers, because now we need to encode the file change. Windbg was coping with our suboptimal output, so this should not be noticeable from the debugger. llvm-svn: 275502
2016-07-15 07:47:15 +08:00
; OBJ: BinaryAnnotations [
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0x8, LineOffset: 1}
[codeview] Add new directives to record inlined call site line info Summary: Previously we were trying to represent this with the "contains" list of the .cv_inline_linetable directive, which was not enough information. Now we directly represent the chain of inlined call sites, so we know what location to emit when we encounter a .cv_loc directive of an inner inlined call site while emitting the line table of an outer function or inlined call site. Fixes PR29146. Also fixes PR29147, where we would crash when .cv_loc directives crossed sections. Now we write down the section of the first .cv_loc directive, and emit an error if any other .cv_loc directive for that function is in a different section. Also fixes issues with discontiguous inlined source locations, like in this example: volatile int unlikely_cond = 0; extern void __declspec(noreturn) abort(); __forceinline void f() { if (!unlikely_cond) abort(); } int main() { unlikely_cond = 0; f(); unlikely_cond = 0; } Previously our tables gave bad location information for the 'abort' call, and the debugger wouldn't snow the inlined stack frame for 'f'. It is important to emit good line tables for this code pattern, because it comes up whenever an asan bug occurs in an inlined function. The __asan_report* stubs are generally placed after the normal function epilogue, leading to discontiguous regions of inlined code. Reviewers: majnemer, amccarth Subscribers: llvm-commits Differential Revision: https://reviews.llvm.org/D24014 llvm-svn: 280822
2016-09-08 00:15:31 +08:00
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0x7, LineOffset: 1}
; OBJ-NEXT: ChangeLineOffset: 1
[X86] Optimization for replacing LEA with MOV at frame index elimination time Summary: Replace a LEA instruction of the form 'lea (%esp), %ebx' --> 'mov %esp, %ebx' MOV is preferable over LEA because usually there are more issue-slots available to execute MOVs than LEAs. Latest processors also support zero-latency MOVs. Fixes pr29022. Reviewers: hfinkel, delena, igorb, myatsina, mkuper Differential Revision: https://reviews.llvm.org/D24705 llvm-svn: 282385
2016-09-26 14:42:07 +08:00
; OBJ-NEXT: ChangeCodeOffset: 0x1D
[codeview] Shrink inlined call site line info tables For a fully inlined call chain like a -> b -> c -> d, we were emitting line info for 'd' 3 separate times: once for d's actual InlineSite line table, and twice for 'b' and 'c'. This is particularly inefficient when all these functions are in different headers, because now we need to encode the file change. Windbg was coping with our suboptimal output, so this should not be noticeable from the debugger. llvm-svn: 275502
2016-07-15 07:47:15 +08:00
; OBJ-NEXT: ChangeCodeLength: 0x7
; OBJ: ]
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: }
[codeview] Change readobj symbol dumping format Avoid duplicating DictScope with hand-written names everywhere. Print the S_-prefixed symbol kind for every record. This should make it easier to search for certain kinds of records when debugging PDB linking. llvm-svn: 307732
2017-07-12 07:41:41 +08:00
; OBJ: InlineSiteSym {
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: PtrParent: 0x0
; OBJ: PtrEnd: 0x0
[codeview] Only emit function ids for inlined functions We aren't referencing any other kind of function currently. Should save a bit on our debug info size. llvm-svn: 263817
2016-03-19 02:54:32 +08:00
; OBJ: Inlinee: foo (0x1003)
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ: BinaryAnnotations [
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0xF, LineOffset: 1}
[X86] Optimization for replacing LEA with MOV at frame index elimination time Summary: Replace a LEA instruction of the form 'lea (%esp), %ebx' --> 'mov %esp, %ebx' MOV is preferable over LEA because usually there are more issue-slots available to execute MOVs than LEAs. Latest processors also support zero-latency MOVs. Fixes pr29022. Reviewers: hfinkel, delena, igorb, myatsina, mkuper Differential Revision: https://reviews.llvm.org/D24705 llvm-svn: 282385
2016-09-26 14:42:07 +08:00
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0x9, LineOffset: 1}
[codeview] Wire up the .cv_inline_linetable directive This directive emits the binary annotations that describe line and code deltas in inlined call sites. Single-stepping through inlined frames in windbg now works. llvm-svn: 259535
2016-02-03 01:41:18 +08:00
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0x6, LineOffset: 1}
; OBJ-NEXT: ChangeCodeOffsetAndLineOffset: {CodeOffset: 0x7, LineOffset: 1}
; OBJ-NEXT: ChangeCodeLength: 0x7
; OBJ: ]
; OBJ: }
; OBJ: InlineSiteEnd {
; OBJ: }
; OBJ: InlineSiteEnd {
; OBJ: }
; OBJ: ProcEnd
; OBJ: ]
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
; ModuleID = 't.cpp'
target datalayout = "e-m:w-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-pc-windows-msvc18.0.0"
@"\01?x@@3HC" = external global i32, align 4
; Function Attrs: norecurse nounwind uwtable
define void @"\01?baz@@YAXXZ"() #0 !dbg !4 {
entry:
%y.i.i = alloca i32, align 4
%0 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !12, !tbaa !13
%add = add nsw i32 %0, 6, !dbg !12
store volatile i32 %add, i32* @"\01?x@@3HC", align 4, !dbg !12, !tbaa !13
%1 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !17, !tbaa !13
%add.i = add nsw i32 %1, 4, !dbg !17
store volatile i32 %add.i, i32* @"\01?x@@3HC", align 4, !dbg !17, !tbaa !13
%2 = bitcast i32* %y.i.i to i8*, !dbg !19
call void @llvm.lifetime.start(i64 4, i8* %2) #2, !dbg !19
store i32 1, i32* %y.i.i, align 4, !dbg !21, !tbaa !13
%3 = ptrtoint i32* %y.i.i to i64, !dbg !22
%4 = trunc i64 %3 to i32, !dbg !22
%5 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !23, !tbaa !13
%add.i.i = add nsw i32 %5, %4, !dbg !23
store volatile i32 %add.i.i, i32* @"\01?x@@3HC", align 4, !dbg !23, !tbaa !13
%6 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !24, !tbaa !13
%add1.i.i = add nsw i32 %6, 2, !dbg !24
store volatile i32 %add1.i.i, i32* @"\01?x@@3HC", align 4, !dbg !24, !tbaa !13
%7 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !25, !tbaa !13
%add2.i.i = add nsw i32 %7, 3, !dbg !25
store volatile i32 %add2.i.i, i32* @"\01?x@@3HC", align 4, !dbg !25, !tbaa !13
call void @llvm.lifetime.end(i64 4, i8* %2) #2, !dbg !26
%8 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !27, !tbaa !13
%add1.i = add nsw i32 %8, 5, !dbg !27
store volatile i32 %add1.i, i32* @"\01?x@@3HC", align 4, !dbg !27, !tbaa !13
%9 = load volatile i32, i32* @"\01?x@@3HC", align 4, !dbg !28, !tbaa !13
%add1 = add nsw i32 %9, 7, !dbg !28
store volatile i32 %add1, i32* @"\01?x@@3HC", align 4, !dbg !28, !tbaa !13
ret void, !dbg !29
}
; Function Attrs: argmemonly nounwind
declare void @llvm.lifetime.start(i64, i8* nocapture) #1
; Function Attrs: argmemonly nounwind
declare void @llvm.lifetime.end(i64, i8* nocapture) #1
attributes #0 = { norecurse nounwind uwtable "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="x86-64" "target-features"="+fxsr,+mmx,+sse,+sse2" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind }
attributes #2 = { nounwind }
!llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!8, !9, !10}
!llvm.ident = !{!11}
[PR27284] Reverse the ownership between DICompileUnit and DISubprogram. Currently each Function points to a DISubprogram and DISubprogram has a scope field. For member functions the scope is a DICompositeType. DIScopes point to the DICompileUnit to facilitate type uniquing. Distinct DISubprograms (with isDefinition: true) are not part of the type hierarchy and cannot be uniqued. This change removes the subprograms list from DICompileUnit and instead adds a pointer to the owning compile unit to distinct DISubprograms. This would make it easy for ThinLTO to strip unneeded DISubprograms and their transitively referenced debug info. Motivation ---------- Materializing DISubprograms is currently the most expensive operation when doing a ThinLTO build of clang. We want the DISubprogram to be stored in a separate Bitcode block (or the same block as the function body) so we can avoid having to expensively deserialize all DISubprograms together with the global metadata. If a function has been inlined into another subprogram we need to store a reference the block containing the inlined subprogram. Attached to https://llvm.org/bugs/show_bug.cgi?id=27284 is a python script that updates LLVM IR testcases to the new format. http://reviews.llvm.org/D19034 <rdar://problem/25256815> llvm-svn: 266446
2016-04-15 23:57:41 +08:00
!0 = distinct !DICompileUnit(language: DW_LANG_C_plus_plus, file: !1, producer: "clang version 3.9.0 ", isOptimized: true, runtimeVersion: 0, emissionKind: LineTablesOnly, enums: !2)
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
!1 = !DIFile(filename: "t.cpp", directory: "D:\5Csrc\5Cllvm\5Cbuild")
!2 = !{}
[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label. In order to set breakpoints on labels and list source code around labels, we need collect debug information for labels, i.e., label name, the function label belong, line number in the file, and the address label located. In order to keep these information in LLVM IR and to allow backend to generate debug information correctly. We create a new kind of metadata for labels, DILabel. The format of DILabel is !DILabel(scope: !1, name: "foo", file: !2, line: 3) We hope to keep debug information as much as possible even the code is optimized. So, we create a new kind of intrinsic for label metadata to avoid the metadata is eliminated with basic block. The intrinsic will keep existing if we keep it from optimized out. The format of the intrinsic is llvm.dbg.label(metadata !1) It has only one argument, that is the DILabel metadata. The intrinsic will follow the label immediately. Backend could get the label metadata through the intrinsic's parameter. We also create DIBuilder API for labels to be used by Frontend. Frontend could use createLabel() to allocate DILabel objects, and use insertLabel() to insert llvm.dbg.label intrinsic in LLVM IR. Differential Revision: https://reviews.llvm.org/D45024 Patch by Hsiangkai Wang. llvm-svn: 331841
2018-05-09 10:40:45 +08:00
!4 = distinct !DISubprogram(name: "baz", scope: !1, file: !1, line: 13, type: !5, isLocal: false, isDefinition: true, scopeLine: 13, flags: DIFlagPrototyped, isOptimized: true, unit: !0, retainedNodes: !2)
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
!5 = !DISubroutineType(types: !2)
[DebugInfo] Add DILabel metadata and intrinsic llvm.dbg.label. In order to set breakpoints on labels and list source code around labels, we need collect debug information for labels, i.e., label name, the function label belong, line number in the file, and the address label located. In order to keep these information in LLVM IR and to allow backend to generate debug information correctly. We create a new kind of metadata for labels, DILabel. The format of DILabel is !DILabel(scope: !1, name: "foo", file: !2, line: 3) We hope to keep debug information as much as possible even the code is optimized. So, we create a new kind of intrinsic for label metadata to avoid the metadata is eliminated with basic block. The intrinsic will keep existing if we keep it from optimized out. The format of the intrinsic is llvm.dbg.label(metadata !1) It has only one argument, that is the DILabel metadata. The intrinsic will follow the label immediately. Backend could get the label metadata through the intrinsic's parameter. We also create DIBuilder API for labels to be used by Frontend. Frontend could use createLabel() to allocate DILabel objects, and use insertLabel() to insert llvm.dbg.label intrinsic in LLVM IR. Differential Revision: https://reviews.llvm.org/D45024 Patch by Hsiangkai Wang. llvm-svn: 331841
2018-05-09 10:40:45 +08:00
!6 = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 8, type: !5, isLocal: true, isDefinition: true, scopeLine: 8, flags: DIFlagPrototyped, isOptimized: true, unit: !0, retainedNodes: !2)
!7 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 2, type: !5, isLocal: true, isDefinition: true, scopeLine: 2, flags: DIFlagPrototyped, isOptimized: true, unit: !0, retainedNodes: !2)
[codeview] Begin to add support for inlined call sites Summary: There are three parts to inlined call frames: 1. The inlinee line subsection 2. The inline site symbol record 3. The function ids referenced by both This change starts by emitting function ids (3) for all subprograms and emitting the base inline site symbol record (2). The actual line numbers in (2) use an encoded format that will come next, along with the inlinee line subsection. Reviewers: majnemer Subscribers: llvm-commits Differential Revision: http://reviews.llvm.org/D16333 llvm-svn: 259217
2016-01-30 02:16:43 +08:00
!8 = !{i32 2, !"CodeView", i32 1}
!9 = !{i32 2, !"Debug Info Version", i32 3}
!10 = !{i32 1, !"PIC Level", i32 2}
!11 = !{!"clang version 3.9.0 "}
!12 = !DILocation(line: 14, column: 5, scope: !4)
!13 = !{!14, !14, i64 0}
!14 = !{!"int", !15, i64 0}
!15 = !{!"omnipotent char", !16, i64 0}
!16 = !{!"Simple C/C++ TBAA"}
!17 = !DILocation(line: 9, column: 5, scope: !6, inlinedAt: !18)
!18 = distinct !DILocation(line: 15, column: 3, scope: !4)
!19 = !DILocation(line: 3, column: 3, scope: !7, inlinedAt: !20)
!20 = distinct !DILocation(line: 10, column: 3, scope: !6, inlinedAt: !18)
!21 = !DILocation(line: 3, column: 7, scope: !7, inlinedAt: !20)
!22 = !DILocation(line: 4, column: 8, scope: !7, inlinedAt: !20)
!23 = !DILocation(line: 4, column: 5, scope: !7, inlinedAt: !20)
!24 = !DILocation(line: 5, column: 5, scope: !7, inlinedAt: !20)
!25 = !DILocation(line: 6, column: 5, scope: !7, inlinedAt: !20)
!26 = !DILocation(line: 7, column: 1, scope: !7, inlinedAt: !20)
!27 = !DILocation(line: 11, column: 5, scope: !6, inlinedAt: !18)
!28 = !DILocation(line: 16, column: 5, scope: !4)
!29 = !DILocation(line: 17, column: 1, scope: !4)