Author: Rong Xu
Date: 2022-10-23T22:33:52-07:00
New Revision: 6cee5393371fdde798605c88bad0ebceb3626257
URL:
https://github.com/llvm/llvm-project/commit/6cee5393371fdde798605c88bad0ebceb3626257
DIFF:
https://github.com/llvm/llvm-project/commit/6cee5393371fdde798605c88bad0ebceb3626257.diff
LOG:
Author: Rong Xu
Date: 2022-08-29T16:15:12-07:00
New Revision: db18f26567be47ec1da7fba48179a3001f0d34ab
URL:
https://github.com/llvm/llvm-project/commit/db18f26567be47ec1da7fba48179a3001f0d34ab
DIFF:
https://github.com/llvm/llvm-project/commit/db18f26567be47ec1da7fba48179a3001f0d34ab.diff
LOG:
Author: Rong Xu
Date: 2021-08-18T18:37:35-07:00
New Revision: 5fdaaf7fd8f35ac9c9de50a45b09e29c7b0d48c4
URL:
https://github.com/llvm/llvm-project/commit/5fdaaf7fd8f35ac9c9de50a45b09e29c7b0d48c4
DIFF:
https://github.com/llvm/llvm-project/commit/5fdaaf7fd8f35ac9c9de50a45b09e29c7b0d48c4.diff
LOG:
Author: Rong Xu
Date: 2021-08-16T10:15:30-07:00
New Revision: b7425e956be60a73004d7ae5bb37da85872c29fb
URL:
https://github.com/llvm/llvm-project/commit/b7425e956be60a73004d7ae5bb37da85872c29fb
DIFF:
https://github.com/llvm/llvm-project/commit/b7425e956be60a73004d7ae5bb37da85872c29fb.diff
LOG:
Author: Rong Xu
Date: 2020-12-17T18:41:12-08:00
New Revision: 3733463dbb58a29892be3872bd32f93cb9af492c
URL:
https://github.com/llvm/llvm-project/commit/3733463dbb58a29892be3872bd32f93cb9af492c
DIFF:
https://github.com/llvm/llvm-project/commit/3733463dbb58a29892be3872bd32f93cb9af492c.diff
LOG:
yes. Also #43 should be #42
I fixed it in
https://reviews.llvm.org/D92493
But that was just to make my test pass.
On Wed, Dec 16, 2020 at 11:35 AM Gulfem Savrun Yeniceri via Phabricator <
revi...@reviews.llvm.org> wrote:
> gulfem added a comment.
>
> > The problem in `llvm/test/Bitcode/attributes
Thanks for catching that.
I messed up the commit from another patch.
Fixed in commit c36f31c.
-Rong
On Mon, Dec 14, 2020 at 4:50 PM dmajor via Phabricator <
revi...@reviews.llvm.org> wrote:
> dmajor added inline comments.
>
>
>
> Comment at: clang/lib/CodeGen/CGCall.cpp:1948
> +
Author: Rong Xu
Date: 2020-12-14T18:41:49-08:00
New Revision: c36f31c4db065e987160a776749f5da81efc24c5
URL:
https://github.com/llvm/llvm-project/commit/c36f31c4db065e987160a776749f5da81efc24c5
DIFF:
https://github.com/llvm/llvm-project/commit/c36f31c4db065e987160a776749f5da81efc24c5.diff
LOG:
Author: Rong Xu
Date: 2020-12-14T15:56:10-08:00
New Revision: 54e03d03a7a4d47f09d40bcbcfe484066a52a077
URL:
https://github.com/llvm/llvm-project/commit/54e03d03a7a4d47f09d40bcbcfe484066a52a077
DIFF:
https://github.com/llvm/llvm-project/commit/54e03d03a7a4d47f09d40bcbcfe484066a52a077.diff
LOG:
Author: Rong Xu
Date: 2020-07-22T15:01:53-07:00
New Revision: 50da55a58534e9207d8d5e31c8b4b5cf0c624175
URL:
https://github.com/llvm/llvm-project/commit/50da55a58534e9207d8d5e31c8b4b5cf0c624175
DIFF:
https://github.com/llvm/llvm-project/commit/50da55a58534e9207d8d5e31c8b4b5cf0c624175.diff
LOG:
Author: Rong Xu
Date: 2020-02-25T08:15:17-08:00
New Revision: 11857d49948b845dcfd7c7f78595095e3add012d
URL:
https://github.com/llvm/llvm-project/commit/11857d49948b845dcfd7c7f78595095e3add012d
DIFF:
https://github.com/llvm/llvm-project/commit/11857d49948b845dcfd7c7f78595095e3add012d.diff
LOG:
OK. I will add a warning but not turn it off at the CodeGenOpt (but it
will not pass to LLVMContext). I will update the patch shortly.
On Fri, Feb 21, 2020 at 1:53 AM serge via Phabricator <
revi...@reviews.llvm.org> wrote:
> serge-sans-paille added a comment.
>
> Let's go that way then. @xur I
Author: Rong Xu
Date: 2020-01-14T16:32:17-08:00
New Revision: c9ee5e996e3c89a751a35e8b771870e0ec24f3c0
URL:
https://github.com/llvm/llvm-project/commit/c9ee5e996e3c89a751a35e8b771870e0ec24f3c0
DIFF:
https://github.com/llvm/llvm-project/commit/c9ee5e996e3c89a751a35e8b771870e0ec24f3c0.diff
LOG:
Author: Rong Xu
Date: 2020-01-14T15:44:57-08:00
New Revision: 60d39479221d6bc09060f7816bcd7c54eb286603
URL:
https://github.com/llvm/llvm-project/commit/60d39479221d6bc09060f7816bcd7c54eb286603
DIFF:
https://github.com/llvm/llvm-project/commit/60d39479221d6bc09060f7816bcd7c54eb286603.diff
LOG:
Thanks for reviewing this.
The new constructor is to avoid initializing the "OS" field (which is not
used by the empty BackendConsumer. Setting this field will affect the
output). I will add a comment there.
-Rong
On Mon, Jan 13, 2020 at 4:50 PM Teresa Johnson via Phabricator <
revi...@reviews.l
r367657 should fix this.
On Fri, Aug 2, 2019 at 12:20 AM Rong Xu wrote:
> yes. I already know this issue as this also breaks windows bolt.
> The fix should be simple.
>
> -Rong
>
> On Thu, Aug 1, 2019 at 11:52 PM Petr Hosek via Phabricator <
> revi...@reviews.llvm.org> wrote:
>
>> phosek added a
Author: xur
Date: Fri Aug 2 00:21:50 2019
New Revision: 367657
URL: http://llvm.org/viewvc/llvm-project?rev=367657&view=rev
Log:
[PGO] Fix bolt failures from r367628
Relaxed the check in a test because the windows bolt generates different
profile variables.
Modified:
cfe/trunk/test/Profile/
yes. I already know this issue as this also breaks windows bolt.
The fix should be simple.
-Rong
On Thu, Aug 1, 2019 at 11:52 PM Petr Hosek via Phabricator <
revi...@reviews.llvm.org> wrote:
> phosek added a comment.
>
> Looks like this change broke `Profile/gcc-flag-compatibility.c` test on
> m
Author: xur
Date: Thu Aug 1 15:36:34 2019
New Revision: 367628
URL: http://llvm.org/viewvc/llvm-project?rev=367628&view=rev
Log:
[PGO] Add PGO support at -O0 in the experimental new pass manager
Add PGO support at -O0 in the experimental new pass manager to sync the
behavior of the legacy pass m
On Fri, Jun 28, 2019 at 5:28 PM Xinliang David Li
wrote:
> I agree that the test coverage needs to be improved eg better check etc.
>
PGO is rarely run under -O0. But I agree that we should improve the test.
I sent out https://reviews.llvm.org/D64029. Chanlder: could you take a look?
Thanks,
they are not doing the exacly the same thing for old pass manager and new
pass manger: old pass manger is doing instrumentation, while the new pass
manager with this change is NOT.
the test is not check instrumentation, (it only check the variables that
generates by InstroProfiling pass).
In this s
Author: xur
Date: Thu Apr 25 14:16:41 2019
New Revision: 359241
URL: http://llvm.org/viewvc/llvm-project?rev=359241&view=rev
Log:
[PGO] Fix buildbot failure in 359215
Revert the part of changes in r359215 that failing in some platforms.
I will re-enable them later.
Modified:
cfe/trunk/test/P
Author: xur
Date: Thu Apr 25 10:52:43 2019
New Revision: 359215
URL: http://llvm.org/viewvc/llvm-project?rev=359215&view=rev
Log:
[PGO] Enable InstrProf lowering for Clang PGO instrumentation in the new pass
manager
Currently InstrProf lowering is not enabled for Clang PGO instrumentation in
the
Author: xur
Date: Wed Mar 6 15:00:38 2019
New Revision: 355563
URL: http://llvm.org/viewvc/llvm-project?rev=355563&view=rev
Log:
[PGO] Re-submit: Clang part of change for context-sensitive PGO (part2)
Part 2 of CSPGO change in Clang: Add test cases.
Differential Revision: https://reviews.llvm.o
Author: xur
Date: Tue Mar 5 15:02:06 2019
New Revision: 355454
URL: http://llvm.org/viewvc/llvm-project?rev=355454&view=rev
Log:
Revert r355432 for buildbot failures in ppc64be-linux and s390x-linux
Removed:
cfe/trunk/test/CodeGen/Inputs/pgotestir.proftext
cfe/trunk/test/CodeGen/Inputs/p
Author: xur
Date: Tue Mar 5 11:09:56 2019
New Revision: 355432
URL: http://llvm.org/viewvc/llvm-project?rev=355432&view=rev
Log:
[PGO] Clang part of change for context-sensitive PGO (part2)
Part 2 of CSPGO change in Clang: Add test cases.
Differential Revision: https://reviews.llvm.org/D54176
Author: xur
Date: Mon Mar 4 12:21:31 2019
New Revision: 355331
URL: http://llvm.org/viewvc/llvm-project?rev=355331&view=rev
Log:
[PGO] Clang part of change for context-sensitive PGO (part1)
Part 1 of CSPGO change in Clang. This includes changes in clang options
and calls to llvm PassManager. Tes
Author: xur
Date: Fri Mar 1 09:50:20 2019
New Revision: 355209
URL: http://llvm.org/viewvc/llvm-project?rev=355209&view=rev
Log:
[PGO] Use the explicit parameter in ProfileSummary API. NFC
Use the explicit parameter in setProfileSummary() and getSummary().
This is a follow-up of r355131.
Modifi
xur added a comment.
Thanks for doing this. I like the patch in general. Hers are my minor
feedbacks:
(1) It seems the original options of -fprofile-instr-generate and
-fprofile-instr-use will be kept. Is so, what is the guideline to use these two
sets of options. Also, we need some documentat
Author: xur
Date: Fri Apr 22 16:19:05 2016
New Revision: 267224
URL: http://llvm.org/viewvc/llvm-project?rev=267224&view=rev
Log:
PGO] PGOFuncName meta data if PGOFuncName is different from function's raw name
Write out the PGOFuncName meta data if PGOFuncName is different from function's
raw nam
This revision was automatically updated to reflect the committed changes.
Closed by commit rL267224: PGO] PGOFuncName meta data if PGOFuncName is
different from function's raw name (authored by xur).
Changed prior to commit:
http://reviews.llvm.org/D18624?vs=54717&id=54725#toc
Repository:
rL
xur added a comment.
tested with povray with full path names in the command line and it works fine.
http://reviews.llvm.org/D18624
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
xur updated this revision to Diff 54717.
xur added a comment.
Handling the path-stripped prefix in PGOFuncName.
This new patch depends on http://reviews.llvm.org/D19433
http://reviews.llvm.org/D18624
Files:
CodeGenPGO.cpp
Index: CodeGenPGO.cpp
===
I was using spec-run where the command line has the pathless filename.
The source in Atam's command-line has the absolute path.
In meta-data creation, we used Module's getSourceFileName() which has the
source name appeared in the command line (in this case, a full patch name).
While in Clang's se
xur added a comment.
I was using spec-run where the command line has the pathless filename.
The source in Atam's command-line has the absolute path.
In meta-data creation, we used Module's getSourceFileName() which has the
source name appeared in the command line (in this case, a full patch name)
xur added a comment.
I tried http://reviews.llvm.org/D17864 and http://reviews.llvm.org/D18624 with
the latest trunk of r267006 on a clean
client.
The built compiler works fine with povray.
the meta data only has the files name.
How did you build povray?
can you show me your command line in bui
I tried D17864 and D18624 with the latest trunk of r267006 on a clean
client.
The built compiler works fine with povray.
the meta data only has the files name.
How did you build povray?
can you show me your command line in building (for example, csg.cpp) for
both--fprofile-instr-use and -fprofile
xur added a comment.
these two patches work fine in my build. My llvm was on top ov r265491. The
opt remarks in the last email were emitted by the compiler.
I'll try to reproduce with r266465.
http://reviews.llvm.org/D18624
___
cfe-commits mailing
these two patches work fine in my build. My llvm was on top ov r265491. The
opt remarks in the last email were emitted by the compiler.
I'll try to reproduce with r266465.
On Apr 20, 2016 4:00 PM, "Adam Nemet" wrote:
> anemet added a comment.
>
> Still does not to work. The metadata has the f
xur added a comment.
the reason for splitting the check is we don't need this check at all in
llvm instrument,-- as it's done per function. but I guess it does not
matter much to move it in to create -- it's only called once per function.
http://reviews.llvm.org/D18624
___
the reason for splitting the check is we don't need this check at all in
llvm instrument,-- as it's done per function. but I guess it does not
matter much to move it in to create -- it's only called once per function.
On Wed, Apr 20, 2016 at 3:04 PM, David Li wrote:
> davidxl added inline co
xur updated this revision to Diff 54275.
xur added a comment.
Previous patch was bad (as David noticed) -- we might not annotate all the
functions that we interested.
This updated patch should work.
BTW, I got the following promotions for 453.provay:
LLVM gold plugin: csg.cpp:157:15: Promote i
I did not test Clang based instrumentation with SPEC. I will try it with
povray today.
On Tue, Apr 19, 2016 at 9:40 AM, Adam Nemet wrote:
> anemet added a comment.
>
> As discussed under http://reviews.llvm.org/D17864, I did a run with this
> and I don't get the indirect call promoted that calls
xur added a comment.
I did not test Clang based instrumentation with SPEC. I will try it with
povray today.
http://reviews.llvm.org/D18624
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-com
xur updated this revision to Diff 52134.
xur added a comment.
sync the change in http://reviews.llvm.org/D18623.
http://reviews.llvm.org/D18624
Files:
lib/CodeGen/CodeGenPGO.cpp
Index: lib/CodeGen/CodeGenPGO.cpp
===
--- lib/Code
xur created this revision.
xur added reviewers: davidxl, bogner, joker.eph.
xur added subscribers: cfe-commits, vsk, xur.
Write out the PGOFuncName meta data if PGOFuncName is different from function's
raw name. This should only apply to internal linkage functions. This is to be
consumed by indi
This revision was automatically updated to reflect the committed changes.
Closed by commit rL262515: [PGO] Change profile use cc1 option to handle IR
level profiles (authored by xur).
Changed prior to commit:
http://reviews.llvm.org/D17737?vs=49655&id=49664#toc
Repository:
rL LLVM
http://re
Author: xur
Date: Wed Mar 2 14:59:36 2016
New Revision: 262515
URL: http://llvm.org/viewvc/llvm-project?rev=262515&view=rev
Log:
[PGO] Change profile use cc1 option to handle IR level profiles
This patch changes cc1 option for PGO profile use from
-fprofile-instr-use= to -fprofile-instrument-use
xur updated this revision to Diff 49655.
xur added a comment.
Integrated with Sean's review comments.
-Rong
http://reviews.llvm.org/D17737
Files:
include/clang/Driver/CC1Options.td
include/clang/Driver/Options.td
include/clang/Frontend/CodeGenOptions.def
include/clang/Frontend/CodeGenO
xur updated this revision to Diff 49647.
xur marked an inline comment as done.
xur added a comment.
I think David's suggestion to use one cc1 option -fprofile-instrument-use-path
is a very good idea.
I update the patch to get rid of -fprofille-instrument-use.
The profile kind is now set by check
xur marked an inline comment as done.
Comment at: lib/CodeGen/CodeGenModule.cpp:154
@@ +153,3 @@
+assert(!CodeGenOpts.ProfileInstrumentUsePath.empty() &&
+ "Need to explicitly specify the profile name.");
+auto ReaderOrErr = llvm::IndexedInstrProfReader::create(
xur updated this revision to Diff 49444.
xur marked 2 inline comments as done.
xur added a comment.
Integrated Sean's review suggestions.
-Rong
http://reviews.llvm.org/D17737
Files:
include/clang/Driver/CC1Options.td
include/clang/Driver/Options.td
include/clang/Frontend/CodeGenOptions.d
xur marked 4 inline comments as done.
Comment at: lib/Frontend/CompilerInvocation.cpp:378
@@ -377,1 +377,3 @@
+enum PGOInstrumentor { ProfileUnknown, ProfileNone, ProfileClang, ProfileLLVM
};
+static PGOInstrumentor getPGOInstrumentor(StringRef S) {
silvas wrot
xur created this revision.
xur added reviewers: davidxl, silvas, bogner.
xur added subscribers: davidxl, xur, cfe-commits.
This patch takes the suggestion by Sean Silva in
https://reviews.llvm.org/D17622 to decouple cc1 option from driver level option.
-fprofile-instr-use={} is now a driver only
Author: xur
Date: Mon Feb 29 12:54:59 2016
New Revision: 262239
URL: http://llvm.org/viewvc/llvm-project?rev=262239&view=rev
Log:
[PGO] clang cc1 option change to enable IR level instrumentation
This patch expands cc1 option -fprofile-instrument= with a new value:
-fprofile-instrument=llvm
which
This revision was automatically updated to reflect the committed changes.
Closed by commit rL262239: [PGO] clang cc1 option change to enable IR level
instrumentation (authored by xur).
Changed prior to commit:
http://reviews.llvm.org/D17622?vs=49243&id=49394#toc
Repository:
rL LLVM
http://r
xur updated this revision to Diff 49243.
xur added a comment.
Sean, Thanks for the review and suggestions.
Here is the patch that only adds -fprofile-instrument=llvm.
I'll have another patch using your suggestion on cc1 options for profile use
compilation.
http://reviews.llvm.org/D17622
File
Create a new review here:
http://reviews.llvm.org/D17622
Thanks,
-Rong
On Wed, Feb 24, 2016 at 9:22 PM, Sean Silva wrote:
> silvas added a comment.
>
> In http://reviews.llvm.org/D15829#360006, @xur wrote:
>
>> Here is the new patch that removes the auto detection of profile kind.
>>
>> In this
xur added a comment.
Create a new review here:
http://reviews.llvm.org/D17622
Thanks,
-Rong
http://reviews.llvm.org/D15829
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
xur created this revision.
xur added reviewers: davidxl, silvas, justinruggles.
xur added subscribers: cfe-commits, xur, mcrosier, davidxl, slingn.
This patch expands cc1 option -fprofile-instrument= with three new values:
(1) -fprofile-instrument=llvm
This enables IR PGO instrumentation.
(2) -fpr
xur updated this revision to Diff 48859.
xur marked 2 inline comments as done.
xur added a comment.
Integrated with David's comments.
Thanks,
-Rong
http://reviews.llvm.org/D15829
Files:
include/clang/Driver/CC1Options.td
include/clang/Driver/Options.td
include/clang/Frontend/CodeGenOpti
xur marked 4 inline comments as done.
Comment at: lib/CodeGen/CodeGenModule.cpp:151
@@ -151,1 +150,3 @@
+ if (CodeGenOpts.hasProfileClangUse() &&
+ !CodeGenOpts.InstrProfileInput.empty()) {
auto ReaderOrErr =
davidxl wrote:
> Is the second condition sti
xur updated this revision to Diff 48846.
xur marked an inline comment as done.
xur added a comment.
Here is the new patch that removes the auto detection of profile kind.
In this patch, I replace the CC1 option of -fprofile-instr-use=<> with
-fprofile-instrument={llvm-use|clang-use}. For the use
xur marked an inline comment as done.
xur added a comment.
I'll send a revised patch soon.
Comment at: lib/CodeGen/CodeGenModule.cpp:150
@@ -149,2 +149,3 @@
- if (!CodeGenOpts.InstrProfileInput.empty()) {
+ if (!CodeGenOpts.hasProfileIRInstr() &&
+ !CodeGenOpts.InstrPro
xur updated this revision to Diff 47331.
xur added a comment.
Ping
(also added the test that missed from last upload).
Thanks,
-Rong
http://reviews.llvm.org/D15829
Files:
include/clang/Driver/CC1Options.td
include/clang/Frontend/CodeGenOptions.h
lib/CodeGen/BackendUtil.cpp
lib/CodeGe
xur updated this revision to Diff 47251.
xur marked an inline comment as done.
xur added a comment.
fixed the typo in comments
http://reviews.llvm.org/D15829
Files:
include/clang/Driver/CC1Options.td
include/clang/Frontend/CodeGenOptions.h
lib/CodeGen/BackendUtil.cpp
lib/CodeGen/CodeGen
Author: xur
Date: Mon Feb 8 12:12:41 2016
New Revision: 260116
URL: http://llvm.org/viewvc/llvm-project?rev=260116&view=rev
Log:
Fix missing space (NFC)
Fixed the inconsistently placed : (missing space) introduced in in r259811.
Modified:
cfe/trunk/include/clang/Basic/DiagnosticDriverKin
xur updated this revision to Diff 46960.
xur added a comment.
Now http://reviews.llvm.org/D16730 has been committed (as r259811)
Here is the patch that adds cc1 option -fprofile-instrument=llvm to enable IR
level PGO generate and use.
The pgo use part of the patch depends http://reviews.llvm.or
Author: xur
Date: Thu Feb 4 12:39:09 2016
New Revision: 259811
URL: http://llvm.org/viewvc/llvm-project?rev=259811&view=rev
Log:
[PGO] cc1 option name change for profile instrumentation
This patch changes cc1 option -fprofile-instr-generate to an enum option
-fprofile-instrument={clang|none}. It
xur added a comment.
To make the review easier, I split the cc1 names change part into a new NFC
patch here.
http://reviews.llvm.org/D16730
http://reviews.llvm.org/D15829
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org
xur updated this revision to Diff 46389.
xur added a comment.
This new patch adds the change the clang test cases (cc1 option
-fprofile-instr-generate to -fprofile-instrument=Clang).
It also replaces cc1 option -fprofile-instr-generate= to
-fprofile-instrument-path=, as suggested by David.
Tha
xur added a comment.
The previous patch was not good as it failed 58 tests that use
-fprofile-instr-generate as a cc1 option.
I can see two methods to solve this:
(1) we change all the 58 tests to use -fprofile-instrument=Clang option.
(2) Fall back to my previous patch: keep -fprofile-instr-gene
The previous patch was not good as it failed 58 tests that use
-fprofile-instr-generate as a cc1 option.
I can see two methods to solve this:
(1) we change all the 58 tests to use -fprofile-instrument=Clang option.
(2) Fall back to my previous patch: keep -fprofile-instr-generate as
the CC1 option
xur updated this revision to Diff 46303.
xur added a comment.
Integrated most recently comments/suggestions from David and Sean.
Thanks,
-Rong
http://reviews.llvm.org/D15829
Files:
include/clang/Basic/DiagnosticDriverKinds.td
include/clang/Driver/CC1Options.td
include/clang/Driver/Optio
xur added a comment.
I'll send an updated patch shortly.
Comment at: include/clang/Frontend/CodeGenOptions.def:106
@@ -105,3 +105,3 @@
-CODEGENOPT(ProfileInstrGenerate , 1, 0) ///< Instrument code to generate
-///< execution counts to us
xur updated this revision to Diff 46199.
xur added a comment.
This new version implemented the usage model David proposed. We now have a cc1
option -fprofile-instrumentor={llvm | clang} to choose which instrumentation to
use. The slight difference from David's proposal is that we can use this op
On Wed, Jan 27, 2016 at 11:31 AM, David Li wrote:
> davidxl added inline comments.
>
>
> Comment at: lib/Driver/Tools.cpp:5520
> @@ +5519,3 @@
> +A->claim();
> +if ((StringRef(A->getValue(0)) == "-fprofile-ir-instr") &&
> +(Args.hasArg(options::OPT_fprofile_instr_g
xur added a comment.
Sean: Adding a new CC1 option ProfileClangInstr will make things
cleaner. But this won't solve the problem. The root of all the mess is
there is no driver level option for IR instrumentation. I need to
either "hijack" the -Xclang option, or move the logic to
CompilerInvocation
Sean: Adding a new CC1 option ProfileClangInstr will make things
cleaner. But this won't solve the problem. The root of all the mess is
there is no driver level option for IR instrumentation. I need to
either "hijack" the -Xclang option, or move the logic to
CompilerInvocation.cpp, which both you d
xur updated this revision to Diff 45929.
xur marked an inline comment as done.
xur added a comment.
Here is the new patch. Changes from previous patch are:
(1) rename codegen option ProfileInstrGenerate to ClangProfInstrGen to avoid
the name confusion (Davidxl). This option is for clang instrumen
xur added a subscriber: xur.
xur added a comment.
OK. Now I understand what you meant. I was trying to avoid driver changes.
It seems you are saying I can change the driver to support cc1 options more
cleanly. This can be done.
Thanks,
Rong
http://reviews.llvm.org/D15829
___
OK. Now I understand what you meant. I was trying to avoid driver changes.
It seems you are saying I can change the driver to support cc1 options more
cleanly. This can be done.
Thanks,
Rong
On Jan 25, 2016 12:43, "Sean Silva" wrote:
> silvas added inline comments.
>
>
> Commen
xur added inline comments.
Comment at: lib/CodeGen/BackendUtil.cpp:440
@@ +439,3 @@
+ if (CodeGenOpts.ProfileIRInstr) {
+// Should not have ProfileInstrGenerate set -- it is for clang
+// instrumentation only.
silvas wrote:
> Then change the existing refe
xur added inline comments.
Comment at: lib/Driver/Tools.cpp:3279
@@ -3278,1 +3278,3 @@
+
+ Args.AddAllArgs(CmdArgs, options::OPT_fprofile_ir_instr);
}
mcrosier wrote:
> I don't think AddAllArgs is what you really want. What if the user specifies
> the option t
xur updated this revision to Diff 45708.
xur added a comment.
This new patches integrates Sean review comments:
(1) make -fprofile-ir-instr a cc1 option instead of driver option.
(2) add one cc1 option test, and test the pass is indeed invoked.
(3) remove the driver test (runtime library).
(4) fix
On Thu, Jan 21, 2016 at 7:32 PM, Sean Silva wrote:
> silvas added a comment.
>
> @slingn and I had a discussion offline about the potential names and came up
> with some ideas, but none is a clear winner.
>
> Overall, my feeling is that from a user's perspective, the frontend stuff is
> probably
xur added a comment.
Ping.
Passmanagerbuilder change has been committed. Could you take a look at this
command line option patch?
Thanks,
-Rong
http://reviews.llvm.org/D15829
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.l
xur updated this revision to Diff 44776.
xur added a comment.
added new tests and comments based on the review comments.
http://reviews.llvm.org/D15829
Files:
include/clang/Driver/Options.td
include/clang/Frontend/CodeGenOptions.def
lib/CodeGen/BackendUtil.cpp
lib/CodeGen/CodeGenModule.
xur marked an inline comment as done.
xur added a comment.
I'll sent the revised patch based on the comments shortly.
Comment at: lib/CodeGen/BackendUtil.cpp:435
@@ +434,3 @@
+ if (CodeGenOpts.ProfileIRInstr) {
+assert (!CodeGenOpts.ProfileInstrGenerate);
+if (!CodeGenO
xur created this revision.
xur added reviewers: davidxl, silvas, bogner.
xur added a subscriber: cfe-commits.
This patch introduce a new toggle option -fprofile-ir-instr that enables IR
level instrumentation. It needs to be used in combination with current PGO
options. Without an existing PGO op
90 matches
Mail list logo
