| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
DFSan changes the ABI of each function in the module. This makes it possible
for a function with the native ABI to be called with the instrumented ABI,
or vice versa, thus possibly invoking undefined behavior. A simple way
of statically detecting instances of this problem is to prepend the prefix
"dfs$" to the name of each instrumented-ABI function.
This will not catch every such problem; in particular function pointers passed
across the instrumented-native barrier cannot be used on the other side.
These problems could potentially be caught dynamically.
Differential Revision: http://llvm-reviews.chandlerc.com/D1373
llvm-svn: 189052
|
|
|
|
|
|
|
|
| |
This uses the ARMcmov pattern that Tim cleaned up in r188995.
Thanks to Simon Tatham for his floating point help!
llvm-svn: 189024
|
|
|
|
|
|
| |
contain Offset when Offset >= 65536.
llvm-svn: 189021
|
|
|
|
| |
llvm-svn: 189020
|
|
|
|
|
|
|
|
|
|
|
| |
The instruction to convert between floating point and fixed point representations
takes an immediate operand for the number of fractional bits of the fixed point
value. ARMARM specifies that when that number of bits is zero, the assembler
should encode floating point/integer conversion instructions.
This patch adds the necessary instruction aliases to achieve this behaviour.
llvm-svn: 189009
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
using GEPs. Previously, it used a number of different heuristics for
analyzing the GEPs. Several of these were conservatively correct, but
failed to fall back to SCEV even when SCEV might have given a reasonable
answer. One was simply incorrect in how it was formulated.
There was good code already to recursively evaluate the constant offsets
in GEPs, look through pointer casts, etc. I gathered this into a form
code like the SLP code can use in a previous commit, which allows all of
this code to become quite simple.
There is some performance (compile time) concern here at first glance as
we're directly attempting to walk both pointers constant GEP chains.
However, a couple of thoughts:
1) The very common cases where there is a dynamic pointer, and a second
pointer at a constant offset (usually a stride) from it, this code
will actually not do any unnecessary work.
2) InstCombine and other passes work very hard to collapse constant
GEPs, so it will be rare that we iterate here for a long time.
That said, if there remain performance problems here, there are some
obvious things that can improve the situation immensely. Doing
a vectorizer-pass-wide memoizer for each individual layer of pointer
values, their base values, and the constant offset is likely to be able
to completely remove redundant work and strictly limit the scaling of
the work to scrape these GEPs. Since this optimization was not done on
the prior version (which would still benefit from it), I've not done it
here. But if folks have benchmarks that slow down it should be straight
forward for them to add.
I've added a test case, but I'm not really confident of the amount of
testing done for different access patterns, strides, and pointer
manipulation.
llvm-svn: 189007
|
|
|
|
|
|
| |
we pass these tests with -verify-machineinstrs.
llvm-svn: 189006
|
|
|
|
| |
llvm-svn: 189005
|
|
|
|
|
|
|
| |
The function call to external function should come with PLT relocation
type if the PIC relocation model is used.
llvm-svn: 189002
|
|
|
|
|
|
|
|
|
|
|
| |
pointers, but accumulate the offset into an APInt in the process of
stripping it.
This is a pretty handy thing to have, such as when trying to determine
if two pointers are at some constant relative offset. I'll be committing
a patch shortly to use it for exactly that purpose.
llvm-svn: 189000
|
|
|
|
|
|
|
|
| |
Follow-up to r188903.
The AllocationGranularity can be 65536 on Win32, even on Cygwin.
llvm-svn: 188998
|
|
|
|
|
|
|
|
|
|
|
|
| |
Back in the mists of time (2008), it seems TableGen couldn't handle the
patterns necessary to match ARM's CMOV node that we convert select operations
to, so we wrote a lot of fairly hairy C++ to do it for us.
TableGen can deal with it now: there were a few minor differences to CodeGen
(see tests), but nothing obviously worse that I could see, so we should
probably address anything that *does* come up in a localised manner.
llvm-svn: 188995
|
|
|
|
|
|
|
| |
The code for 'Q' and 'R' operand modifiers needs to look through tied
operands to discover the register class.
llvm-svn: 188990
|
|
|
|
|
|
|
|
| |
parentmbb into a successmbb, include any DBG_VALUE MI.
Fix for PR16954.
llvm-svn: 188987
|
|
|
|
| |
llvm-svn: 188980
|
|
|
|
|
|
|
|
|
| |
Indirect tail-calls shouldn't use R9 for the branch destination, as
it's not reliably a call-clobbered register.
rdar://14793425
llvm-svn: 188967
|
|
|
|
| |
llvm-svn: 188957
|
|
|
|
| |
llvm-svn: 188956
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When truncated vector stores were being custom lowered in
VectorLegalizer::LegalizeOp(), the old (illegal) and new (legal) node pair
was not being added to LegalizedNodes list. Instead of the legalized
result being passed to VectorLegalizer::TranslateLegalizeResult(),
the result was being passed back into VectorLegalizer::LegalizeOp(),
which ended up adding a (new, new) pair to the list instead.
This was causing an assertion failure when a custom lowered truncated
vector store was the last instruction a basic block and the VectorLegalizer
was unable to find it in the LegalizedNodes list when updating the
DAG root.
llvm-svn: 188953
|
|
|
|
|
|
| |
Spotted by Bill Wendling.
llvm-svn: 188942
|
|
|
|
|
|
| |
Replace "(255 & value)" with "(0xFF & value)" to improve clarity.
llvm-svn: 188941
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The small utility function that pattern matches Base + Index +
Offset patterns for loads and stores fails to recognize the base
pointer for loads/stores from/into an array at offset 0 inside a
loop. As a result DAGCombiner::MergeConsecutiveStores was not able
to merge all stores.
This commit fixes the issue by adding an additional pattern match
and also a test case.
Reviewer: Nadav
llvm-svn: 188936
|
|
|
|
| |
llvm-svn: 188932
|
|
|
|
|
|
| |
Also obsessively reorder the headers to be in something closer to alphabetical order.
llvm-svn: 188928
|
|
|
|
| |
llvm-svn: 188926
|
|
|
|
| |
llvm-svn: 188925
|
|
|
|
| |
llvm-svn: 188924
|
|
|
|
| |
llvm-svn: 188916
|
|
|
|
| |
llvm-svn: 188915
|
|
|
|
|
|
|
| |
def imm0_63 : Operand<i32>, ImmLeaf<i32, [{ return Imm >= 0 && Imm < 63;}]>{
As it seems Imm <63 should be Imm <= 63. ImmLeaf is used in pattern match, but there is already a function check the shift amount range, so just remove ImmLeaf. Also add a test to check 63.
llvm-svn: 188911
|
|
|
|
|
|
|
| |
Offset in mmap(3) should be aligned to gepagesize(), 64k, or mmap(3) would fail.
TODO: Invetigate places where 4096 would be required as pagesize, or 4096 would satisfy.
llvm-svn: 188903
|
|
|
|
|
|
|
| |
According to the ARM specification, "mov" is a valid mnemonic for all Thumb2 MOV encodings.
To achieve this, the patch adds one instruction alias with a special range condition to avoid collision with the Thumb1 MOV.
llvm-svn: 188901
|
|
|
|
| |
llvm-svn: 188899
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The initial port used MLG(R) for i64 UMUL_LOHI but left the other three
combinations as not-legal-or-custom. Although 32x32->{32,32}
multiplications exist, they're not as quick as doing a normal 64-bit
multiplication, so it didn't seem like i32 SMUL_LOHI and UMUL_LOHI
would be useful. There's also no direct instruction for i64 SMUL_LOHI,
so it needs to be implemented in terms of UMUL_LOHI.
However, not defining these patterns means that we don't convert
division by a constant into multiplication, so this patch fills
in the other cases. The new i64 SMUL_LOHI sequence is simpler
than the one that we used previously for 64x64->128 multiplication,
so int-mul-08.ll now tests the full sequence.
llvm-svn: 188898
|
|
|
|
|
|
|
|
|
|
| |
I accidentally changed the encoding of the MSA registers to zero instead of 0
to 31. This change restores the encoding the registers had prior to r188893.
This didn't show up in the existing tests because direct-object emission isn't
implemented yet for MSA.
llvm-svn: 188896
|
|
|
|
| |
llvm-svn: 188895
|
|
|
|
|
|
|
| |
These are extensions of the existing FI[EDX]BR instructions, but use a spare
bit to suppress inexact conditions.
llvm-svn: 188894
|
|
|
|
|
|
| |
No functional change
llvm-svn: 188893
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Like yaml ObjectFiles, this will be very useful for testing the MC CFG
implementation (mostly MCObjectDisassembler), by matching the output
with YAML, and for potential users of the MC CFG, by using it as an input.
There isn't much to the actual format, it is just a serialization of the
MCModule class. Of note:
- Basic block references (pred/succ, ..) are represented by the BB's
start address.
- Just as in the MC CFG, instructions are MCInsts with a size.
- Operands have a prefix representing the type (only register and
immediate supported here).
- Instruction opcodes are represented by their names; enum values aren't
stable, enum names mostly are: usually, a change to a name would need
lots of changes in the backend anyway.
Same with registers.
All in all, an example is better than 1000 words, here goes:
A simple binary:
Disassembly of section __TEXT,__text:
_main:
100000f9c: 48 8b 46 08 movq 8(%rsi), %rax
100000fa0: 0f be 00 movsbl (%rax), %eax
100000fa3: 3b 04 25 48 00 00 00 cmpl 72, %eax
100000faa: 0f 8c 07 00 00 00 jl 7 <.Lend>
100000fb0: 2b 04 25 48 00 00 00 subl 72, %eax
.Lend:
100000fb7: c3 ret
And the (pretty verbose) generated YAML:
---
Atoms:
- StartAddress: 0x0000000100000F9C
Size: 20
Type: Text
Content:
- Inst: MOV64rm
Size: 4
Ops: [ RRAX, RRSI, I1, R, I8, R ]
- Inst: MOVSX32rm8
Size: 3
Ops: [ REAX, RRAX, I1, R, I0, R ]
- Inst: CMP32rm
Size: 7
Ops: [ REAX, R, I1, R, I72, R ]
- Inst: JL_4
Size: 6
Ops: [ I7 ]
- StartAddress: 0x0000000100000FB0
Size: 7
Type: Text
Content:
- Inst: SUB32rm
Size: 7
Ops: [ REAX, REAX, R, I1, R, I72, R ]
- StartAddress: 0x0000000100000FB7
Size: 1
Type: Text
Content:
- Inst: RET
Size: 1
Ops: [ ]
Functions:
- Name: __text
BasicBlocks:
- Address: 0x0000000100000F9C
Preds: [ ]
Succs: [ 0x0000000100000FB7, 0x0000000100000FB0 ]
<snip>
...
llvm-svn: 188890
|
|
|
|
| |
llvm-svn: 188889
|
|
|
|
| |
llvm-svn: 188888
|
|
|
|
|
|
| |
Used to detect calls to function symbol stubs (future commit).
llvm-svn: 188887
|
|
|
|
|
|
|
|
|
|
| |
Supports:
- entrypoint, using LC_MAIN.
- static ctors/dtors, using __mod_{init,exit}_func
- translation between effective and object load address, using
dyld's VM address slide.
llvm-svn: 188886
|
|
|
|
|
|
|
|
| |
It can now disassemble code in situations where the effective load
address is different than the load address declared in the object file.
This happens for PIC, hence "dynamic".
llvm-svn: 188884
|
|
|
|
|
|
|
|
|
| |
This is the behavior of sequential disassemblers (llvm-objdump, ...),
when there is no instruction size hint (fixed-length, ...)
While there, also do some minor cleanup.
llvm-svn: 188883
|
|
|
|
|
|
| |
For now, this isn't implemented for any format.
llvm-svn: 188882
|
|
|
|
|
|
|
|
| |
When an MCTextAtom is split, all MCBasicBlocks backed by it are
automatically split, with a fallthrough between both blocks, and
the successors moved to the second block.
llvm-svn: 188881
|
|
|
|
|
|
| |
While there, do some minor cleanup.
llvm-svn: 188880
|
|
|
|
|
|
|
| |
Only implemented in the Mach-O ObjectSymbolizer.
The testcase sadly introduces a new binary.
llvm-svn: 188879
|
|
|
|
| |
llvm-svn: 188878
|