view gen/naked.cpp @ 1047:6bb04dbee21f

Some calling convention work for x86-64: - Implement x86-64 extern(C), hopefully correctly. - Tried to be a bit smarter about extern(D) while I was there. Interestingly, this code seems to be generating more efficient code than gcc and llvm-gcc in some edge cases, like returning a `{ [7 x i8] }` loaded from a stack slot from an extern(C) function. (gcc generates 7 1-byte loads, while this code generates a 4-byte, a 2-byte and a 1-byte load) I also added some changes to make sure structs being returned from functions or passed in as parameters are stored in memory where the rest of the backend seems to expect them to be. These should be removed when support for first-class aggregates improves.
author Frits van Bommel <fvbommel wxs.nl>
date Fri, 06 Mar 2009 16:00:47 +0100
parents 474d7dd54d43
children dbe4af57b240
line wrap: on
line source

#include "gen/llvm.h"

#include "expression.h"
#include "statement.h"
#include "declaration.h"

#include <cassert>

#include "gen/logger.h"
#include "gen/irstate.h"
#include "gen/llvmhelpers.h"
#include "gen/tollvm.h"

//////////////////////////////////////////////////////////////////////////////////////////

void Statement::toNakedIR(IRState *p)
{
    error("not allowed in naked function");
}

//////////////////////////////////////////////////////////////////////////////////////////

void CompoundStatement::toNakedIR(IRState *p)
{
    Logger::println("CompoundStatement::toNakedIR(): %s", loc.toChars());
    LOG_SCOPE;

    if (statements)
    for (unsigned i = 0; i < statements->dim; i++)
    {
        Statement* s = (Statement*)statements->data[i];
        if (s) s->toNakedIR(p);
    }
}

//////////////////////////////////////////////////////////////////////////////////////////

void ExpStatement::toNakedIR(IRState *p)
{
    Logger::println("ExpStatement::toNakedIR(): %s", loc.toChars());
    LOG_SCOPE;

    // only expstmt supported in declarations
    if (exp->op != TOKdeclaration)
    {
        Statement::toNakedIR(p);
        return;
    }

    DeclarationExp* d = (DeclarationExp*)exp;
    VarDeclaration* vd = d->declaration->isVarDeclaration();
    FuncDeclaration* fd = d->declaration->isFuncDeclaration();
    EnumDeclaration* ed = d->declaration->isEnumDeclaration();

    // and only static variable/function declaration
    // no locals or nested stuffies!
    if (!vd && !fd && !ed)
    {
        Statement::toNakedIR(p);
        return;
    }
    else if (vd && !vd->isDataseg())
    {
        error("non-static variable '%s' not allowed in naked function", vd->toChars());
        return;
    }
    else if (fd && !fd->isStatic())
    {
        error("non-static nested function '%s' not allowed in naked function", fd->toChars());
        return;
    }
    // enum decls should always be safe

    // make sure the symbols gets processed
    d->declaration->toObjFile(0);
}

//////////////////////////////////////////////////////////////////////////////////////////

void LabelStatement::toNakedIR(IRState *p)
{
    Logger::println("LabelStatement::toNakedIR(): %s", loc.toChars());
    LOG_SCOPE;

    p->nakedAsm << p->func()->decl->mangle() << "_" << ident->toChars() << ":";

    if (statement)
        statement->toNakedIR(p);
}

//////////////////////////////////////////////////////////////////////////////////////////

void DtoDefineNakedFunction(FuncDeclaration* fd)
{
    Logger::println("DtoDefineNakedFunction(%s)", fd->mangle());
    LOG_SCOPE;

    assert(fd->ir.irFunc);
    gIR->functions.push_back(fd->ir.irFunc);

    // we need to do special processing on the body, since we only want
    // to allow actual inline asm blocks to reach the final asm output

    std::ostringstream& asmstr = gIR->nakedAsm;

    // build function header

    // FIXME: could we perhaps use llvm asmwriter to give us these details ?

    const char* mangle = fd->mangle();
    std::ostringstream tmpstr;

    // osx is different
    // also mangling has an extra underscore prefixed
    if (global.params.os == OSMacOSX)
    {
        std::string section = "text";
        bool weak = false;
        if (DtoIsTemplateInstance(fd))
        {
            tmpstr << "section\t__TEXT,__textcoal_nt,coalesced,pure_instructions";
            section = tmpstr.str();
            weak = true;
        }
        asmstr << "\t." << section << std::endl;
        asmstr << "\t.align\t4,0x90" << std::endl;
        asmstr << "\t.globl\t_" << mangle << std::endl;
        if (weak)
        {
            asmstr << "\t.weak_definition\t_" << mangle << std::endl;
        }
        asmstr << "_" << mangle << ":" << std::endl;
    }
    // this works on linux x86 32 and 64 bit
    // assume it works everywhere else as well for now
    else
    {
        const char* linkage = "globl";
        std::string section = "text";
        if (DtoIsTemplateInstance(fd))
        {
            linkage = "weak";
            tmpstr << "section\t.gnu.linkonce.t." << mangle << ",\"ax\",@progbits";
            section = tmpstr.str();
        }
        asmstr << "\t." << section << std::endl;
        asmstr << "\t.align\t16" << std::endl;
        asmstr << "\t." << linkage << "\t" << mangle << std::endl;
        asmstr << "\t.type\t" << mangle << ",@function" << std::endl;
        asmstr << mangle << ":" << std::endl;
    }

    // emit body
    fd->fbody->toNakedIR(gIR);

    // emit size after body
    // llvm does this on linux, but not on osx
    if (global.params.os != OSMacOSX)
    {
        asmstr << "\t.size\t" << mangle << ", .-" << mangle << std::endl << std::endl;
    }

    gIR->module->appendModuleInlineAsm(asmstr.str());
    asmstr.str("");

    gIR->functions.pop_back();
}

//////////////////////////////////////////////////////////////////////////////////////////

void emitABIReturnAsmStmt(IRAsmBlock* asmblock, Loc loc, FuncDeclaration* fdecl)
{
    Logger::println("emitABIReturnAsmStmt(%s)", fdecl->mangle());
    LOG_SCOPE;

    IRAsmStmt* as = new IRAsmStmt;

    const LLType* llretTy = DtoType(fdecl->type->nextOf());
    asmblock->retty = llretTy;
    asmblock->retn = 1;

    // FIXME: This should probably be handled by the TargetABI somehow.
    //        It should be able to do this for a greater variety of types.

    // x86
    if (global.params.cpu == ARCHx86)
    {
        LINK l = fdecl->linkage;
        assert((l == LINKd || l == LINKc || l == LINKwindows) && "invalid linkage for asm implicit return");

        Type* rt = fdecl->type->nextOf()->toBasetype();
        if (rt->isintegral() || rt->ty == Tpointer || rt->ty == Tclass || rt->ty == Taarray)
        {
            if (rt->size() == 8) {
                as->out_c = "=A,";
            } else {
                as->out_c = "={ax},";
            }
        }
        else if (rt->isfloating())
        {
            if (rt->iscomplex()) {
                if (fdecl->linkage == LINKd) {
                    // extern(D) always returns on the FPU stack
                    as->out_c = "={st},={st(1)},";
                    asmblock->retn = 2;
                } else if (rt->ty == Tcomplex32) {
                    // extern(C) cfloat is return as i64
                    as->out_c = "=A,";
                    asmblock->retty = LLType::Int64Ty;
                } else {
                    // cdouble and creal extern(C) are returned in pointer
                    // don't add anything!
                    asmblock->retty = LLType::VoidTy;
                    asmblock->retn = 0;
                    return;
                }
            } else {
                as->out_c = "={st},";
            }
        }
        else if (rt->ty == Tarray || rt->ty == Tdelegate)
        {
            as->out_c = "={ax},={dx},";
            asmblock->retn = 2;
        #if 0
            // this is to show how to allocate a temporary for the return value
            // in case the appropriate multi register constraint isn't supported.
            // this way abi return from inline asm can still be emulated.
            // note that "$<<out0>>" etc in the asm will translate to the correct
            // numbered output when the asm block in finalized

            // generate asm
            as->out_c = "=*m,=*m,";
            LLValue* tmp = DtoAlloca(llretTy, ".tmp_asm_ret");
            as->out.push_back( tmp );
            as->out.push_back( DtoGEPi(tmp, 0,1) );
            as->code = "movd %eax, $<<out0>>" "\n\t" "mov %edx, $<<out1>>";

            // fix asmblock
            asmblock->retn = 0;
            asmblock->retemu = true;
            asmblock->asmBlock->abiret = tmp;

            // add "ret" stmt at the end of the block
            asmblock->s.push_back(as);

            // done, we don't want anything pushed in the front of the block
            return;
        #endif
        }
        else
        {
            error(loc, "unimplemented return type '%s' for implicit abi return", rt->toChars());
            fatal();
        }
    }

    // x86_64
    else if (global.params.cpu == ARCHx86_64)
    {
        LINK l = fdecl->linkage;
        /* TODO: Check if this works with extern(Windows), completely untested.
         *       In particular, returning cdouble may not work with
         *       extern(Windows) since according to X86CallingConv.td it
         *       doesn't allow XMM1 to be used.
         * (So is extern(C), but that should be fine as the calling convention
         * is identical to that of extern(D))
         */
        assert((l == LINKd || l == LINKc || l == LINKwindows) && "invalid linkage for asm implicit return");

        Type* rt = fdecl->type->nextOf()->toBasetype();
        if (rt->isintegral() || rt->ty == Tpointer || rt->ty == Tclass || rt->ty == Taarray)
        {
            as->out_c = "={ax},";
        }
        else if (rt->isfloating())
        {
            if (rt == Type::tcomplex80) {
                // On x87 stack, re=st, im=st(1)
                as->out_c = "={st},={st(1)},";
                asmblock->retn = 2;
            } else if (rt == Type::tfloat80 || rt == Type::timaginary80) {
                // On x87 stack
                as->out_c = "={st},";
            } else if (l != LINKd && rt == Type::tcomplex32) {
                // LLVM and GCC disagree on how to return {float, float}.
                // For compatibility, use the GCC/LLVM-GCC way for extern(C/Windows)
                // extern(C) cfloat -> %xmm0 (extract two floats)
                as->out_c = "={xmm0},";
                asmblock->retty = LLType::DoubleTy;
            } else if (rt->iscomplex()) {
                // cdouble and extern(D) cfloat -> re=%xmm0, im=%xmm1
                as->out_c = "={xmm0},={xmm1},";
                asmblock->retn = 2;
            } else {
                // Plain float/double/ifloat/idouble
                as->out_c = "={xmm0},";
            }
        }
        else if (rt->ty == Tarray || rt->ty == Tdelegate)
        {
            as->out_c = "={ax},={dx},";
            asmblock->retn = 2;
        }
        else
        {
            error(loc, "unimplemented return type '%s' for implicit abi return", rt->toChars());
            fatal();
        }
    }

    // unsupported
    else
    {
        error(loc, "this target (%s) does not implement inline asm falling off the end of the function", global.params.targetTriple);
        fatal();
    }

    // return values always go in the front
    asmblock->s.push_front(as);
}