phire / llvm Goto Github PK

Videocore Backend for llvm

License: Other

Shell 1.30% C 10.28% OCaml 0.89% Python 0.77% CSS 0.05% C++ 66.07% Objective-C 6.53% Assembly 13.95% Perl 0.09% Emacs Lisp 0.03% Racket 0.01% Vim Script 0.03%

llvm's Introduction

Low Level Virtual Machine (LLVM)
================================

This directory and its subdirectories contain source code for the Low Level
Virtual Machine, a toolkit for the construction of highly optimized compilers,
optimizers, and runtime environments.

LLVM is open source software. You may freely distribute it under the terms of
the license agreement found in LICENSE.txt.

Please see the documentation provided in docs/ for further
assistance with LLVM, and in particular docs/GettingStarted.rst for getting
started with LLVM and docs/README.txt for an overview of LLVM's
documentation setup.

If you're writing a package for LLVM, see docs/Packaging.rst for our
suggestions.

llvm's People

Contributors

Stargazers

Watchers

Forkers

ryarnyah twilmer uplinkcoder mfzhang jeanvaljean2

llvm's Issues

Assembly/Disassembly of push/pop instructions.

Assembly/Disassembly of floating point instructions.

48bit vector operand decoding is incomplete

Operands don't show scalar register offsets and will potentially show the wrong orientation.

Load/stores with a displacement crash

phire@nirrit:~/projects/llvm/build$ echo "st r0, (r3+sp)" | bin/llvm-mc -arch=videocore  -show-encoding
    .text
llvm-mc: /home/phire/projects/llvm/llvm32/lib/Target/Videocore/MCTargetDesc/VideocoreMCCodeEmitter.cpp:134: unsigned int {anonymous}::VideocoreMCCodeEmitter::getMachineOpValue(const llvm::MCInst&, const llvm::MCOperand&, llvm::SmallVectorImpl<llvm::MCFixup>&) const: Assertion `Kind == MCExpr::SymbolRef' failed.
0  llvm-mc         0x000000000060c21d
1  llvm-mc         0x000000000060c49a
2  llvm-mc         0x000000000060bf13
3  libpthread.so.0 0x00007fd48e807870
4  libc.so.6       0x00007fd48da50369 gsignal + 57
5  libc.so.6       0x00007fd48da51768 abort + 328
6  libc.so.6       0x00007fd48da49456
7  libc.so.6       0x00007fd48da49502
8  llvm-mc         0x00000000004cb8a2
9  llvm-mc         0x00000000004cbe15
10 llvm-mc         0x00000000004cb661
11 llvm-mc         0x000000000055ebf4
12 llvm-mc         0x000000000055f77d
13 llvm-mc         0x0000000000421633
14 llvm-mc         0x00000000005129f7
15 llvm-mc         0x000000000050ea91
16 llvm-mc         0x000000000040c455
17 llvm-mc         0x000000000040d138
18 libc.so.6       0x00007fd48da3cb05 __libc_start_main + 245
19 llvm-mc         0x000000000040b6a9
Stack dump:
0.  Program arguments: bin/llvm-mc -arch=videocore -show-encoding 
Aborted

Clean up .td files

It might be possible to remove duplication in the instruction formats and multiclasses by using multiple inheritance.

Scalar Register Offsets on 48bit Vector Instructions

There is only a single scalar offset field for 48bit instructions.

The destination vector uses the z flag to specify if the scalar offset field is used.

The sources vector(s) are forced to have the same orientation as the destination and the orientation flag is used to specify if the scalar offset field is used.