about summary refs log tree commit diff stats
Commit message (Collapse)AuthorAgeFilesLines
* Fix ExprInt repr to include sizeAjax2017-01-131-1/+2
|
* Merge pull request #472 from commial/feature/faster-parityserpilliere2017-01-123-12/+16
|\ | | | | Feature/faster parity
| * CC: declare parity as a macro to avoid function callsAjax2017-01-092-7/+1
| |
| * LLVM: use llvm.ctpop for 'parity' operationAjax2017-01-091-5/+15
|/
* Merge pull request #467 from serpilliere/fix_dg_emulCamille Mougey2017-01-095-18/+43
|\ | | | | Fix dg emul
| * DepGraph: Add emul regression testsFabrice Desclaux2017-01-094-0/+13
| |
| * DepGraph: fix emulationFabrice Desclaux2017-01-091-18/+30
|/
* Merge pull request #469 from serpilliere/updt_call_effects_apiCamille Mougey2017-01-067-62/+89
|\ | | | | Updt call effects api
| * x86: IRA don't stack retaddr in default call modelFabrice Desclaux2017-01-061-31/+8
| |
| * IR: Split addbloc codeFabrice Desclaux2017-01-061-23/+68
| |
| * IR: Call_effects API modificationFabrice Desclaux2017-01-066-9/+14
| | | | | | | | | | | | | | | | | | | | | | Old API: def call_effects(self, addr): New API: def call_effects(self, addr, instr): The addr is the address of the called function 'instr' is the instruction responsible for the call. The new API is a bit more flexible for model a function.
* | Merge pull request #471 from commial/feature/llvm-cacheserpilliere2017-01-0610-62/+282
|\ \ | | | | | | Feature/llvm cache
| * | Test/jitter for each jitter engineAjax2017-01-064-11/+9
| | |
| * | Add a test for jit_maxline optionAjax2017-01-061-0/+22
| | |
| * | Add a test for max_exec_per_call optionAjax2017-01-062-0/+82
| | |
| * | Remove useless jit_call wrapperAjax2017-01-052-14/+5
| | |
| * | Jitter: avoid getting all gpregs, just for one valueAjax2017-01-051-1/+1
| | |
| * | Jitter: remove useless VmMngr argumentAjax2017-01-053-11/+10
| | |
| * | LLVM: handle max_exec_per_call optionAjax2017-01-052-11/+21
| | |
| * | LLVM: get rid of vm_mngr argumentAjax2017-01-052-4/+6
| | |
| * | LLVM: enable final object cachingAjax2017-01-052-22/+111
| | |
| * | LLVM: avoid going back to Python while next block are already jittedAjax2017-01-052-7/+34
| | |
* | | Merge pull request #470 from a-vincent/masterserpilliere2017-01-0674-93/+49
|\ \ \ | |/ / |/| | Tidy up which files are executable and which are not
| * | make this one executable, it's probably the most used toplevel executableAymeric Vincent2017-01-051-0/+2
| | |
| * | these files are not top level executables, adaptAymeric Vincent2017-01-053-24/+0
| | |
| * | use python2 as executable name, give adequate permissionsAymeric Vincent2017-01-0537-37/+47
| | |
| * | remove #! command line on files not supposed to be run at top levelAymeric Vincent2017-01-0532-32/+0
| | |
| * | fix permissionsAymeric Vincent2017-01-051-0/+0
|/ /
* | Merge pull request #468 from commial/feature/llvm-floatserpilliere2017-01-054-5/+43
|\ \ | |/ |/| Feature/llvm float
| * x86 BT*: include shift inside addr computation (segm case)Ajax2017-01-051-1/+8
| |
| * x86 sem: fix a type mismatchAjax2017-01-051-1/+3
| |
| * LLVM: add support for cpuid operationAjax2017-01-051-0/+4
| |
| * Enable mn_float unit test for LLVMAjax2017-01-051-1/+1
| |
| * Promote ExprId to float if they belong to archiitecture float_listAjax2017-01-051-2/+6
| | | | | | | | There aren't distinguishable based on the Expr instance
| * LLVM: handle basic float operationsAjax2017-01-051-0/+18
| |
| * Add missing x86 offset for floatAjax2017-01-051-0/+3
|/
* Merge pull request #466 from serpilliere/fix_empty_dstCamille Mougey2017-01-051-2/+7
|\ | | | | IR: fix set_empty_dst using c_next
| * IR: fix set_empty_dst using c_nextFabrice Desclaux2017-01-051-2/+7
| |
* | Merge pull request #458 from commial/feature-llvm-jitterserpilliere2017-01-0510-622/+897
|\ \ | |/ |/| Feature llvm jitter
| * Simplify LLVM translation of ExprCondAjax2017-01-041-34/+2
| |
| * Force the undefined behavior of shifts to be 0Ajax2017-01-041-6/+20
| | | | | | | | | | This is an arbitrary convention describing the behavior of '>>', '<<' and 'a>>' Miasm IR operator (contrary to the usual ones)
| * Update Travis for Numba llvmliteAjax2017-01-041-0/+11
| |
| * Remove unused code, regroup similar function by categoryAjax2017-01-041-398/+169
| |
| * Clean-up LLVM ModuleRef manipulationAjax2017-01-042-44/+44
| |
| * LLVM: add support for optimisationAjax2017-01-041-28/+14
| |
| * LLVM: Handle '-' operation with 2 operandsAjax2017-01-041-2/+3
| | | | | | | | | | At the time of writting, x86 AAS instruction, for instance, generate this kind of Expression
| * LLVM: Handle asm_bad_blockAjax2017-01-041-4/+17
| |
| * LLVM: avoid division by zero, using Miasm wrapper insteadAjax2017-01-041-4/+12
| |
| * LLVM: handle the specific size of segm2addr operationAjax2017-01-041-5/+17
| |
| * LLVM: Add missing operation BSR/BSFAjax2017-01-041-0/+8
| |