summaryrefslogtreecommitdiff
path: root/compiler/nativeGen/X86
Commit message (Collapse)AuthorAgeFilesLines
...
* Fix compile failure on non x86/x86-64 (#7054).Erik de Castro Lopo2012-07-091-0/+5
|
* Allow the register allocator access to argument regs (R1.., F1.., etc.)Simon Marlow2012-07-064-73/+46
| | | | | | | | | | | | | | | This was made possible by the recent change to codeGen to attach the live GlobalRegs to every CmmJump, and we'll be relying on it quite heavily in the new code generator too. What this means essentially is that when we see x = R1 the register allocator will automatically assign x to R1 and generate no code at all (also known as "coalescing"). It wasn't possible before because the register allocator had to assume that R1 was always live, because it didn't have access to accurate liveness information.
* Remove lots of commented out 'in' keywordsIan Lynagh2012-06-131-30/+0
|
* Remove PlatformOutputableIan Lynagh2012-06-132-10/+6
| | | | | We can now get the Platform from the DynFlags inside an SDoc, so we no longer need to pass the Platform in.
* Use SDoc rather than Doc in the native gensIan Lynagh2012-06-122-80/+77
| | | | This avoid lots of converting back and forth between the two types.
* Change how macros like ASSERT are definedIan Lynagh2012-06-051-0/+1
| | | | | By using Haskell's debugIsOn rather than CPP's "#ifdef DEBUG", we don't need to kludge things to keep the warning checker happy etc.
* Add an X86/amd64 implementation for quotRemWord2Ian Lynagh2012-04-211-20/+50
|
* Add a quotRemWord2 primopIan Lynagh2012-04-211-6/+7
| | | | | | | | It allows you to do (high, low) `quotRem` d provided high < d. Currently only has an inefficient fallback implementation.
* Fix compile for CPUs other than x86/x86_64.Erik de Castro Lopo2012-03-221-2/+4
|
* Implement the Adjustor for Win64Ian Lynagh2012-03-211-2/+6
|
* Fixes for the calling convention on Win64Ian Lynagh2012-03-212-19/+62
| | | | In particular, fixes for FP arguments
* Rename allArgRegs to allIntArgRegsIan Lynagh2012-03-212-9/+9
|
* Fix for Win64 codegenIan Lynagh2012-03-201-7/+22
| | | | | We need to leave stack space for arguments that we are passing in registers.
* Define allArgRegs correctly for Win64Ian Lynagh2012-03-191-0/+4
|
* Fix the unregisterised build; fixes #5901Ian Lynagh2012-02-271-4/+4
|
* Add x86 implementations of the quotRem, Mul2 and Add2 primopsIan Lynagh2012-02-241-2/+59
|
* Implement 2-word-multiply for x86_64Ian Lynagh2012-02-243-0/+19
|
* Add a 2-word-multiply operatorIan Lynagh2012-02-241-0/+1
| | | | Currently no NCGs support it
* Add x86_64 support for the add-with-carry opIan Lynagh2012-02-231-0/+13
|
* Add a Word add-with-carry primopIan Lynagh2012-02-231-19/+17
| | | | No special-casing in any NCGs yet
* Call expandCallishMachOp in the x86_64 codegen tooIan Lynagh2012-02-231-0/+4
| | | | | Currently it does nothing, as x86_64 supports all the callishMachOps that expandCallishMachOp expands, but it might be needed in the future.
* Add a primop for unsigned quotRem; part of #5598Ian Lynagh2012-02-171-15/+23
| | | | Only amd64 has an efficient implementation currently.
* Small refactorIan Lynagh2012-02-171-87/+92
| | | | Moved the default case of genCCall64 out into a separate function
* Define a quotRem CallishMachOp; fixes #5598Ian Lynagh2012-02-141-4/+27
| | | | | This means we no longer do a division twice when we are using quotRem (on platforms on which the op is supported; currently only amd64).
* Track STG live register information for use in LLVMDavid Terei2012-01-091-1/+1
| | | | | | | | | We now carry around with CmmJump statements a list of the STG registers that are live at that jump site. This is used by the LLVM backend so it can avoid unnesecarily passing around dead registers, improving perfromance. This gives us the framework to finally fix trac #4308.
* Remove unused arg field of CmmReturnDavid Terei2012-01-051-1/+1
|
* Remove unused argument field on CmmJumpDavid Terei2012-01-051-1/+1
|
* small refactoringSimon Marlow2012-01-051-2/+3
|
* We must emit DELTA pseudo-instructions when moving %esp (#5747)Simon Marlow2012-01-051-1/+3
|
* Make getDynFlags* functions use HasDynFlags/getDynFlags tooIan Lynagh2011-12-191-13/+13
|
* some small optimisationsSimon Marlow2011-12-131-33/+33
|
* Get rid of the "safety" field of CmmCall (OldCmm)Simon Marlow2011-11-291-2/+2
| | | | | This field was doing nothing. I think it originally appeared in a very old incarnation of the new code generator.
* Explicitly handle unsupported Cmm prim ops.David Terei2011-11-221-2/+4
|
* Better documentation for stack alignment designDavid Terei2011-11-171-25/+22
|
* Fix #4211: No need to fixup stack using mangler on OSXDavid Terei2011-11-171-3/+3
| | | | | | | We now manage the stack correctly on both x86 and i386, keeping the stack align at (16n bytes - word size) on function entry and at (16n bytes) on function calls. This gives us compatability with LLVM and GCC.
* Allow the use of R9 and R10 in primops; fixes trac #5423Ian Lynagh2011-11-061-0/+6
|
* Use -fwarn-tabs when validatingIan Lynagh2011-11-042-0/+14
| | | | | We only use it for "compiler" sources, i.e. not for libraries. Many modules have a -fno-warn-tabs kludge for now.
* Ignore stdcall c-call in native codegen on x86_64David M Peixotto2011-11-011-2/+3
| | | | | | | | | | | | | | The stdcall calling convention is not supported on x86_64. When an ffi import requests stdcall, a warning is issued as desired by #3336. However, the native codegen was still generating code that expected the callee to cleanup the stack arguments when calling a c function that requests stdcall. This patch changes the codegen to actually use the ccall calling convention as intended. Signed-off-by: David Terei <davidterei@gmail.com>
* Change stack alignment to 16+8 bytes in STG codeDavid M Peixotto2011-11-011-7/+9
| | | | | | | | | | | | | | | | | This patch changes the STG code so that %rsp to be aligned to a 16-byte boundary + 8. This is the alignment required by the x86_64 ABI on entry to a function. Previously we kept %rsp aligned to a 16-byte boundary, but this was causing problems for the LLVM backend (see #4211). We now don't need to invoke llvm stack mangler on x86_64 targets. Since the stack is now 16+8 byte algined in STG land on x86_64, we don't need to mangle the stack manipulations with the llvm mangler. This patch only modifies the alignement for x86_64 backends. Signed-off-by: David Terei <davidterei@gmail.com>
* Eliminate all uses of IF_ARCH_i386, and remove the definitionIan Lynagh2011-10-232-22/+32
|
* Eliminate a couple of IF_ARCH_i386'sIan Lynagh2011-10-231-2/+2
|
* More CPP removalIan Lynagh2011-10-233-48/+46
|
* Correct a commentIan Lynagh2011-10-231-1/+1
|
* A little more CPP removalIan Lynagh2011-10-191-22/+16
|
* Remove all Haskell conditionals on HAVE_SUBSECTIONS_VIA_SYMBOLSIan Lynagh2011-10-191-18/+18
| | | | We now use the value from the targetPlatform instead.
* Revert "Remove ArchUnknown"Ian Lynagh2011-10-191-0/+1
| | | | | | | This reverts commit 2dea11a442e1d14d86fa661804de06a721943bf0. On second thoughts, this does make sense, for unregisterised via-C arches at least.
* Remove ArchUnknownIan Lynagh2011-10-181-1/+0
| | | | | | | It doesn't make sense. If platformArch is ArchUnknown then we don't know the answer to any questions about the arch. So now if we don't recognise the arch we just fail, and the new arch will need to be added to the datatype.
* Tweak a commentIan Lynagh2011-10-141-2/+1
|
* More CPP removal: pprDynamicLinkerAsmLabel in CLabelIan Lynagh2011-10-022-56/+62
| | | | And some knock-on changes
* Make an import unconditionalIan Lynagh2011-09-061-4/+0
| | | | Fixes build on non-Intel/PPC platforms
View Lorry Controller Status