- implemented __builtin_return_address(0)
[libfirm] / ir / be / ia32 / bearch_ia32_t.h
index 3241445..a85ca4b 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 1995-2007 University of Karlsruhe.  All right reserved.
+ * Copyright (C) 1995-2008 University of Karlsruhe.  All right reserved.
  *
  * This file is part of libFirm.
  *
@@ -26,8 +26,7 @@
 #ifndef FIRM_BE_IA32_BEARCH_IA32_T_H
 #define FIRM_BE_IA32_BEARCH_IA32_T_H
 
-#include "firm_config.h"
-
+#include "config.h"
 #include "pmap.h"
 #include "debug.h"
 #include "ia32_nodes_attr.h"
@@ -41,7 +40,7 @@
 #ifdef NDEBUG
 #define SET_IA32_ORIG_NODE(n, o)
 #else  /* ! NDEBUG */
-#define SET_IA32_ORIG_NODE(n, o) set_ia32_orig_node(n, o);
+#define SET_IA32_ORIG_NODE(n, o) set_ia32_orig_node(n, o)
 #endif /* NDEBUG */
 
 /* some typedefs */
@@ -49,106 +48,6 @@ typedef enum ia32_optimize_t ia32_optimize_t;
 typedef enum cpu_support     cpu_support;
 typedef enum fp_support      fp_support;
 
-/**
- * Bitmask for the backend optimization settings.
- */
-enum ia32_optimize_t {
-       IA32_OPT_INCDEC    = 1 << 0,   /**< optimize add/sub 1/-1 to inc/dec */
-       IA32_OPT_CC        = 1 << 1,   /**< optimize calling convention of private
-                                           functions */
-       IA32_OPT_UNSAFE_FLOATCONV = 1 << 2, /**< disrespect current floating
-                                  point rounding mode at entry and exit of
-                                  functions (this is ok for programs that don't
-                                  explicitly change the rounding mode) */
-};
-
-/**
- * CPU features.
- */
-enum cpu_arch_features {
-       arch_feature_intel    = 0x80000000,                      /**< Intel CPU */
-       arch_feature_amd      = 0x40000000,                      /**< AMD CPU */
-       arch_feature_p6       = 0x20000000,                      /**< P6 instructions */
-       arch_feature_mmx      = 0x10000000,                      /**< MMX instructions */
-       arch_feature_sse1     = 0x08000000 | arch_feature_mmx,   /**< SSE1 instructions, include MMX */
-       arch_feature_sse2     = 0x04000000 | arch_feature_sse1,  /**< SSE2 instructions, include SSE1 */
-       arch_feature_sse3     = 0x02000000 | arch_feature_sse2,  /**< SSE3 instructions, include SSE2 */
-       arch_feature_ssse3    = 0x01000000 | arch_feature_sse3,  /**< SSSE3 instructions, include SSE3 */
-       arch_feature_3DNow    = 0x00800000,                      /**< 3DNow! instructions */
-       arch_feature_3DNowE   = 0x00400000 | arch_feature_3DNow, /**< Enhanced 3DNow! instructions */
-       arch_feature_netburst = 0x00200000 | arch_feature_intel, /**< Netburst architecture */
-       arch_feature_64bit    = 0x00100000 | arch_feature_sse2,  /**< x86_64 support, include SSE2 */
-};
-
-/**
- * Architectures.
- */
-enum cpu_support {
-       /* intel CPU's */
-       arch_generic     =  0,
-
-       arch_i386        =  1,
-       arch_i486        =  2,
-       arch_pentium     =  3 | arch_feature_intel,
-       arch_pentium_mmx =  4 | arch_feature_intel | arch_feature_mmx,
-       arch_pentium_pro =  5 | arch_feature_intel | arch_feature_p6,
-       arch_pentium_2   =  6 | arch_feature_intel | arch_feature_p6 | arch_feature_mmx,
-       arch_pentium_3   =  7 | arch_feature_intel | arch_feature_p6 | arch_feature_sse1,
-       arch_pentium_4   =  8 | arch_feature_netburst | arch_feature_p6 | arch_feature_sse2,
-       arch_pentium_m   =  9 | arch_feature_intel | arch_feature_p6 | arch_feature_sse2,
-       arch_core        = 10 | arch_feature_intel | arch_feature_p6 | arch_feature_sse3,
-       arch_prescott    = 11 | arch_feature_netburst | arch_feature_p6 | arch_feature_sse3,
-       arch_core2       = 12 | arch_feature_intel | arch_feature_p6 | arch_feature_64bit | arch_feature_ssse3,
-
-       /* AMD CPU's */
-       arch_k6          = 13 | arch_feature_amd | arch_feature_mmx,
-       arch_k6_2        = 14 | arch_feature_amd | arch_feature_mmx | arch_feature_3DNow,
-       arch_k6_3        = 15 | arch_feature_amd | arch_feature_mmx | arch_feature_3DNow,
-       arch_athlon      = 16 | arch_feature_amd | arch_feature_mmx | arch_feature_3DNowE | arch_feature_p6,
-       arch_athlon_xp   = 17 | arch_feature_amd | arch_feature_sse1 | arch_feature_3DNowE | arch_feature_p6,
-       arch_opteron     = 18 | arch_feature_amd | arch_feature_64bit | arch_feature_3DNowE | arch_feature_p6,
-
-       /* other */
-       arch_winchip_c6  = 19 | arch_feature_mmx,
-       arch_winchip2    = 20 | arch_feature_mmx | arch_feature_3DNow,
-       arch_c3          = 21 | arch_feature_mmx | arch_feature_3DNow,
-       arch_c3_2        = 22 | arch_feature_sse1,  /* really no 3DNow! */
-};
-
-/** checks for l <= x <= h */
-#define _IN_RANGE(x, l, h)  ((unsigned)((x) - (l)) <= (unsigned)((h) - (l)))
-
-/** returns true if it's Intel architecture */
-#define ARCH_INTEL(x)       ((x) & arch_feature_intel)
-
-/** returns true if it's AMD architecture */
-#define ARCH_AMD(x)         ((x) & arch_feature_amd)
-
-/** return true if it's a Athlon/Opteron */
-#define ARCH_ATHLON(x)      _IN_RANGE((x), arch_athlon, arch_opteron)
-
-/** return true if the CPU has MMX support */
-#define ARCH_MMX(x)         ((x) & arch_feature_mmx)
-
-/** return true if the CPU has 3DNow! support */
-#define ARCH_3DNow(x)       ((x) & arch_feature_3DNow)
-
-/** return true if the CPU has P6 features (CMOV) */
-#define IS_P6_ARCH(x)       ((x) & arch_feature_p6)
-
-/** floating point support */
-enum fp_support {
-       fp_none,  /**< no floating point instructions are used */
-       fp_x87,   /**< use x87 instructions */
-       fp_sse2   /**< use SSE2 instructions */
-};
-
-/** Returns non-zero if the current floating point architecture is SSE2. */
-#define USE_SSE2(cg) ((cg)->fp_kind == fp_sse2)
-
-/** Returns non-zero if the current floating point architecture is x87. */
-#define USE_x87(cg)  ((cg)->fp_kind == fp_x87)
-
 typedef struct ia32_isa_t            ia32_isa_t;
 typedef struct ia32_code_gen_t       ia32_code_gen_t;
 typedef struct ia32_irn_ops_t        ia32_irn_ops_t;
@@ -160,43 +59,36 @@ typedef struct ia32_intrinsic_env_t  ia32_intrinsic_env_t;
 struct ia32_code_gen_t {
        const arch_code_generator_if_t *impl;          /**< implementation */
        ir_graph                       *irg;           /**< current irg */
-       const arch_env_t               *arch_env;      /**< the arch env */
        set                            *reg_set;       /**< set to memorize registers for non-ia32 nodes (e.g. phi nodes) */
        ia32_isa_t                     *isa;           /**< for fast access to the isa object */
        be_irg_t                       *birg;          /**< The be-irg (contains additional information about the irg) */
        ir_node                        **blk_sched;    /**< an array containing the scheduled blocks */
-       ia32_optimize_t                opt;            /**< contains optimization information */
-       int                            arch;           /**< instruction architecture */
-       int                            opt_arch;       /**< optimize for architecture */
-       char                           fp_kind;        /**< floating point kind */
-       char                           do_x87_sim;     /**< set to 1 if x87 simulation should be enforced */
-       char                           dump;           /**< set to 1 if graphs should be dumped */
-       ir_node                       *unknown_gp;     /**< unique Unknown_GP node */
-       ir_node                       *unknown_vfp;    /**< unique Unknown_VFP node */
-       ir_node                       *unknown_xmm;    /**< unique Unknown_XMM node */
-       ir_node                       *noreg_gp;       /**< unique NoReg_GP node */
-       ir_node                       *noreg_vfp;      /**< unique NoReg_VFP node */
-       ir_node                       *noreg_xmm;      /**< unique NoReg_XMM node */
-
-       ir_node                       *fpu_trunc_mode; /**< truncate fpu mode */
-
-       struct obstack                *obst;
+       unsigned                       do_x87_sim:1;   /**< set to 1 if x87 simulation should be enforced */
+       unsigned                       dump:1;         /**< set to 1 if graphs should be dumped */
+       unsigned                       gprof:1;        /**< set to 1 grof profiling is in use */
+       ir_node                        *unknown_gp;     /**< unique Unknown_GP node */
+       ir_node                        *unknown_vfp;    /**< unique Unknown_VFP node */
+       ir_node                        *unknown_xmm;    /**< unique Unknown_XMM node */
+       ir_node                        *noreg_gp;       /**< unique NoReg_GP node */
+       ir_node                        *noreg_vfp;      /**< unique NoReg_VFP node */
+       ir_node                        *noreg_xmm;      /**< unique NoReg_XMM node */
+
+       ir_node                        *fpu_trunc_mode; /**< truncate fpu mode */
+       ir_node                        *get_eip;        /**< get eip node */
+
+       struct obstack                 *obst;
 };
 
 /**
  * IA32 ISA object
  */
 struct ia32_isa_t {
-       arch_isa_t            arch_isa;       /**< must be derived from arch_isa_t */
+       arch_env_t            arch_env;       /**< must be derived from arch_env_t */
        pmap                  *regs_16bit;    /**< Contains the 16bits names of the gp registers */
        pmap                  *regs_8bit;     /**< Contains the 8bits names of the gp registers */
        pmap                  *regs_8bit_high; /**< contains the hight part of the 8 bit names of the gp registers */
        pmap                  *types;         /**< A map of modes to primitive types */
        pmap                  *tv_ent;        /**< A map of entities that store const tarvals */
-       ia32_optimize_t       opt;            /**< contains optimization information */
-       int                   arch;           /**< instruction architecture */
-       int                   opt_arch;       /**< optimize for architecture */
-       int                   fp_kind;        /**< floating point kind */
        ia32_code_gen_t       *cg;            /**< the current code generator */
        const be_machine_t    *cpu;           /**< the abstract machine */
 #ifndef NDEBUG
@@ -204,28 +96,34 @@ struct ia32_isa_t {
 #endif /* NDEBUG */
 };
 
-struct ia32_irn_ops_t {
-       const arch_irn_ops_if_t *impl;
-       ia32_code_gen_t         *cg;
-};
-
 /**
  * A helper type collecting needed info for IA32 intrinsic lowering.
  */
 struct ia32_intrinsic_env_t {
        ia32_isa_t *isa;          /**< the isa object */
        ir_graph   *irg;          /**< the irg, these entities belong to */
-       ir_entity  *ll_div_op1;   /**< entity for first div operand (move into FPU) */
-       ir_entity  *ll_div_op2;   /**< entity for second div operand (move into FPU) */
-       ir_entity  *ll_d_conv;    /**< entity for converts ll -> d */
-       ir_entity  *d_ll_conv;    /**< entity for converts d -> ll */
        ir_entity  *divdi3;       /**< entity for __divdi3 library call */
        ir_entity  *moddi3;       /**< entity for __moddi3 library call */
        ir_entity  *udivdi3;      /**< entity for __udivdi3 library call */
        ir_entity  *umoddi3;      /**< entity for __umoddi3 library call */
-       tarval     *u64_bias;     /**< bias value for conversion from float to unsigned 64 */
 };
 
+typedef enum transformer_t {
+       TRANSFORMER_DEFAULT,
+#ifdef FIRM_GRGEN_BE
+       TRANSFORMER_PBQP,
+       TRANSFORMER_RAND
+#endif
+} transformer_t;
+
+#ifdef FIRM_GRGEN_BE
+/** The selected transformer. */
+extern transformer_t be_transformer;
+
+#else
+#define be_transformer TRANSFORMER_DEFAULT
+#endif
+
 /** The mode for the floating point control word. */
 extern ir_mode *mode_fpcw;
 
@@ -247,20 +145,16 @@ ir_node *ia32_new_Unknown_gp(ia32_code_gen_t *cg);
 ir_node *ia32_new_Unknown_xmm(ia32_code_gen_t *cg);
 ir_node *ia32_new_Unknown_vfp(ia32_code_gen_t *cg);
 
-/**
- * Returns the unique per irg FP NoReg node.
- */
-ir_node *ia32_new_NoReg_fp(ia32_code_gen_t *cg);
-
 /**
  * Returns the unique per irg FPU truncation mode node.
  */
 ir_node *ia32_new_Fpu_truncate(ia32_code_gen_t *cg);
 
 /**
- * Returns gp_noreg or fp_noreg, depending on input requirements.
+ * Split instruction with source AM into Load and separate instruction.
+ * @return result of the Load
  */
-ir_node *ia32_get_admissible_noreg(ia32_code_gen_t *cg, ir_node *irn, int pos);
+ir_node *turn_back_am(ir_node *node);
 
 /**
  * Maps all intrinsic calls that the backend support
@@ -282,4 +176,9 @@ ir_entity *ia32_create_intrinsic_fkt(ir_type *method, const ir_op *op,
                                      const ir_mode *imode, const ir_mode *omode,
                                      void *context);
 
+/**
+ * Return the stack entity that contains the return address.
+ */
+ir_entity *ia32_get_return_address_entity(void);
+
 #endif