Added support for unpredictable mcrr/mcrr2/mrrc/mrrc2 ARM instruction in the disassembler. Since the upredicability conditions are complex, C++ code was added to handle them.
Benjamin Kramer [Wed, 18 Apr 2012 10:37:32 +0000 (10:37 +0000)]
SmallPtrSet: Reuse DenseMapInfo's pointer hash function instead of inventing a bad one ourselves.
DenseMap's hash function uses slightly more entropy and reduces hash collisions
significantly. I also experimented with Hashing.h, but it didn't gave a lot of
improvement while being much more expensive to compute.
Bill Wendling [Wed, 18 Apr 2012 06:00:09 +0000 (06:00 +0000)]
Use a heavy hammer to fix PR12573.
If the loop contains invoke instructions, whose unwind edge escapes the loop,
then don't try to unswitch the loop. Doing so may cause the unwind edge to be
split, which not only is non-trivial but doesn't preserve loop simplify
information.
Andrew Trick [Wed, 18 Apr 2012 04:00:10 +0000 (04:00 +0000)]
loop-reduce: Add an early bailout to catch extremely large loops.
This introduces a threshold of 200 IV Users, which is very
conservative but should be sufficient to avoid serious compile time
sink or stack overflow. The llvm test-suite with LTO never exceeds 190
users per loop.
The bug doesn't relate to a specific type of loop. Checking in an
arbitrary giant loop as a unit test would be silly.
Joe Groff [Tue, 17 Apr 2012 23:05:54 +0000 (23:05 +0000)]
fix pr12559: mark unavailable win32 math libcalls
also fix SimplifyLibCalls to use TLI rather than compile-time conditionals to enable optimizations on floor, ceil, round, rint, and nearbyint
Joel Jones [Tue, 17 Apr 2012 22:23:10 +0000 (22:23 +0000)]
Fixes a problem in instruction selection with testing whether or not the
transformation:
(X op C1) ^ C2 --> (X op C1) & ~C2 iff (C1&C2) == C2
should be done.
This change has been tested:
Using a debug+asserts build:
on the specific test case that brought this bug to light
make check-all
lnt nt
using this clang to build a release version of clang
Using the release+asserts clang-with-clang build:
on the specific test case that brought this bug to light
make check-all
lnt nt
Checking in because Evan wants it checked in. Test case forthcoming after
scrubbing.
Lang Hames [Tue, 17 Apr 2012 04:15:51 +0000 (04:15 +0000)]
SlotIndexes used to store the index list in a crufty custom linked-list. I can't
for the life of me remember why I wrote it this way, but I can't see any good
reason for it now. This patch replaces the custom linked list with an ilist.
This change should preserve the existing numberings exactly, so no generated code
should change (if it does, file a bug!).
Kevin Enderby [Tue, 17 Apr 2012 00:49:27 +0000 (00:49 +0000)]
Fix ARM disassembly of VLD2 (single 2-element structure to all lanes)
instructions with writebacks. And add test a case for all opcodes handed by
DecodeVLD2DupInstruction() in ARMDisassembler.cpp .
Implement GDB integration for source level debugging of code JITed using
the MCJIT execution engine.
The GDB JIT debugging integration support works by registering a loaded
object image with a pre-defined function that GDB will monitor if GDB
is attached. GDB integration support is implemented for ELF only at this
time. This integration requires GDB version 7.0 or newer.
Fix updateTerminator to be resiliant to degenerate terminators where
both fallthrough and a conditional branch target the same successor.
Gracefully delete the conditional branch and introduce any unconditional
branch needed to reach the actual successor. This fixes memory
corruption in 2009-06-15-RegScavengerAssert.ll and possibly other tests.
Also, while I'm here fix a latent bug I spotted by inspection. I never
applied the same fundamental fix to this fallthrough successor finding
logic that I did to the logic used when there are no conditional
branches. As a consequence it would have selected landing pads had they
be aligned in just the right way here. I don't have a test case as
I spotted this by inspection, and the previous time I found this
required have of TableGen's source code to produce it. =/ I hate backend
bugs. ;]
Thanks to Jim Grosbach for helping me reason through this and reviewing
the fix.
1. CHECKNEXT was used instead of CHECK-NEXT which caused the line to be
ignored which in turn hid the next 2 problems:
2. ('sh_offset', 0x{{{[0-9,a-f]+}}) had one too many leading curly braces and
failed to do it's job of accepting all hex digits and:
3. The check for the hex values for the code instructions didn't account for
blank separators.
Make it possible to indicate relaxed floating point requirements at the IR level
through the use of 'fpmath' metadata. Currently this only provides a 'fpaccuracy'
value, which may be a number in ULPs or the keyword 'fast', however the intent is
that this will be extended with additional information about NaN's, infinities
etc later. No optimizations have been hooked up to this so far.
Flip the new block-placement pass to be on by default.
This is mostly to test the waters. I'd like to get results from FNT
build bots and other bots running on non-x86 platforms.
This feature has been pretty heavily tested over the last few months by
me, and it fixes several of the execution time regressions caused by the
inlining work by preventing inlining decisions from radically impacting
block layout.
I've seen very large improvements in yacr2 and ackermann benchmarks,
along with the expected noise across all of the benchmark suite whenever
code layout changes. I've analyzed all of the regressions and fixed
them, or found them to be impossible to fix. See my email to llvmdev for
more details.
I'd like for this to be in 3.1 as it complements the inliner changes,
but if any failures are showing up or anyone has concerns, it is just
a flag flip and so can be easily turned off.
I'm switching it on tonight to try and get at least one run through
various folks' performance suites in case SPEC or something else has
serious issues with it. I'll watch bots and revert if anything shows up.
Remove an overly brittle test. This test will no longer be interesting
once we start changing the block layout, so just nuke it. If anyone has
ideas about how to craft a code layout agnostic form of the test please
let me know.
Add a somewhat hacky heuristic to do something different from whole-loop
rotation. When there is a loop backedge which is an unconditional
branch, we will end up with a branch somewhere no matter what. Try
placing this backedge in a fallthrough position above the loop header as
that will definitely remove at least one branch from the loop iteration,
where whole loop rotation may not.
I haven't seen any benchmarks where this is important but loop-blocks.ll
tests for it, and so this will be covered when I flip the default.
Add convenience methods to MDBuilder for attaching metadata to instructions,
and retrieving it from instructions. I don't have a use for this but is seems
logical for it to exist. While there, remove some 'const' markings from methods
which are in fact 'const' in practice, but aren't logically 'const'.
Richard Barton [Mon, 16 Apr 2012 11:32:10 +0000 (11:32 +0000)]
Add -disassemble support for -show-inst and -show-encode capability llvm-mc. Also refactor so all MC paraphernalia are created once for all uses as much as possible.
The test change is to account for the fact that the default disassembler behaviour has changed with regards to specifying the assembly syntax to use.
Tweak the loop rotation logic to check whether the loop is naturally
laid out in a form with a fallthrough into the header and a fallthrough
out of the bottom. In that case, leave the loop alone because any
rotation will introduce unnecessary branches. If either side looks like
it will require an explicit branch, then the rotation won't add any, do
it to ensure the branch occurs outside of the loop (if possible) and
maximize the benefit of the fallthrough in the bottom.
Change type profile for vpermv back to using operand type for the mask argument to match intrinsic behavior. Add a bitcast to the lowering code to convert mask from v8i32 to v8f32 for vpermps.
Flip the arguments when converting vpermd/vpermps intrinsics into instructions. The intrinsic has the mask as the last operand, but the instruction has it as the second.
projects / llvm / log