Commit Graph

12 Commits

Author SHA1 Message Date
Xinliang David Li 3106d9a122 [PGO] test case cleanups
1. Make test case more focused and robust by focusing on what to be tested (linkage, icall) -- make it easier to validate
2. Testing linkages of data and counter variables instead of names. Counters and data are more relavant to be tested.

llvm-svn: 259067
2016-01-28 18:25:53 +00:00
Xinliang David Li ddbdb1e0be [PGO] make profile prefix even shorter and more readable
llvm-svn: 255587
2015-12-15 00:33:12 +00:00
Xinliang David Li 8608513b69 [PGO] Shorten profile symbol prefixes
(test case update)
Profile symbols have long prefixes which waste space and creating pressure for linker.
This patch shortens the prefixes to minimal length without losing verbosity.

Differential Revision: http://reviews.llvm.org/D15503

llvm-svn: 255576
2015-12-14 23:26:46 +00:00
Justin Bogner e9fe0a298c InstrProf: Make profile variables private to reduce binary size overhead
When we instrument a program for profiling, we copy the linkage of an
instrumented function so that our datastructures merge in the same way
as the function. This avoids redundant copies for things like
linkonce, but ends up emitting names we never need to reference for
normal and internal symbols. Promoting internal and external linkage
to private for these variables reduces the size overhead of profiling
drastically.

llvm-svn: 232799
2015-03-20 06:34:38 +00:00
Justin Bogner f07e34e9d8 InstrProf: Simplify a couple of tests after r230383
Most of the checks in these two tests were actually testing the
behaviour of the instrprof LLVM pass. Now that we're testing that
specifically in LLVM's test suite, it's better if we only test the
frontend's behaviour here.

llvm-svn: 230387
2015-02-24 21:49:28 +00:00
Justin Bogner 970ac60573 InstrProf: Use LLVM's -instrprof pass for profiling
The logic for lowering profiling counters has been moved to an LLVM
pass. Emit the intrinsics rather than duplicating the whole pass in
clang.

llvm-svn: 223683
2014-12-08 19:04:51 +00:00
Duncan P. N. Exon Smith 9121220804 InstrProf: Set profile data to visibility hidden
Shared objects are fairly broken for InstrProf right now -- a follow-up
commit in compiler-rt will fix the rest of this.

The main problem here is that at link time, profile data symbols in the
shared object might get used instead of symbols from the main
executable, creating invalid profile data sections.

<rdar://problem/16918688>

llvm-svn: 208939
2014-05-16 01:24:00 +00:00
Duncan P. N. Exon Smith 4bc7731a29 InstrProf: Calculate a better function hash
The function hash should change when control flow changes.  This patch
hashes the type of each AST node that affects counters, rather than just
counting how many there are.  These types are combined into a small
enumerator that currently has 16 values.

The new hash algorithm packs the enums for consecutively visited types
into a `uint64_t`.  In order to save space for new types, the types are
assumed to be 6-bit values (instead of 4-bit).  In order to minimize
overhead for functions with little control flow, the `uint64_t` is used
directly as a hash if it never fills up; if it does, it's passed through
an MD5 context.

<rdar://problem/16435801>

llvm-svn: 206397
2014-04-16 16:03:27 +00:00
Duncan P. N. Exon Smith d971cd1b18 InstrProf: Emit runtime hook directly in IRGen
-u behaviour is apparently not portable between linkers (see cfe-commits
discussions for r204379 and r205012).  I've moved the logic to IRGen,
where it should have been in the first place.

I don't have a Linux system to test this on, so it's possible this logic
*still* doesn't pull in the instrumented profiling runtime on Linux.

I'm in the process of getting tests going on the compiler-rt side
(llvm-commits "[PATCH] InstrProf: Add initial compiler-rt test").  Once
we have tests for the full flow there, the runtime logic should get a
whole lot less brittle.

<rdar://problem/16458307>

llvm-svn: 205023
2014-03-28 17:53:22 +00:00
Duncan P. N. Exon Smith a7807637bf PGO: Change runtime prefix from pgo to profile
These functions are in the profile runtime.  PGO comes later.

Unfortunately, there's only room for 16 characters in a Darwin section,
so use __llvm_prf_ instead of __llvm_profile_ for section names.

<rdar://problem/15943240>

llvm-svn: 204390
2014-03-20 20:00:41 +00:00
Justin Bogner b4416f58d5 CodeGen: Include a function hash in instrumentation based profiling
The hash itself is still the number of counters, which isn't all that
useful, but this separates the API changes from the actual
implementation of the hash and will make it easier to transition to
the ProfileData library once it's implemented.

llvm-svn: 204186
2014-03-18 21:58:06 +00:00
Duncan P. N. Exon Smith 2fe531cb07 PGO: Statically generate data structures
In instrumentation-based profiling, we need a set of data structures to
represent the counters.  Previously, these were built up during static
initialization.  Now, they're shoved into a specially-named section so
that they show up as an array.

As a consequence of the reorganizing symbols, instrumentation data
structures for linkonce functions are now correctly coalesced.

This is the first step in a larger project to minimize runtime overhead
and dependencies in instrumentation-based profilng.  The larger picture
includes removing all initialization overhead and making the dependency
on libc optional.

<rdar://problem/15943240>

llvm-svn: 204080
2014-03-17 21:18:30 +00:00