maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Other/change-printer.ll

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

327 lines
20 KiB
LLVM
Raw Normal View History

Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; Simple checks of -print-changed functionality
;
; Note that (mostly) only the banners are checked.
;
; Simple functionality check.
; RUN: opt -S -print-changed -passes=instsimplify 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-SIMPLE
;
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; Simple functionality check.
; RUN: opt -S -print-changed= -passes=instsimplify 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-SIMPLE
;
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; Check that only the passes that change the IR are printed and that the
; others (including g) are filtered out.
; RUN: opt -S -print-changed -passes=instsimplify -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FUNC-FILTER
;
; Check that the reporting of IRs respects -print-module-scope
; RUN: opt -S -print-changed -passes=instsimplify -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-PRINT-MOD-SCOPE
;
; Check that the reporting of IRs respects -print-module-scope
; RUN: opt -S -print-changed -passes=instsimplify -filter-print-funcs=f -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FUNC-FILTER-MOD-SCOPE
;
; Check that reporting of multiple functions happens
; RUN: opt -S -print-changed -passes=instsimplify -filter-print-funcs="f,g" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FILTER-MULT-FUNC
;
; Check that the reporting of IRs respects -filter-passes
; RUN: opt -S -print-changed -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FILTER-PASSES
;
; Check that the reporting of IRs respects -filter-passes with multiple passes
; RUN: opt -S -print-changed -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FILTER-MULT-PASSES
;
; Check that the reporting of IRs respects both -filter-passes and -filter-print-funcs
; RUN: opt -S -print-changed -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FILTER-FUNC-PASSES
;
; Check that the reporting of IRs respects -filter-passes, -filter-print-funcs and -print-module-scope
; RUN: opt -S -print-changed -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" -filter-print-funcs=f -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FILTER-FUNC-PASSES-MOD-SCOPE
;
; Check that repeated passes that change the IR are printed and that the
; others (including g) are filtered out. Note that the second time
; instsimplify is run on f, it does not change the IR
; RUN: opt -S -print-changed -passes="instsimplify,instsimplify" -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-MULT-PASSES-FILTER-FUNC
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
;
; Simple print-before-changed functionality check.
; RUN: opt -S -print-changed -print-before-changed -passes=instsimplify 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-SIMPLE-BEFORE
;
; Check print-before-changed obeys the function filtering
; RUN: opt -S -print-changed -print-before-changed -passes=instsimplify -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FUNC-FILTER-BEFORE
;
; Check that the reporting of IRs with -print-before-changed respects -print-module-scope
; RUN: opt -S -print-changed -print-before-changed -passes=instsimplify -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-PRINT-MOD-SCOPE-BEFORE
;
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; Simple checks of -print-changed=quiet functionality
;
; Simple functionality check.
; RUN: opt -S -print-changed=quiet -passes=instsimplify 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-SIMPLE
;
; Check that only the passes that change the IR are printed and that the
; others (including g) are filtered out.
; RUN: opt -S -print-changed=quiet -passes=instsimplify -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FUNC-FILTER
;
; Check that the reporting of IRs respects -print-module-scope
; RUN: opt -S -print-changed=quiet -passes=instsimplify -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-PRINT-MOD-SCOPE
;
; Check that the reporting of IRs respects -print-module-scope
; RUN: opt -S -print-changed=quiet -passes=instsimplify -filter-print-funcs=f -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FUNC-FILTER-MOD-SCOPE
;
; Check that reporting of multiple functions happens
; RUN: opt -S -print-changed=quiet -passes=instsimplify -filter-print-funcs="f,g" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-MULT-FUNC
;
; Check that the reporting of IRs respects -filter-passes
; RUN: opt -S -print-changed=quiet -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-PASSES-NONE --allow-empty
;
; Check that the reporting of IRs respects -filter-passes with multiple passes
; RUN: opt -S -print-changed=quiet -passes="instsimplify" -filter-passes="NoOpFunctionPass,InstSimplifyPass" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-PASSES
;
; Check that the reporting of IRs respects -filter-passes with multiple passes
; RUN: opt -S -print-changed=quiet -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-MULT-PASSES
;
; Check that the reporting of IRs respects both -filter-passes and -filter-print-funcs
; RUN: opt -S -print-changed=quiet -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-FUNC-PASSES
;
; Check that the reporting of IRs respects -filter-passes, -filter-print-funcs and -print-module-scope
; RUN: opt -S -print-changed=quiet -passes="instsimplify,no-op-function" -filter-passes="NoOpFunctionPass,InstSimplifyPass" -filter-print-funcs=f -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-FILTER-FUNC-PASSES-MOD-SCOPE
;
; Check that repeated passes that change the IR are printed and that the
; others (including g) are filtered out. Note that the second time
; instsimplify is run on f, it does not change the IR
; RUN: opt -S -print-changed=quiet -passes="instsimplify,instsimplify" -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-QUIET-MULT-PASSES-FILTER-FUNC
;
; Simple print-before-changed functionality check.
; RUN: opt -S -print-changed=quiet -print-before-changed -passes=instsimplify 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-SIMPLE-BEFORE-QUIET
;
; Check print-before-changed obeys the function filtering
; RUN: opt -S -print-changed=quiet -print-before-changed -passes=instsimplify -filter-print-funcs=f 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-FUNC-FILTER-BEFORE-QUIET
;
; Check that the reporting of IRs with -print-before-changed respects -print-module-scope
; RUN: opt -S -print-changed=quiet -print-before-changed -passes=instsimplify -print-module-scope 2>&1 -o /dev/null < %s | FileCheck %s --check-prefix=CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET
;
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
define i32 @g() {
entry:
%a = add i32 2, 3
ret i32 %a
}
define i32 @f() {
entry:
%a = add i32 2, 3
ret i32 %a
}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-SIMPLE-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE: *** IR Dump After VerifierPass on [module] omitted because no change ***
; CHECK-SIMPLE: *** IR Dump After InstSimplifyPass on g ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-SIMPLE-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE: *** IR Pass PassManager{{.*}} on g ignored ***
; CHECK-SIMPLE: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-SIMPLE-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE: *** IR Pass PassManager{{.*}} on f ignored ***
; CHECK-SIMPLE: *** IR Pass ModuleToFunctionPassAdaptor on [module] ignored ***
; CHECK-SIMPLE: *** IR Dump After VerifierPass on [module] omitted because no change ***
; CHECK-SIMPLE: *** IR Dump After PrintModulePass on [module] omitted because no change ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-SIMPLE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FUNC-FILTER-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FUNC-FILTER: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FUNC-FILTER-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-PRINT-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE: *** IR Dump After InstSimplifyPass on g ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-PRINT-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-PRINT-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-MOD-SCOPE: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FUNC-FILTER-MOD-SCOPE-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-MOD-SCOPE: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FUNC-FILTER-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FUNC-FILTER-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-FUNC: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-FUNC-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-FUNC: *** IR Dump After InstSimplifyPass on g ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-FUNC-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-FUNC: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-FUNC-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-PASSES: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FILTER-PASSES: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-PASSES-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-PASSES: *** IR Dump After NoOpFunctionPass on g omitted because no change ***
; CHECK-FILTER-PASSES: *** IR Dump After InstSimplifyPass on f filtered out ***
; CHECK-FILTER-PASSES: *** IR Dump After NoOpFunctionPass on f omitted because no change ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-PASSES: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-PASSES-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-PASSES: *** IR Dump After InstSimplifyPass on g ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-PASSES-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-PASSES: *** IR Dump After NoOpFunctionPass on g omitted because no change ***
; CHECK-FILTER-MULT-PASSES: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-MULT-PASSES-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-MULT-PASSES: *** IR Dump After NoOpFunctionPass on f omitted because no change ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FILTER-FUNC-PASSES: *** IR Dump After NoOpFunctionPass on g filtered out ***
; CHECK-FILTER-FUNC-PASSES: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-FUNC-PASSES-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-FUNC-PASSES-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES: *** IR Dump After NoOpFunctionPass on f omitted because no change ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump After NoOpFunctionPass on g filtered out ***
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump After NoOpFunctionPass on f omitted because no change ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-MULT-PASSES-FILTER-FUNC: *** IR Dump At Start ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-MULT-PASSES-FILTER-FUNC-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-MULT-PASSES-FILTER-FUNC: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-MULT-PASSES-FILTER-FUNC: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-MULT-PASSES-FILTER-FUNC: *** IR Dump After InstSimplifyPass on f ***
Reland No.3: Add new hidden option -print-changed which only reports changes to IR A new hidden option -print-changed is added along with code to support printing the IR as it passes through the opt pipeline in the new pass manager. Only those passes that change the IR are reported, with others only having the banner reported, indicating that they did not change the IR, were filtered out or ignored. Filtering of output via the -filter-print-funcs is supported and a new supporting hidden option -filter-passes is added. The latter takes a comma separated list of pass names and filters the output to only show those passes in the list that change the IR. The output can also be modified via the -print-module-scope function. The code introduces an abstract template base class that generalizes the comparison of IRs that takes an IR representation as template parameter. Derived classes provide overrides that provide an event based API for generalized reporting of IRs as they are changed in the opt pipeline through the new pass manager. The first of several instantiations is provided that prints the IR in a form similar to that produced by -print-after-all with the above mentioned filtering capabilities. This version, and the others to follow will be introduced at the upcoming developer's conference. Reviewed By: aeubanks (Arthur Eubanks), yrouban (Yevgeny Rouban), ychen (Yuanfang Chen), MaskRay (Fangrui Song) Differential Revision: https://reviews.llvm.org/D86360
2020-10-02 01:39:02 +08:00
; CHECK-MULT-PASSES-FILTER-FUNC-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-MULT-PASSES-FILTER-FUNC: *** IR Dump After InstSimplifyPass on f omitted because no change ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Dump At Start ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Dump After VerifierPass on [module] omitted because no change ***
; CHECK-SIMPLE-BEFORE: *** IR Dump Before InstSimplifyPass on g ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Dump After InstSimplifyPass on g ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Pass PassManager{{.*}} on g ignored ***
; CHECK-SIMPLE-BEFORE: *** IR Dump Before InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Dump After InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE: *** IR Pass PassManager{{.*}} on f ignored ***
; CHECK-SIMPLE-BEFORE: *** IR Pass ModuleToFunctionPassAdaptor on [module] ignored ***
; CHECK-SIMPLE-BEFORE: *** IR Dump After VerifierPass on [module] omitted because no change ***
; CHECK-SIMPLE-BEFORE: *** IR Dump After PrintModulePass on [module] omitted because no change ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-SIMPLE-BEFORE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-BEFORE: *** IR Dump At Start ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-FUNC-FILTER-BEFORE-NEXT: ; ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-BEFORE: *** IR Dump After InstSimplifyPass on g filtered out ***
; CHECK-FUNC-FILTER-BEFORE: *** IR Dump Before InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-FUNC-FILTER-BEFORE-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-BEFORE: *** IR Dump After InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-FUNC-FILTER-BEFORE-NEXT: define i32 @f()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE: *** IR Dump At Start ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE: *** IR Dump Before InstSimplifyPass on g ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE: *** IR Dump After InstSimplifyPass on g ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE: *** IR Dump Before InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE: *** IR Dump After InstSimplifyPass on f ***
Introduce -print-before-changed, making -print-changed also print before passes that modify IR Summary: Add an option -print-before-changed that modifies the print-changed behaviour so that it prints the IR before a pass that changed it in addition to printing the IR after the pass. Note that the option does nothing in isolation. The filtering options work as expected. Lit tests are included. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D88757
2020-11-12 23:20:46 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-NEXT: ModuleID = {{.+}}
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-SIMPLE-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-SIMPLE: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-SIMPLE-NEXT: define i32 @g()
; CHECK-QUIET-SIMPLE-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-SIMPLE: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-SIMPLE-NEXT: define i32 @f()
; CHECK-QUIET-SIMPLE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FUNC-FILTER-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FUNC-FILTER: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FUNC-FILTER-NEXT: define i32 @f()
; CHECK-QUIET-FUNC-FILTER-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-PRINT-MOD-SCOPE-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-PRINT-MOD-SCOPE: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-PRINT-MOD-SCOPE-NEXT: ModuleID = {{.+}}
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-PRINT-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-PRINT-MOD-SCOPE-NEXT: ModuleID = {{.+}}
; CHECK-QUIET-PRINT-MOD-SCOPE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FUNC-FILTER-MOD-SCOPE-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FUNC-FILTER-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FUNC-FILTER-MOD-SCOPE-NEXT: ModuleID = {{.+}}
; CHECK-QUIET-FUNC-FILTER-MOD-SCOPE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-MULT-FUNC-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-MULT-FUNC: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-MULT-FUNC-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-MULT-FUNC: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-MULT-FUNC-NEXT: define i32 @f()
; CHECK-QUIET-FILTER-MULT-FUNC-NOT: *** IR
; CHECK-QUIET-FILTER-PASSES-NONE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-PASSES-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-PASSES: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-PASSES-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-PASSES-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-PASSES: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-PASSES-NEXT: define i32 @f()
; CHECK-QUIET-FILTER-PASSES-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-MULT-PASSES-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-MULT-PASSES: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-MULT-PASSES-NEXT: define i32 @g()
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-MULT-PASSES-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-MULT-PASSES: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-MULT-PASSES-NEXT: define i32 @f()
; CHECK-QUIET-FILTER-MULT-PASSES-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-FUNC-PASSES-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-FUNC-PASSES: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-FUNC-PASSES-NEXT: define i32 @f()
; CHECK-QUIET-FILTER-FUNC-PASSES-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-FILTER-FUNC-PASSES-MOD-SCOPE-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-FILTER-FUNC-PASSES-MOD-SCOPE: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-FILTER-FUNC-PASSES-MOD-SCOPE-NEXT: ModuleID = {{.+}}
; CHECK-QUIET-FILTER-FUNC-PASSES-MOD-SCOPE-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-QUIET-MULT-PASSES-FILTER-FUNC-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-QUIET-MULT-PASSES-FILTER-FUNC: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-QUIET-MULT-PASSES-FILTER-FUNC-NEXT: define i32 @f()
; CHECK-QUIET-MULT-PASSES-FILTER-FUNC-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE-QUIET-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-SIMPLE-BEFORE-QUIET: *** IR Dump Before InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-SIMPLE-BEFORE-QUIET-NEXT: define i32 @g()
; CHECK-SIMPLE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE-QUIET: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-SIMPLE-BEFORE-QUIET-NEXT: define i32 @g()
; CHECK-SIMPLE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE-QUIET: *** IR Dump Before InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-SIMPLE-BEFORE-QUIET-NEXT: define i32 @f()
; CHECK-SIMPLE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-SIMPLE-BEFORE-QUIET: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-SIMPLE-BEFORE-QUIET-NEXT: define i32 @f()
; CHECK-SIMPLE-BEFORE-QUIET-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-BEFORE-QUIET-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-FUNC-FILTER-BEFORE-QUIET: *** IR Dump Before InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-FUNC-FILTER-BEFORE-QUIET-NEXT: define i32 @f()
; CHECK-FUNC-FILTER-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-FUNC-FILTER-BEFORE-QUIET: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-FUNC-FILTER-BEFORE-QUIET-NEXT: define i32 @f()
; CHECK-FUNC-FILTER-BEFORE-QUIET-NOT: *** IR
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NOT: *** IR Dump {{.*(At Start|no change|ignored|filtered out)}} ***
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET: *** IR Dump Before InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NEXT: ModuleID = {{.+}}
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET: *** IR Dump After InstSimplifyPass on g ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NEXT: ModuleID = {{.+}}
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET: *** IR Dump Before InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NEXT: ModuleID = {{.+}}
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NOT: *** IR Dump {{.*(no change|ignored|filtered out)}} ***
[NewPM] Cleanup IR printing instrumentation Being lazy with printing the banner seems hard to reason with, we should print it unconditionally first (it could also lead to duplicate banners if we have multiple functions in -filter-print-funcs). The printIR() functions were doing too many things. I separated out the call from PrintPassInstrumentation since we were essentially doing two completely separate things in printIR() from different callers. There were multiple ways to generate the name of some IR. That's all been moved to getIRName(). The printing of the IR name was also inconsistent, now it's always "IR Dump on $foo" where "$foo" is the name. For a function, it's the function name. For a loop, it's what's printed by Loop::print(), which is more detailed. For an SCC, it's the list of functions in parentheses. For a module it's "[module]", to differentiate between a possible SCC with a function called "module". To preserve D74814, we have to check if we're going to print anything at all first. This is unfortunate, but I would consider this a special case that shouldn't be handled in the core logic. Reviewed By: jamieschmeiser Differential Revision: https://reviews.llvm.org/D100231
2021-04-15 05:52:50 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET: *** IR Dump After InstSimplifyPass on f ***
Introduce new quiet mode and new option handling for -print-changed. Summary: Introduce a new mode of operation for -print-changed that only reports after a pass changes the IR with all of the other messages suppressed (ie, no initial IR and no messages about ignored, filtered or non-modifying passes). The option processing for -print-changed is changed to take an optional string indicating options for print-changed. Initially, the only option supported is quiet (as described above). This new quiet mode is specified with -print-changed=quiet while -print-changed will continue to function in the same way. It is intended that there will be more options in the future. Author: Jamie Schmeiser <schmeise@ca.ibm.com> Reviewed By: aeubanks (Arthur Eubanks) Differential Revision: https://reviews.llvm.org/D92589
2021-01-12 03:11:00 +08:00
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NEXT: ModuleID = {{.+}}
; CHECK-PRINT-MOD-SCOPE-BEFORE-QUIET-NOT: *** IR