Previously we reliedy on pseudo probe descriptors to look up precomputed GUID during probe emission for inlined probes. Since we are moving to always using unique linkage names, GUID for functions can be computed in place from dwarf names. This eliminates the need of importing pseudo probe descs in thinlto, since those descs should be emitted by the original modules.
This significantly reduces thinlto memory footprint in some extreme case where the number of imported modules for a single module is massive.
Test Plan:
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D105248
The linker or post-link optimizer can create an ELF image with multiple executable segments each of which will be loaded separately at run time. This breaks the assumption of llvm-profgen that currently only supports one base load address. What it ends up with is that the subsequent mmap events will be treated as an overwrite of the first mmap event which will in turn screw up address mapping. While it is non-trivial to support multiple separate load addresses and given that on x64 those segments will always be loaded at consecutive addresses (though via separate mmap
sys calls), I'm adding an error checking logic to bail out if that's violated and keep using a single load address which is the address of the first executable segment.
Also changing the disassembly output from printing section offset to printing the virtual address instead, which matches the behavior of objdump.
Differential Revision: https://reviews.llvm.org/D103178
Allow
```
%x:_<2 x p0> = G_INTTOPTR %y:_<2 x s64>
```
This shows up when building clang for AArch64 with GlobalISel.
Also show that we can select it.
This should match SDAG's behaviour: https://godbolt.org/z/33oqYoaYv
Differential Revision: https://reviews.llvm.org/D105944
Currently we allocate MemoryMapper per size class.
MemoryMapper mmap's and munmap's internal buffer.
This results in 50 mmap/munmap calls under the global
allocator mutex. Reuse MemoryMapper and the buffer
for all size classes. This radically reduces number of
mmap/munmap calls. Smaller size classes tend to have
more objects allocated, so it's highly likely that
the buffer allocated for the first size class will
be enough for all subsequent size classes.
Reviewed By: melver
Differential Revision: https://reviews.llvm.org/D105778
To help with debugging non-trivial unswitching issues.
Don't care about the legacy pass, nobody is using it.
If a pass's string params are empty (e.g. "simple-loop-unswitch"), don't
default to the empty constructor for the pass params. We should still
let the parser take care of it in case the parser has its own defaults.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D105933
AMDGPU normally spills SGPRs to VGPRs. Previously, since all register
classes are handled at the same time, this was problematic. We don't
know ahead of time how many registers will be needed to be reserved to
handle the spilling. If no VGPRs were left for spilling, we would have
to try to spill to memory. If the spilled SGPRs were required for exec
mask manipulation, it is highly problematic because the lanes active
at the point of spill are not necessarily the same as at the restore
point.
Avoid this problem by fully allocating SGPRs in a separate regalloc
run from VGPRs. This way we know the exact number of VGPRs needed, and
can reserve them for a second run. This fixes the most serious
issues, but it is still possible using inline asm to make all VGPRs
unavailable. Start erroring in the case where we ever would require
memory for an SGPR spill.
This is implemented by giving each regalloc pass a callback which
reports if a register class should be handled or not. A few passes
need some small changes to deal with leftover virtual registers.
In the AMDGPU implementation, a new pass is introduced to take the
place of PrologEpilogInserter for SGPR spills emitted during the first
run.
One disadvantage of this is currently StackSlotColoring is no longer
used for SGPR spills. It would need to be run again, which will
require more work.
Error if the standard -regalloc option is used. Introduce new separate
-sgpr-regalloc and -vgpr-regalloc flags, so the two runs can be
controlled individually. PBQB is not currently supported, so this also
prevents using the unhandled allocator.
The documentation on these was out of sync with the implementation. Also
the declaration of inputs was repeated when it is already part of the
ArithmeticCastOp definition.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D105934
This patch is in a series of patches to provide builtins for compatibility
with the XL compiler. This patch adds the builtins and instrisics for compare
and multiply related operations.
Reviewed By: nemanjai, #powerpc
Differential revision: https://reviews.llvm.org/D102875
Adds zero-preserving unary operators from std. Also adds xor.
Performs minor refactoring to remove "zero" node, and pushed
the irregular logic for negi (not support in std) into one place.
Reviewed By: gussmith23
Differential Revision: https://reviews.llvm.org/D105928
This is a follow up to https://reviews.llvm.org/D104080, and ca3bdb57fa (diff-e64a48fabe31db213a631fdc5f2acb51bdddf3f16a8fb2928784f4c579229585). The implementation of call graph profile was changed from a black box section to relocation approach. This was done to be compatible with post processing tools like strip/objcopy, and llvm equivalent. When they are invoked on object file before the final linking step with this new approach the symbol indices correctness is preserved.
The GNU binutils tools change the REL section to RELA section, unlike llvm tools. For example when strip -S is run on the ELF object files, as an intermediate step before linking. To preserve compatibility this patch extends implementation in LLD and ELFDumper to support both REL and RELA sections for call graph profile.
Reviewed By: MaskRay, jhenderson
Differential Revision: https://reviews.llvm.org/D105217
Redefined FPBits.h and LongDoubleBitsX86 so its implementation works for the Windows
and Linux platform while maintaining a packed memory alignment of the precision floating
point numbers. For its size in memory to be the same as the data type of the float point number.
This change was necessary because the previous attribute((packed)) specification in the struct was not working
for Windows like it was for Linux and consequently static_asserts in the FPBits.h file were failing.
Reviewed By: aeubanks, sivachandra
Differential Revision: https://reviews.llvm.org/D105561
The current compile options function hardcodes the -fpie and
-ffreestanding flags, which don't exist on Windows. This patch sets the
compilation flags conditionally based on the OS specifics.
Reviewed By: sivachandra, aeubanks
Differential Revision: https://reviews.llvm.org/D105643
This is split from D105216, but the code is hoisted much earlier into the path where we can actually get a zero stride flowing through. Some fairly simple proofs handle the cases which show up in practice. The only test changes are the cases where we really do need a non-zero divider to produce the right result.
Differential Revision: https://reviews.llvm.org/D105921
Defined constant that express the number of bits for exponent in single and double precision. Added bit masks values and other properties for quad precision floating point numbers that specifically targets architectures defined in PlatfromDefs.h. The exponentWidth values were added to be used in LongDoubleBitsX86.h where the implementation to set the exponent component uses this and the bitWidth value. The need occurred because of the 80-bit quad precision implementation.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D105153
This patch is a followup patch to https://reviews.llvm.org/D105760 which adds this relocation. This handles the relocation in lld.
The s_branch family of instruction does the following:
PC = PC + signext(simm * 4) + 4
so we we do the opposite on the target address before writing it in the instruction stream.
Reviewed By: MaskRay
Differential Revision: https://reviews.llvm.org/D105761
Make sure getMinusSCEV() didn't return a pointer. The following check
would never succeed if it was a pointer, anyway, but calling
getMulExpr() on a pointer SCEV now asserts.
Split off from D105216 to simplify review. Rewritten with a lambda to be easier to follow. Comments clarified.
Sorry for no test case, this is tricky to exercise with the current structure of the code. It's about to be hit more frequently in a follow up patch, and the change itself is simple.
Add tests for D105088, as well as an option to disable the
(generally) unsound inttoptr of ptrtoint optimization.
Differential Revision: https://reviews.llvm.org/D105771
This will make it easier to update these tests as we add support for generating
more SIMD loads and stores with custom alignments.
Differential Revision: https://reviews.llvm.org/D105862
[NFC] This patch adds features for pwr7, pwr8, and pwr9 that can be
used for semachecking builtin functions that are only valid for certain
versions of ppc.
Reviewed By: nemanjai, #powerpc
Authored By: Quinn Pham <Quinn.Pham@ibm.com>
Differential revision: https://reviews.llvm.org/D105501
Hongtao Yu