skipping indirectly recursive inline chains.
To do this, we implicitly build an inline stack for each callsite and
check prior to inlining that doing so would not form a cycle. This uses
the exact same technique and even shares some code with the legacy PM
inliner.
This solution remains deeply unsatisfying to me because it means we
cannot actually iterate the inliner externally. Doing so would not be
able to easily detect and avoid such cycles. Some day I would very much
like to have a solution that works without this internal state to detect
cycles, but this is not that day.
llvm-svn: 290590
Nothing really interesting here, but I had to improve the test to use
variables rather than hard coding value names as we happen to end up
with different value names in the new PM.
llvm-svn: 290589
This mostly involved converting from grep to FileCheck and tidying up
the IR used.
In one case (invoke_test-3.ll) the test had become completely pointless
as we use 'resume' rather than 'unwind' now, and even then it did not
occur at the end of the line.
llvm-svn: 290570
inside of `InlineFunction`. Prior to this, call instructions are
specifically being rewritten and replaced within the inlined region,
invalidating some of the call sites.
Several of these regions are using the same technique to walk the
inlined region so this seems clearly safe up to this point.
I've also added a short circuit to the scan for call sites based on what
other code is doing.
With this, the most common crash I've found in the new inliner code is
fixed. I've turned it on for another test case that covers this
scenario.
I'll make my way through most of the other inliner test cases
just to get some easy coverage next.
llvm-svn: 290562
removing fully-dead comdats without removing dead entries in comdats
with live members.
This factors the core logic out of the current inliner's internals to
a reusable utility and leverages that in both places. The factored out
code should also be (minorly) more efficient in cases where we have very
few dead functions or dead comdats to consider.
I've added a test case to cover this behavior of the always inliner.
This is the last significant bug in the new PM's always inliner I've
found (so far).
llvm-svn: 290557
whether functions are removed, and fix the new PM's always inliner to
actually pass this test.
Without this, the new PM's always inliner leaves all the functions
kicking around which won't work out very well given the semantics of
always inline.
Doing this really highlights how frustrating the current alwaysinline
semantic contract is though -- why can we put it on *external*
functions, etc?
Also I've added a number of tricky and interesting test cases for
removing functions with the always inliner. There is one remaining case
not handled -- fully removing comdats -- and I've left a FIXME about
this.
llvm-svn: 290457
This patch renumbers the metadata nodes in debug info testcases after
https://reviews.llvm.org/D26769. This is a separate patch because it
causes so much churn. This was implemented with a python script that
pipes the testcases through llvm-as - | llvm-dis - and then goes
through the original and new output side-by side to insert all
comments at a close-enough location.
Differential Revision: https://reviews.llvm.org/D27765
llvm-svn: 290292
This doesn't implement *every* feature of the existing inliner, but
tries to implement the most important ones for building a functional
optimization pipeline and beginning to sort out bugs, regressions, and
other problems.
Notable, but intentional omissions:
- No alloca merging support. Why? Because it isn't clear we want to do
this at all. Active discussion and investigation is going on to remove
it, so for simplicity I omitted it.
- No support for trying to iterate on "internally" devirtualized calls.
Why? Because it adds what I suspect is inappropriate coupling for
little or no benefit. We will have an outer iteration system that
tracks devirtualization including that from function passes and
iterates already. We should improve that rather than approximate it
here.
- Optimization remarks. Why? Purely to make the patch smaller, no other
reason at all.
The last one I'll probably work on almost immediately. But I wanted to
skip it in the initial patch to try to focus the change as much as
possible as there is already a lot of code moving around and both of
these *could* be skipped without really disrupting the core logic.
A summary of the different things happening here:
1) Adding the usual new PM class and rigging.
2) Fixing minor underlying assumptions in the inline cost analysis or
inline logic that don't generally hold in the new PM world.
3) Adding the core pass logic which is in essence a loop over the calls
in the nodes in the call graph. This is a bit duplicated from the old
inliner, but only a handful of lines could realistically be shared.
(I tried at first, and it really didn't help anything.) All told,
this is only about 100 lines of code, and most of that is the
mechanics of wiring up analyses from the new PM world.
4) Updating the LazyCallGraph (in the new PM) based on the *newly
inlined* calls and references. This is very minimal because we cannot
form cycles.
5) When inlining removes the last use of a function, eagerly nuking the
body of the function so that any "one use remaining" inline cost
heuristics are immediately refined, and queuing these functions to be
completely deleted once inlining is complete and the call graph
updated to reflect that they have become dead.
6) After all the inlining for a particular function, updating the
LazyCallGraph and the CGSCC pass manager to reflect the
function-local simplifications that are done immediately and
internally by the inline utilties. These are the exact same
fundamental set of CG updates done by arbitrary function passes.
7) Adding a bunch of test cases to specifically target CGSCC and other
subtle aspects in the new PM world.
Many thanks to the careful review from Easwaran and Sanjoy and others!
Differential Revision: https://reviews.llvm.org/D24226
llvm-svn: 290161
This patch implements PR31013 by introducing a
DIGlobalVariableExpression that holds a pair of DIGlobalVariable and
DIExpression.
Currently, DIGlobalVariables holds a DIExpression. This is not the
best way to model this:
(1) The DIGlobalVariable should describe the source level variable,
not how to get to its location.
(2) It makes it unsafe/hard to update the expressions when we call
replaceExpression on the DIGLobalVariable.
(3) It makes it impossible to represent a global variable that is in
more than one location (e.g., a variable with multiple
DW_OP_LLVM_fragment-s). We also moved away from attaching the
DIExpression to DILocalVariable for the same reasons.
This reapplies r289902 with additional testcase upgrades and a change
to the Bitcode record for DIGlobalVariable, that makes upgrading the
old format unambiguous also for variables without DIExpressions.
<rdar://problem/29250149>
https://llvm.org/bugs/show_bug.cgi?id=31013
Differential Revision: https://reviews.llvm.org/D26769
llvm-svn: 290153
This reverts commit 289920 (again).
I forgot to implement a Bitcode upgrade for the case where a DIGlobalVariable
has not DIExpression. Unfortunately it is not possible to safely upgrade
these variables without adding a flag to the bitcode record indicating which
version they are.
My plan of record is to roll the planned follow-up patch that adds a
unit: field to DIGlobalVariable into this patch before recomitting.
This way we only need one Bitcode upgrade for both changes (with a
version flag in the bitcode record to safely distinguish the record
formats).
Sorry for the churn!
llvm-svn: 289982
This patch implements PR31013 by introducing a
DIGlobalVariableExpression that holds a pair of DIGlobalVariable and
DIExpression.
Currently, DIGlobalVariables holds a DIExpression. This is not the
best way to model this:
(1) The DIGlobalVariable should describe the source level variable,
not how to get to its location.
(2) It makes it unsafe/hard to update the expressions when we call
replaceExpression on the DIGLobalVariable.
(3) It makes it impossible to represent a global variable that is in
more than one location (e.g., a variable with multiple
DW_OP_LLVM_fragment-s). We also moved away from attaching the
DIExpression to DILocalVariable for the same reasons.
This reapplies r289902 with additional testcase upgrades.
<rdar://problem/29250149>
https://llvm.org/bugs/show_bug.cgi?id=31013
Differential Revision: https://reviews.llvm.org/D26769
llvm-svn: 289920
This patch implements PR31013 by introducing a
DIGlobalVariableExpression that holds a pair of DIGlobalVariable and
DIExpression.
Currently, DIGlobalVariables holds a DIExpression. This is not the
best way to model this:
(1) The DIGlobalVariable should describe the source level variable,
not how to get to its location.
(2) It makes it unsafe/hard to update the expressions when we call
replaceExpression on the DIGLobalVariable.
(3) It makes it impossible to represent a global variable that is in
more than one location (e.g., a variable with multiple
DW_OP_LLVM_fragment-s). We also moved away from attaching the
DIExpression to DILocalVariable for the same reasons.
<rdar://problem/29250149>
https://llvm.org/bugs/show_bug.cgi?id=31013
Differential Revision: https://reviews.llvm.org/D26769
llvm-svn: 289902
so we can stop using DW_OP_bit_piece with the wrong semantics.
The entire back story can be found here:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20161114/405934.html
The gist is that in LLVM we've been misinterpreting DW_OP_bit_piece's
offset field to mean the offset into the source variable rather than
the offset into the location at the top the DWARF expression stack. In
order to be able to fix this in a subsequent patch, this patch
introduces a dedicated DW_OP_LLVM_fragment operation with the
semantics that we used to apply to DW_OP_bit_piece, which is what we
actually need while inside of LLVM. This patch is complete with a
bitcode upgrade for expressions using the old format. It does not yet
fix the DWARF backend to use DW_OP_bit_piece correctly.
Implementation note: We discussed several options for implementing
this, including reserving a dedicated field in DIExpression for the
fragment size and offset, but using an custom operator at the end of
the expression works just fine and is more efficient because we then
only pay for it when we need it.
Differential Revision: https://reviews.llvm.org/D27361
rdar://problem/29335809
llvm-svn: 288683
In r286814, the algorithm for calculating inline costs changed. This
caused more inlining to take place which is especially apparent
in optsize and minsize modes.
As the cost calculation removed a skewed behaviour (we were inconsistent
about the cost of calls) it isn't possible to update the thresholds to
get exactly the same behaviour as before. However, this threshold change
accounts for the very common case where an inline candidate has no
calls within it. In this case, r286814 would inline around 5-6 more (IR)
instructions.
The changes to -Oz have been heavily benchmarked. The "obvious" value
for the inline threshold at -Oz is zero, but due to inaccuracies in the
inline heuristics this can actually cause code size increases due to
not inlining key thunk functions (that then disappear). Experimentally,
5 was the sweet spot for code size over the test-suite.
For -Os, this change removes the outlier results shown up by green dragon
(http://104.154.54.203/db_default/v4/nts/13248).
Fixes D26848.
llvm-svn: 288024
When calculating the cost of a call instruction we were applying a heuristic penalty as well as the cost of the instruction itself.
However, when calculating the benefit from inlining we weren't discounting the equivalent penalty for the call instruction that would be removed! This caused skew in the calculation and meant we wouldn't inline in the following, trivial case:
int g() {
h();
}
int f() {
g();
}
llvm-svn: 286814
The r283656 did this in the remark arguments. We also need to do this
in the main function attribute as that is written to YAML as well.
llvm-svn: 286482
With this we get a new field in the YAML record if the value being
streamed out has a debug location. For examples, please see the changes
to the tests.
This is then used in opt-viewer to display a link for the callee
function in the inlining remarks.
Differential Revision: https://reviews.llvm.org/D26366
llvm-svn: 286169
Value names may be prefixed with a binary '1' to indicate that the
backend should not modify the symbols due to any platform naming
convention.
This should not show up in the YAML opt record file because it breaks
the YAML parser.
llvm-svn: 283656
The purpose of the YAML diagnostic output file is to collect information on
optimizations performed, or not performed, for later processing by tools that
help users (and compiler developers) understand how code was optimized. As
such, the diagnostics that appear in the file should not be coupled to what a
user might want to see summarized for them as the compiler runs, and in fact,
because the user likely does not know what optimization diagnostics their tools
might want to use, the user cannot provide a useful filter regardless. As such,
we shouldn't filter the diagnostics going to the output file.
Differential Revision: https://reviews.llvm.org/D25224
llvm-svn: 283236
There is really no reason for these to be separate.
The vectorizer started this pretty bad tradition that the text of the
missed remarks is pretty meaningless, i.e. vectorization failed. There,
you have to query analysis to get the full picture.
I think we should just explain the reason for missing the optimization
in the missed remark when possible. Analysis remarks should provide
information that the pass gathers regardless whether the optimization is
passing or not.
llvm-svn: 282542
(Re-committed after moving the template specialization under the yaml
namespace. GCC was complaining about this.)
This allows various presentation of this data using an external tool.
This was first recommended here[1].
As an example, consider this module:
1 int foo();
2 int bar();
3
4 int baz() {
5 return foo() + bar();
6 }
The inliner generates these missed-optimization remarks today (the
hotness information is pulled from PGO):
remark: /tmp/s.c:5:10: foo will not be inlined into baz (hotness: 30)
remark: /tmp/s.c:5:18: bar will not be inlined into baz (hotness: 30)
Now with -pass-remarks-output=<yaml-file>, we generate this YAML file:
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 10 }
Function: baz
Hotness: 30
Args:
- Callee: foo
- String: will not be inlined into
- Caller: baz
...
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 18 }
Function: baz
Hotness: 30
Args:
- Callee: bar
- String: will not be inlined into
- Caller: baz
...
This is a summary of the high-level decisions:
* There is a new streaming interface to emit optimization remarks.
E.g. for the inliner remark above:
ORE.emit(DiagnosticInfoOptimizationRemarkMissed(
DEBUG_TYPE, "NotInlined", &I)
<< NV("Callee", Callee) << " will not be inlined into "
<< NV("Caller", CS.getCaller()) << setIsVerbose());
NV stands for named value and allows the YAML client to process a remark
using its name (NotInlined) and the named arguments (Callee and Caller)
without parsing the text of the message.
Subsequent patches will update ORE users to use the new streaming API.
* I am using YAML I/O for writing the YAML file. YAML I/O requires you
to specify reading and writing at once but reading is highly non-trivial
for some of the more complex LLVM types. Since it's not clear that we
(ever) want to use LLVM to parse this YAML file, the code supports and
asserts that we're writing only.
On the other hand, I did experiment that the class hierarchy starting at
DiagnosticInfoOptimizationBase can be mapped back from YAML generated
here (see D24479).
* The YAML stream is stored in the LLVM context.
* In the example, we can probably further specify the IR value used,
i.e. print "Function" rather than "Value".
* As before hotness is computed in the analysis pass instead of
DiganosticInfo. This avoids the layering problem since BFI is in
Analysis while DiagnosticInfo is in IR.
[1] https://reviews.llvm.org/D19678#419445
Differential Revision: https://reviews.llvm.org/D24587
llvm-svn: 282539
This allows various presentation of this data using an external tool.
This was first recommended here[1].
As an example, consider this module:
1 int foo();
2 int bar();
3
4 int baz() {
5 return foo() + bar();
6 }
The inliner generates these missed-optimization remarks today (the
hotness information is pulled from PGO):
remark: /tmp/s.c:5:10: foo will not be inlined into baz (hotness: 30)
remark: /tmp/s.c:5:18: bar will not be inlined into baz (hotness: 30)
Now with -pass-remarks-output=<yaml-file>, we generate this YAML file:
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 10 }
Function: baz
Hotness: 30
Args:
- Callee: foo
- String: will not be inlined into
- Caller: baz
...
--- !Missed
Pass: inline
Name: NotInlined
DebugLoc: { File: /tmp/s.c, Line: 5, Column: 18 }
Function: baz
Hotness: 30
Args:
- Callee: bar
- String: will not be inlined into
- Caller: baz
...
This is a summary of the high-level decisions:
* There is a new streaming interface to emit optimization remarks.
E.g. for the inliner remark above:
ORE.emit(DiagnosticInfoOptimizationRemarkMissed(
DEBUG_TYPE, "NotInlined", &I)
<< NV("Callee", Callee) << " will not be inlined into "
<< NV("Caller", CS.getCaller()) << setIsVerbose());
NV stands for named value and allows the YAML client to process a remark
using its name (NotInlined) and the named arguments (Callee and Caller)
without parsing the text of the message.
Subsequent patches will update ORE users to use the new streaming API.
* I am using YAML I/O for writing the YAML file. YAML I/O requires you
to specify reading and writing at once but reading is highly non-trivial
for some of the more complex LLVM types. Since it's not clear that we
(ever) want to use LLVM to parse this YAML file, the code supports and
asserts that we're writing only.
On the other hand, I did experiment that the class hierarchy starting at
DiagnosticInfoOptimizationBase can be mapped back from YAML generated
here (see D24479).
* The YAML stream is stored in the LLVM context.
* In the example, we can probably further specify the IR value used,
i.e. print "Function" rather than "Value".
* As before hotness is computed in the analysis pass instead of
DiganosticInfo. This avoids the layering problem since BFI is in
Analysis while DiagnosticInfo is in IR.
[1] https://reviews.llvm.org/D19678#419445
Differential Revision: https://reviews.llvm.org/D24587
llvm-svn: 282499
This patch reverses the edge from DIGlobalVariable to GlobalVariable.
This will allow us to more easily preserve debug info metadata when
manipulating global variables.
Fixes PR30362. A program for upgrading test cases is attached to that
bug.
Differential Revision: http://reviews.llvm.org/D20147
llvm-svn: 281284
This would create a bitcast use which fails the verifier: swifterror values may
only be used by loads, stores, and as function arguments.
rdar://28233244
llvm-svn: 281114
Summary:
The inliner may need to determine where a given funclet unwinds to,
and this determination may depend on other funclets throughout the
funclet tree. The code that performs this walk in getUnwindDestToken
memoizes results to avoid redundant computations. In the case that
a funclet's unwind destination is derived from its ancestor, there's
code to walk back down the tree from the ancestor updating the memo
map of its descendants to record the unwind destination. This change
fixes that code to account for the case that some descendant has a
different unwind destination, which can happen if that unwind dest
is a descendant of the EHPad being queried and thus didn't determine
its unwind destination.
Also update test inline-funclets.ll, which is supposed to cover such
scenarios, to include a case that fails an assertion without this fix
but passes with it.
Fixes PR29151.
Reviewers: majnemer
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D24117
llvm-svn: 280610
Summary:
This is obviously an interesting case because it may motivate code
restructuring or LTO.
Reporting this requires instantiation of ORE in the loop where the call
sites are first gathered. I've checked compile-time
overhead *with* -Rpass-with-hotness and the worst slow-down was 6% in
mcf and quickly tailing off. As before without -Rpass-with-hotness
there is no overhead.
Because this could be a pretty noisy diagnostics, it is currently
qualified as 'verbose'. As of this patch, 'verbose' diagnostics are
only emitted with -Rpass-with-hotness, i.e. when the output is expected
to be filtered.
Reviewers: eraman, chandlerc, davidxl, hfinkel
Subscribers: tejohnson, Prazek, davide, llvm-commits
Differential Revision: https://reviews.llvm.org/D23415
llvm-svn: 279860
CGSCC use a WeakVH to track call sites. RAUW a call within a function
can result in that WeakVH getting confused about whether or not the call
site is still around.
llvm-svn: 279268
minimal and boring form than the old pass manager's version.
This pass does the very minimal amount of work necessary to inline
functions declared as always-inline. It doesn't support a wide array of
things that the legacy pass manager did support, but is alse ... about
20 lines of code. So it has that going for it. Notably things this
doesn't support:
- Array alloca merging
- To support the above, bottom-up inlining with careful history
tracking and call graph updates
- DCE of the functions that become dead after this inlining.
- Inlining through call instructions with the always_inline attribute.
Instead, it focuses on inlining functions with that attribute.
The first I've omitted because I'm hoping to just turn it off for the
primary pass manager. If that doesn't pan out, I can add it here but it
will be reasonably expensive to do so.
The second should really be handled by running global-dce after the
inliner. I don't want to re-implement the non-trivial logic necessary to
do comdat-correct DCE of functions. This means the -O0 pipeline will
have to be at least 'always-inline,global-dce', but that seems
reasonable to me. If others are seriously worried about this I'd like to
hear about it and understand why. Again, this is all solveable by
factoring that logic into a utility and calling it here, but I'd like to
wait to do that until there is a clear reason why the existing
pass-based factoring won't work.
The final point is a serious one. I can fairly easily add support for
this, but it seems both costly and a confusing construct for the use
case of the always inliner running at -O0. This attribute can of course
still impact the normal inliner easily (although I find that
a questionable re-use of the same attribute). I've started a discussion
to sort out what semantics we want here and based on that can figure out
if it makes sense ta have this complexity at O0 or not.
One other advantage of this design is that it should be quite a bit
faster due to checking for whether the function is a viable candidate
for inlining exactly once per function instead of doing it for each call
site.
Anyways, hopefully a reasonable starting point for this pass.
Differential Revision: https://reviews.llvm.org/D23299
llvm-svn: 278896
They aren't static, and moving them to the entry block across something
else will only result in tears.
Root cause of http://crbug.com/636558.
llvm-svn: 278571
Summary:
The inliner not being a function pass requires the work-around of
generating the OptimizationRemarkEmitter and in turn BFI on demand.
This will go away after the new PM is ready.
BFI is only computed inside ORE if the user has requested hotness
information for optimization diagnostitics (-pass-remark-with-hotness at
the 'opt' level). Thus there is no additional overhead without the
flag.
Reviewers: hfinkel, davidxl, eraman
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D22694
llvm-svn: 278185
Summary: Hot callsites should have higher threshold than inline hints. This patch uses separate threshold parameter for hot callsites.
Reviewers: davidxl, eraman
Subscribers: Prazek, llvm-commits
Differential Revision: https://reviews.llvm.org/D22368
llvm-svn: 277860
PR28848 had a very nice reduction of the underlying cause of the bug.
Our ValueMap had, in an entry for an Instruction, a ConstantInt.
This is not at all unexpected but should be handled properly.
llvm-svn: 277773
This reverts commit r277611 and the followup r277614.
Bootstrap builds and chromium builds are crashing during inlining after
this change.
llvm-svn: 277642
We were able to figure out that the result of a call is some constant.
While propagating that fact, we added the constant to the value map.
This is problematic because it results in us losing the call site when
processing the value map.
This fixes PR28802.
llvm-svn: 277611
Summary:
copypasta doc of ImportedFunctionsInliningStatistics class
\brief Calculate and dump ThinLTO specific inliner stats.
The main statistics are:
(1) Number of inlined imported functions,
(2) Number of imported functions inlined into importing module (indirect),
(3) Number of non imported functions inlined into importing module
(indirect).
The difference between first and the second is that first stat counts
all performed inlines on imported functions, but the second one only the
functions that have been eventually inlined to a function in the importing
module (by a chain of inlines). Because llvm uses bottom-up inliner, it is
possible to e.g. import function `A`, `B` and then inline `B` to `A`,
and after this `A` might be too big to be inlined into some other function
that calls it. It calculates this statistic by building graph, where
the nodes are functions, and edges are performed inlines and then by marking
the edges starting from not imported function.
If `Verbose` is set to true, then it also dumps statistics
per each inlined function, sorted by the greatest inlines count like
- number of performed inlines
- number of performed inlines to importing module
Reviewers: eraman, tejohnson, mehdi_amini
Subscribers: mehdi_amini, llvm-commits
Differential Revision: https://reviews.llvm.org/D22491
llvm-svn: 277089
The public InlineFunction utility assumes that the passed in
InlineFunctionInfo has a valid AssumptionCacheTracker.
Patch by River Riddle!
Differential Revision: https://reviews.llvm.org/D22706
llvm-svn: 276609
Summary:
For sample-based PGO, using BFI to calculate callsite count is sometime not accurate. This is because with sampling based approach, if a callsite resides in a hot loop deeply nested in a bunch of cold branches, the callsite's BFI frequency would be inaccurately calculated due to lack of samples in the cold branch.
E.g.
if (A1 && A2 && A3 && ..... && A10) {
for (i=0; i < 100000000; i++) {
callsite();
}
}
Assume that A1 to A100 are all 100% taken, and callsite has 1000 samples and thus is considerred hot. Because the loop's trip count is huge, it's normal that all branches outside the loop has no sample at all. As a result, we can only use static branch probability to derive the the frequency of the loop header. Assuming that static heuristic thinks each branch is 50% taken, then the count calculated from BFI will be 1/(2^10) of the actual value.
In order to get more accurate callsite count, we directly annotate the weight on the call instruction, and directly use it when checking callsite hotness.
Note that this mechanism can also be shared by instrumentation based callsite hotness analysis. The side benefit is that it breaks the dependency from Inliner to BFI as call count is embedded in the IR.
Reviewers: davidxl, eraman, dnovillo
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D22118
llvm-svn: 275073
r273711 was reverted by r273743. The inliner needs to know about any
call sites in the inlined function. These were obscured if we replaced
a call to undef with an undef but kept the call around.
This fixes PR28298.
llvm-svn: 273753
Chandler Carruth