/* Copyright (C) 2021 Free Software Foundation, Inc. Contributed by Oracle. This file is part of GNU Binutils. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA. */ #ifndef _COMP_COM_H #define _COMP_COM_H #include #include /* * This file describes format for the compiler-commentary * section to be added to .o's and propagated to the a.out. It reflects * information the compiler can expose to the user about his or her * program. The section should be generated for all compiles where * the user has specified -g on the compile line. * * In the analyzer, display of the messages will be governed by a user UI * that sets a vis_bits bitmap, and matches it against a show_bits * bitmap table, which is maintained separately from the producer * code. For any message, if (vis_bits&show_bits) is non-zero, the * message is shown. If zero, the message is not shown. A similar * mechanism would be used for a stand-alone source or disassembly browser. * * * The .compcom Section * -------------------- * The section will be named ".compcom"; it is generated for each * .o, and aggregated into a single section in the a.out. In that * section, each .o's data is separate, and the tools will loop * over the data for each .o in order to find the subsection for * the particular .o being annotated. * * * Since the header is fixed-length, and the total size of the section * can be easily determined as: * * sizeof(stuct compcomhdr) * + msgcount * sizeof(struct compmsg) * + paramcount * sizeof(int32_t) * + stringlen * * there is no need to have the size in the header. */ typedef struct { /* Header describing the section */ int32_t srcname; /* index into strings of source file path */ int32_t version; /* a version number for the .compcom format */ int32_t msgcount; /* count of messages in the section */ int32_t paramcount; /* count of parameters in the section */ int32_t stringcount; /* count of strings in the section */ int32_t stringlen; /* count of total bytes in strings */ } compcomhdr; /* * The data for the .o after the header as: * * compmsg msgs[msgcount]; the array of messages * int32_t param[paramcount]; the parameters used in the messages * parameters are either integers or * string-indices * char msgstrings[stringlen]; the strings used in the messages */ /* * Message Classes and Visualization Bits * -------------------------------------- * Each of the messages above may belong to zero or more visualization * classes, governed by a table using zero or more of the following symbolic * names for the classes: */ typedef enum { CCMV_WANT = 0x000, /* High-priority RFE -- used only for human */ /* reading of message list */ CCMV_UNIMPL = 0x000, /* Unimplemented -- used only for human */ /* reading of message list */ CCMV_OBS = 0x000, /* Obsolete -- to be replaced by a different */ /* message with different parameters -- */ /* used only for human reading of message */ /* list */ CCMV_VER = 0x001, /* Versioning messages */ CCMV_WARN = 0x002, /* Warning messages */ CCMV_PAR = 0x004, /* Parallelization messages */ CCMV_QUERY = 0x008, /* Compiler queries */ CCMV_LOOP = 0x010, /* Loop detail messages */ CCMV_PIPE = 0x020, /* Pipelining messages */ CCMV_INLINE = 0x040, /* Inlining information */ CCMV_MEMOPS = 0x080, /* Messages concerning memory operations */ CCMV_FE = 0x100, /* Front-end messages (all compilers) */ CCMV_CG = 0x200, /* Code-generator messages (all compilers) */ CCMV_BASIC = 0x400, /* for default messages */ CCMV_ALL = 0x7FFFFFFF /* for all messages */ } COMPCLASS_ID; typedef enum ccm_msgid { /* Group: Versioning Messages */ /* All of these are global to the .o, and will */ /* have lineno = pcoffset = 0 */ CCM_MODDATE=0x00100, /* Source file , last modified on date */ CCM_COMPVER, /* Component , version */ /* [Emitted for each component of the compiler.] */ CCM_COMPDATE, /* Compilation date */ /* [ is an I18n string with the date and time] */ CCM_COMPOPT, /* Compilation options */ /* [As specified by the user] */ CCM_ACOMPOPT, /* Actual Compilation options */ /* [As expanded by the driver] */ /* Group: Warning Messages */ CCM_VAR_ALIAS=0x00200, /* Variable aliased to */ CCM_FBIRDIFF, /* Profile feedback data inconsistent with */ /* intermediate representation file; check compiler */ /* version, flags and source file */ CCM_OPTRED_SWAP, /* Optimization level for reduced from to */ /* due to insufficient swap space */ CCM_OPTRED_CPLX, /* Optimization level for reduced from to */ /* due to program complexity */ CCM_UNKNOWN, /* Unexpected compiler comment */ /* Group: Parallelization Messages */ CCM_UNPAR_CALL=0x00400, /* Loop below not parallelized because it contains a */ /* call to */ /* CCMV_WANT: Don't generate CCM_PAR_SER; always use CCM_PAR_SER_VER */ CCM_PAR_SER, /* Both serial and parallel versions generated for */ /* loop below */ CCM_PAR_SER_VER, /* Both serial and parallel versions generated for */ /* loop below; with parallel version used if , */ /* serial otherwise */ CCM_PAR_DRECTV, /* Loop below parallelized by explicit user */ /* directive */ CCM_APAR, /* Loop below autoparallelized */ CCM_AUTOPAR, /* Loop below autoparallelized; equivalent */ /* explict directive is */ CCM_UNPAR_DD, /* Loop below could not be parallelized because of a */ /* data dependency on , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_UNPAR_DDA, /* Loop below could not be parallelized because of a */ /* data dependency or aliasing of , , ... */ CCM_UNPAR_ANONDD, /* Loop below could not be parallelized because of */ /* an anonymous data dependency */ CCM_UNPAR_ANONDDA, /* Loop below could not be parallelized because of */ /* an anonymous data dependency or aliasing */ CCM_PAR_WORK, /* Loop below parallelized, but might not contain */ /* enough work to be efficiently run in parallel */ CCM_UNPAR_EXIT, /* Loop below not parallelized because it contains */ /* multiple exit points */ CCM_UNPAR_STRNG, /* Loop below not parallelized because it contains a */ /* strange flow of control */ CCM_UNPAR_IO, /* Loop below not parallelized because it contains */ /* I/O or other MT-unsafe calls */ CCM_PAR_BODY_NAME, /* Parallel loop-body code is in function */ CCM_UNPAR_NLOOPIDX, /* Loop below not parallelized because loop index */ /* not found */ CCM_UNPAR_DRECTV, /* Loop below not parallelized because of explicit */ /* user directive */ CCM_UNPAR_NOTPROFIT, /* Loop below not parallelized because it was not */ /* profitable to do so */ CCM_UNPAR_NEST, /* Loop below not parallelized because it was */ /* nested in a parallel loop */ CCM_UNPAR, /* Loop below not parallelized */ CCM_UNPAR_NOAUTO, /* Loop below not parallelized because */ /* autoparallelization is not enabled */ CCM_PR_L_VAR, /* Private variables in loop below: */ /* , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_SH_L_VAR, /* Shared variables in loop below: */ /* , , ... */ CCM_TP_L_VAR, /* Threadprivate variables in loop below: */ /* , , ... */ CCM_RV_L_VAR, /* Reduction variables in loop below: */ /* , , ... */ CCM_IM_L_VAR, /* Implicit variables in loop below: */ /* , , ... */ CCM_PR_O_VAR, /* Private variables in OpenMP construct below: */ /* , , ... */ CCM_SH_O_VAR, /* Shared variables in OpenMP construct below: */ /* , , ... */ CCM_TP_O_VAR, /* Threadprivate variables in OpenMP construct */ /* below: , , ... */ CCM_RV_O_VAR, /* Reduction variables in OpenMP construct below: */ /* , , ... */ CCM_IM_O_VAR, /* Implicit variables in OpenMP construct below: */ /* , , ... */ CCM_UNPAR_IN_OMP, /* Loop below not parallelized because it is inside */ /* an OpenMP region */ CCM_FP_O_VAR, /* Firstprivate variables in OpenMP construct below: */ /* , , ... */ CCM_LP_O_VAR, /* Lastprivate variables in OpenMP construct below: */ /* , , ... */ CCM_CP_O_VAR, /* Copyprivate variables in OpenMP construct below: */ /* , , ... */ CCM_PR_OAS_VAR, /* Variables autoscoped as PRIVATE in OpenMP */ /* construct below: , , ... */ CCM_SH_OAS_VAR, /* Variables autoscoped as SHARED in OpenMP */ /* construct below: , , ... */ CCM_FP_OAS_VAR, /* Variables autoscoped as FIRSTPRIVATE in OpenMP */ /* construct below: , , ... */ CCM_LP_OAS_VAR, /* Variables autoscoped as LASTPRIVATE in OpenMP */ /* construct below: , , ... */ CCM_RV_OAS_VAR, /* Variables autoscoped as REDUCTION in OpenMP */ /* construct below: , , ... */ CCM_FAIL_OAS_VAR, /* Variables cannot be autoscoped in OpenMP */ /* construct below: , , ... */ CCM_SERIALIZE_OAS, /* OpenMP parallel region below is serialized */ /* because autoscoping has failed */ CCM_UNPAR_CALL_2, /* not parallelized because it contains calls */ /* to: , , ... */ CCM_PAR_DRECTV_2, /* parallelized by explicit user directive */ CCM_APAR_2, /* autoparallelized */ CCM_AUTOPAR_2, /* autoparallelized; equivalent */ /* explict directive is */ CCM_UNPAR_DD_2, /* could not be parallelized because of */ /* data dependences on: , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_UNPAR_DDA_2, /* could not be parallelized because of a */ /* data dependence or aliasing of: , , ... */ CCM_UNPAR_ANONDD_2, /* could not be parallelized because of an */ /* anonymous data dependence */ CCM_UNPAR_ANONDDA_2, /* could not be parallelized because of an */ /* anonymous data dependence or aliasing */ CCM_PAR_WORK_2, /* parallelized, but might not contain */ /* enough work to run efficiently in parallel */ CCM_UNPAR_EXIT_2, /* not parallelized because it contains */ /* multiple exit points */ CCM_UNPAR_STRANGE_2, /* not parallelized because it contains a */ /* strange flow of control */ CCM_UNPAR_IO_2, /* not parallelized because it contains */ /* I/O or other MT-unsafe calls */ CCM_PAR_BODY_NAME_2, /* parallel loop-body code placed in */ /* function along with inner loops */ CCM_UNPAR_NLOOPIDX_2, /* not parallelized because loop index not */ /* found */ CCM_UNPAR_DRECTV_2, /* not parallelized because of explicit */ /* user directive */ CCM_UNPAR_NOTPROFIT_2, /* not parallelized because it was not */ /* profitable to do so */ CCM_UNPAR_NEST_2, /* not parallelized because it was */ /* nested within a parallel loop */ CCM_UNPAR_2, /* not parallelized */ CCM_UNPAR_NOAUTO_2, /* not parallelized because */ /* autoparallelization is not enabled */ CCM_PR_L_VAR_2, /* Private variables in : */ /* , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_SH_L_VAR_2, /* Shared variables in : */ /* , , ... */ CCM_TP_L_VAR_2, /* Threadprivate variables in : */ /* , , ... */ CCM_RV_L_VAR_2, /* Reduction variables of operator in : */ /* , , ... */ CCM_IM_L_VAR_2, /* Implicit variables in : */ /* , , ... */ CCM_PR_O_VAR_2, /* Private variables in : */ /* , , ... */ CCM_SH_O_VAR_2, /* Shared variables in : */ /* , , ... */ CCM_TP_O_VAR_2, /* Threadprivate variables in : */ /* , , ... */ CCM_RV_O_VAR_2, /* Reduction variables of operator in : */ /* , , ... */ CCM_IM_O_VAR_2, /* Implicit variables in : */ /* , , ... */ CCM_UNPAR_IN_OMP_2, /* not parallelized because it is inside */ /* OpenMP region */ CCM_FP_O_VAR_2, /* Firstprivate variables in : */ /* , , ... */ CCM_LP_O_VAR_2, /* Lastprivate variables in : */ /* , , ... */ CCM_CP_O_VAR_2, /* Copyprivate variables in : */ /* , , ... */ CCM_PR_OAS_VAR_2, /* Variables autoscoped as PRIVATE in : */ /* , , ... */ CCM_SH_OAS_VAR_2, /* Variables autoscoped as SHARED in : */ /* , , ... */ CCM_FP_OAS_VAR_2, /* Variables autoscoped as FIRSTPRIVATE in : */ /* , , ... */ CCM_LP_OAS_VAR_2, /* Variables autoscoped as LASTPRIVATE in : */ /* , , ... */ CCM_RV_OAS_VAR_2, /* Variables autoscoped as REDUCTION of operator */ /* in : , , ... */ CCM_FAIL_OAS_VAR_2, /* Variables treated as shared because they cannot */ /* be autoscoped in : , , ... */ CCM_SERIALIZE_OAS_2, /* will be executed by a single thread because */ /* autoscoping for some variables was not successful */ /* Group: Parallelization Questions asked of the user */ /* How will the user answer these questions? */ CCM_QPERMVEC=0x00800, /* Is a permutation vector during execution of */ /* ? */ CCM_QEXPR, /* Is expression true for ? */ CCM_QSAFECALL, /* Is subroutine MP-safe as used in ? */ /* Group: Loop Optimization Messages */ CCM_LCOST=0x01000, /* Loop below estimated to cost cycles per */ /* iteration */ CCM_UNROLL, /* Loop below unrolled times */ /* CCMV_WANT: the next one should be replaced by CCM_IMIX2 */ CCM_IMIX, /* Loop below has loads, stores, */ /* prefetches, FPadds, FPmuls, and */ /* FPdivs per iteration */ CCM_SPILLS, /* Loop below required integer register spills, */ /* FP register spills, and used */ /* integer registers and FP registers */ CCM_LFISSION, /* Loop below fissioned into loops */ CCM_LPEEL, /* Loop below had iterations peeled off for better */ /* unrolling and/or parallelization */ CCM_LBLOCKED, /* Loop below blocked by for improved cache */ /* performance */ CCM_LTILED, /* Loop below tiled for better performance */ CCM_LUNRJAM, /* Loop below unrolled and jammed */ CCM_LWHILE2DO, /* Bounds test for loop below moved to top of loop */ CCM_L2CALL, /* Loop below replaced by a call to */ CCM_LDEAD, /* Loop below deleted as dead code */ CCM_LINTRCHNG, /* Loop below interchanged with loop on line */ CCM_FUSEDTO, /* Loop below fused with loop on line */ CCM_FUSEDFROM, /* Loop from line fused with loop below */ CCM_VECINTRNSC, /* Loop below transformed to use calls to vector */ /* intrinsic , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_LSTRIPMINE, /* Loop below strip-mined */ CCM_LNEST2LOOPS, /* Loop below collapsed with loop on line */ CCM_LREVERSE, /* Loop below has had its iteration direction */ /* reversed */ CCM_IMIX2, /* Loop below has loads, stores, */ /* prefetches, FPadds, FPmuls, */ /* FPdivs, FPsubs, and FPsqrts per */ /* iteration */ CCM_LUNRFULL, /* Loop below fully unrolled */ CCM_ELIM_NOAMORTINST, /* Loop below was eliminated as it contains no */ /* non-amortizable instructions */ CCM_COMP_DALIGN, /* Performance of loop below could be improved */ /* by compiling with -dalign */ CCM_INTIMIX, /* Loop below has int-loads, int-stores, */ /* alu-ops, muls, int-divs and */ /* shifts per iteration */ CCM_LMULTI_VERSION, /* multi-versioned. Specialized version */ /* is */ CCM_LCOST_2, /* estimated to cost cycles per iteration */ CCM_UNROLL_2, /* unrolled times */ /* CCMV_WANT: the next one should be replaced by CCM_IMIX2_B or CCM_IMIX3_B */ CCM_IMIX_B, /* has loads, stores, */ /* prefetches, FPadds, FPmuls, and */ /* FPdivs per iteration */ CCM_SPILLS_2, /* required integer register spills, */ /* FP register spills, and used */ /* integer registers and FP registers */ CCM_LFISSION_2, /* fissioned into loops, generating: */ /* , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_LFISSION_FRAG, /* contains code from lines: , , ... */ CCM_LPEEL_2, /* had iterations peeled off for better */ /* unrolling and/or parallelization */ CCM_LBLOCKED_2, /* blocked by for improved memory */ /* hierarchy performance, new inner loop */ CCM_LOUTER_UNROLL, /* is outer-unrolled times as part */ /* of unroll and jam */ CCM_LJAMMED, /* All copies of are fused together */ /* as part of unroll and jam */ CCM_LWHILE2DO_2, /* Bounds test for moved to top of loop */ CCM_L2CALL_2, /* replaced by a call to */ CCM_LDEAD_2, /* deleted as dead code */ CCM_LINTRCHNG_2, /* interchanged with */ CCM_LINTRCHNG_ORDER, /* For loop nest below, the final order of loops */ /* after interchanging and subsequent */ /* transformations is: , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_FUSED_2, /* fused with , new loop */ CCM_VECINTRNSC_2, /* transformed to use calls to vector */ /* intrinsics: , , ... */ CCM_LSTRIPMINE_2, /* strip-mined by , new inner loop */ CCM_LNEST2LOOPS_2, /* collapsed with , new loop */ CCM_LREVERSE_2, /* has had its iteration direction reversed */ CCM_IMIX2_B, /* has loads, stores, */ /* prefetches, FPadds, FPmuls, */ /* FPdivs, FPsubs, and FPsqrts per */ /* iteration */ CCM_LUNRFULL_2, /* fully unrolled */ CCM_ELIM_NOAMORTINST_2, /* was eliminated as it contains no */ /* non-amortizable instructions */ CCM_COMP_DALIGN_2, /* Performance of could be improved by */ /* compiling with -dalign */ CCM_INTIMIX_2, /* has int-loads, int-stores, */ /* alu-ops, muls, int-divs and */ /* shifts per iteration */ CCM_OMP_REGION, /* Source OpenMP region below has tag */ CCM_LMICROVECTORIZE, /* is micro-vectorized */ CCM_LMULTI_VERSION_2, /* multi-versioned for . */ /* Specialized version is */ CCM_LCLONED, /* cloned for . Clone is */ CCM_LUNSWITCHED, /* is unswitched. New loops */ /* are and */ CCM_LRESWITCHED, /* Loops and and their surrounding */ /* conditional code have been merged to */ /* form loop */ CCM_LSKEWBLOCKED, /* skew-blocked by with slope */ /* for improved memory hierarchy */ /* performance, new inner loop */ CCM_IVSUB, /* Induction variable substitution performed on */ CCM_ONEITER_REPLACED, /* determined to have a trip count of 1; */ /* converted to straight-line code */ CCM_IMIX3_B, /* has loads, stores, */ /* prefetches, FPadds, FPmuls, */ /* FPmuladds, FPdivs, and FPsqrts per */ /* iteration */ /* Group: Pipelining Messages */ CCM_PIPELINE=0x02000, /* Loop below pipelined */ CCM_PIPESTATS, /* Loop below scheduled with steady-state cycle */ /* count = */ CCM_NOPIPE_CALL, /* Loop could not be pipelined because it contains */ /* calls */ CCM_NOPIPE_INTCC, /* Loop could not be pipelined because it sets */ /* multiple integer condition codes. */ CCM_NOPIPE_MBAR, /* Loop could not be pipelined because it contains a */ /* memory barrier instruction */ CCM_NOPIPE_MNMX, /* Loop could not be pipelined because it contains */ /* a minimum or a maximum operation */ CCM_NOPIPE_U2FLT, /* Loop could not be pipelined because it contains */ /* an unsigned to float conversion */ CCM_NOPIPE_GOT, /* Loop could not be pipelined because it sets the */ /* Global Offset Table pointer */ CCM_NOPIPE_IDIV, /* Loop could not be pipelined because it contains */ /* an integer divide */ CCM_NOPIPE_PRFTCH, /* Loop could not be pipelined because it contains */ /* a prefetch operation */ CCM_NOPIPE_EXIT, /* Loop could not be pipelined because it contains */ /* an exit operation */ CCM_NOPIPE_REG, /* Loop could not be pipelined because it contains */ /* instructions that set the %gsr or %fsr register */ CCM_NOPIPE_UNS, /* Loop could not be pipelined because it has an */ /* unsigned loop counter */ CCM_NOPIPE_UNSUIT, /* Loop was unsuitable for pipelining */ CCM_NOPIPE_INTRINSIC, /* Loop could not be pipelined because it has an */ /* intrinsic call to */ CCM_NOPIPE_BIG, /* Loop could not be pipelined as it is too big */ CCM_NOPIPE_INVINTPR, /* Loop could not be pipelined as it contains too */ /* many loop invariant integers = */ CCM_NOPIPE_INVFLTPR, /* Loop could not be pipelined as it contains too */ /* many loop invariant floats = */ CCM_NOPIPE_INVDBLPR, /* Loop could not be pipelined as it contains too */ /* many loop invariant doubles = */ CCM_PIPE_SCHEDAFIPR, /* Loop below was adversely affected by high */ /* integer register pressure = */ CCM_PIPE_SCHEDAFDPR, /* Loop below was adversely affected by high */ /* double register pressure = */ CCM_PIPE_SCHEDAFFPR, /* Loop below was adversely affected by high */ /* float register pressure = */ CCM_NOPIPE_INTPR, /* Loop could not be pipelined due to high */ /* integer register pressure = */ CCM_NOPIPE_DBLPR, /* Loop could not be pipelined due to high */ /* double register pressure = */ CCM_NOPIPE_FLTPR, /* Loop could not be pipelined due to high */ /* float register pressure = */ CCM_PIPELINE_2, /* pipelined */ CCM_PIPESTATS_2, /* scheduled with steady-state cycle */ /* count = */ CCM_NOPIPE_CALL_2, /* could not be pipelined because it contains */ /* calls */ CCM_NOPIPE_INTCC_2, /* could not be pipelined because it sets */ /* multiple integer condition codes. */ CCM_NOPIPE_MBAR_2, /* could not be pipelined because it contains */ /* a memory barrier instruction */ CCM_NOPIPE_MNMX_2, /* could not be pipelined because it contains */ /* a minimum or a maximum operation */ CCM_NOPIPE_U2FLT_2, /* could not be pipelined because it contains */ /* an unsigned to float conversion */ CCM_NOPIPE_GOT_2, /* could not be pipelined because it sets the */ /* Global Offset Table pointer */ CCM_NOPIPE_IDIV_2, /* could not be pipelined because it contains */ /* an integer divide */ CCM_NOPIPE_PRFTCH_2, /* could not be pipelined because it contains */ /* a prefetch operation */ CCM_NOPIPE_EXIT_2, /* could not be pipelined because it contains */ /* an exit operation */ CCM_NOPIPE_REG_2, /* could not be pipelined because it contains */ /* instructions that set the %gsr or %fsr register */ CCM_NOPIPE_UNS_2, /* could not be pipelined because it has an */ /* unsigned loop counter */ CCM_NOPIPE_UNSUIT_2, /* is unsuitable for pipelining */ CCM_NOPIPE_INTRINSIC_2, /* could not be pipelined because it contains */ /* a call to intrinsic */ CCM_NOPIPE_BIG_2, /* could not be pipelined as it is too big */ CCM_NOPIPE_INVINTPR_2, /* could not be pipelined as it contains too */ /* many loop invariant integers = */ CCM_NOPIPE_INVFLTPR_2, /* could not be pipelined as it contains too */ /* many loop invariant floats = */ CCM_NOPIPE_INVDBLPR_2, /* could not be pipelined as it contains too */ /* many loop invariant doubles = */ CCM_PIPE_SCHEDAFIPR_2, /* was adversely affected by high */ /* integer register pressure = */ CCM_PIPE_SCHEDAFDPR_2, /* was adversely affected by high */ /* double register pressure = */ CCM_PIPE_SCHEDAFFPR_2, /* was adversely affected by high */ /* float register pressure = */ CCM_NOPIPE_INTPR_2, /* could not be pipelined due to high */ /* integer register pressure = */ CCM_NOPIPE_DBLPR_2, /* could not be pipelined due to high */ /* double register pressure = */ CCM_NOPIPE_FLTPR_2, /* could not be pipelined due to high */ /* float register pressure = */ /* Group: Inlining Messages */ CCM_INLINE=0x04000, /* Function inlined from source file into */ /* the code for the following line */ CCM_INLINE2, /* Function inlined from source file into */ /* inline copy of function */ CCM_INLINE_TMPLT, /* Function inlined from template file */ /* into the code for the following line */ CCM_INLINE_TMPLT2, /* Function inlined from template file */ /* into inline copy of function */ CCM_INLINE_OUT_COPY, /* Out-of-line copy of inlined function from */ /* source file generated */ CCM_NINLINE_REC, /* Recursive function inlined only up to */ /* depth */ CCM_NINLINE_NEST, /* Function not inlined because inlining is */ /* already nested too deeply */ CCM_NINLINE_CMPLX, /* Function not inlined because it contains */ /* too many operations */ CCM_NINLINE_FB, /* Function not inlined because the */ /* profile-feedback execution count is too low */ CCM_NINLINE_PAR, /* Function not inlined because it contains */ /* explicit parallel pragmas */ CCM_NINLINE_OPT, /* Function not inlined because it is */ /* compiled with optimization level <= 2 */ CCM_NINLINE_USR, /* Function not inlined because either command */ /* line option or source code pragma prohibited it, */ /* or it's not safe to inline it */ CCM_NINLINE_AUTO, /* Function not inlined because doing so */ /* would make automatic storage for too large */ CCM_NINLINE_CALLS, /* Function not inlined because it contains */ /* too many calls */ CCM_NINLINE_ACTUAL, /* Function not inlined because it has more */ /* actual parameters than formal parameters */ CCM_NINLINE_FORMAL, /* Function not inlined because it has more */ /* formal parameters than actual parameters */ CCM_NINLINE_TYPE, /* Function not inlined because formal */ /* argument type does not match actual type */ CCM_NINLINE_ATYPE, /* Function not inlined because array formal */ /* argument does not match reshaped array actual */ /* argument type */ CCM_NINLINE_RETTYPE, /* Function not inlined because return type */ /* does not match */ CCM_NINLINE_EXCPT, /* Function not inlined because it */ /* guarded by an exception handler */ CCM_NINLINE_UNSAFE, /* Function not inlined because it might be */ /* unsafe (call alloca(), etc) */ CCM_NINLINE_ALIAS, /* Function not inlined because inlining it */ /* will make the alias analysis in the calling */ /* function more conservative */ CCM_NINLINE_FEMARK, /* Function not inlined because it contains */ /* setjmp/longjmp, or indirect goto, etc */ CCM_NINLINE_RAREX, /* Function not inlined because it is known */ /* to be rarely executed */ CCM_CLONING, /* Function from source file cloned, */ /* creating cloned function ; constant */ /* parameters propagated to clone */ CCM_INLINE_B, /* Function inlined from source file into */ /* the code for the following line. loops */ /* inlined */ CCM_INLINE2_B, /* Function inlined from source file into */ /* inline copy of function . loops inlined */ CCM_INLINE_LOOP, /* Loop in function , line has */ /* tag */ CCM_NINLINE_MULTIENTRY, /* Function not inlined because it */ /* contains an ENTRY statement */ CCM_NINLINE_VARARGS, /* Function not inlined because variable */ /* argument routines cannot be inlined */ CCM_NINLINE_UNSEEN_BODY, /* Function not inlined because the compiler */ /* has not seen the body of the function. Use */ /* -xcrossfile or -xipo in order to inline it */ CCM_NINLINE_UPLEVEL, /* Function not inlined because it is a */ /* nested routine containing references to */ /* variables defined in an outer function */ CCM_NINLINE_CMDLINE, /* Function not inlined because either */ /* -xinline or source code pragma prohibited it */ CCM_NINLINE_CALL_CMPLX, /* Call to not inlined because of the */ /* complexity of the calling routine */ CCM_NINLINE_LANG_MISMATCH, /* Call to not inlined because it is in */ /* a different language */ CCM_NINLINE_RTN_WEAK, /* Function not inlined because it */ /* is marked weak */ CCM_NINLINE_CALL_WEAKFILE, /* Call to not inlined because it is */ /* in a different file and it contains a */ /* call to a weak routine */ CCM_NINLINE_CALL_TRYCATCH, /* Call to not inlined because it is */ /* in a different file and contains an */ /* explicit try/catch */ CCM_NINLINE_CALL_REGP, /* Call to not inlined because it would */ /* cause excessive register pressure */ CCM_NINLINE_RTN_REGP, /* Function not inlined because it would */ /* cause excessive register pressure */ CCM_NINLINE_CALL_XPENSV, /* Call to not inlined because analysis */ /* exceeds the compilation time limit */ CCM_NINLINE_READONLYIR, /* Function not inlined because it is in a file */ /* specified as read-only by -xipo_archive=readonly */ /* and it contains calls to static functions */ CCM_NINLINE_CALL_THUNK, /* Call to not inlined because it is in a */ /* compiler-generated function that does not */ /* permit inlining */ CCM_NINLINE_CALL_XTARGETS, /* Indirect callsite has too many targets; */ /* callsite marked do not inline */ CCM_NINLINE_SELFTAIL_RECURSIVE, /* Function not inlined because */ /* of a recursive tail-call to itself */ CCM_NINLINE_PRAGMA, /* Function not inlined because it contains */ /* explicit parallel or alias pragmas */ CCM_NINLINE_CMPLX2, /* Function not inlined because it contains too */ /* many operations. Increase max_inst_hard in order */ /* to inline it: -xinline_param=max_inst_hard:n */ CCM_NINLINE_RARE, /* Function not inlined because the call */ /* is rarely executed */ CCM_NINLINE_PAR2, /* Function not inlined because it is called */ /* within a region guarded by an explicit */ /* parallel pragmas */ CCM_NINLINE_G_LIMIT, /* Function not inlined because it would exceed */ /* the permitted global code size growth limit. Try */ /* to increase max_growth in order to inline it: */ /* -xinline_param=max_growth:n */ CCM_NINLINE_L_LIMIT, /* Function not inlined because it would exceed */ /* the maximum function size growth limit. Increase */ /* max_function_inst in order to inline it: */ /* -xinline_param=max_function_inst:n */ CCM_NINLINE_REC2, /* Recursive function is inlined only up to */ /* levels and up to size. Increase */ /* max_recursive_deptha or max_recursive_inst in */ /* order to inline it: */ /* -xinline_param=max_recursive_depth:n, */ /* -xinline_param=max_recursive_inst:n */ CCM_NINLINE_FB2, /* Function not inlined because the */ /* profile-feedback execution count is too */ /* low. Decrease min_counter in order to inline it: */ /* -xinline_param:min_counter:n */ CCM_NINLINE_CS_CMPLX, /* Function not inlined because called */ /* function's size is too big. Increase */ /* max_inst_soft in order to inline it: */ /* -xinline_param=max_inst_soft:n */ CCM_NINLINE_R_EXCPT, /* Function not inlined because it contains */ /* an exception handler */ CCM_NINLINE_ASM, /* Function not inlined because */ /* it contains asm statements */ CCM_NINLINE_R_READONLYIR, /* Function not inlined because it is in a file */ /* specified as read-only by -xipo_archive=readonly */ /* and it is a static function */ CCM_NINLINE_C_READONLYIR, /* Call to not inlined because the calling */ /* function is in a file specified as read-only */ /* by -xipo_archive=readonly */ CCM_NINLINE_NEVERRETURN, /* Function not inlined because it */ /* never returns */ /* Group: Messages Concerning Memory Operations */ /* Notes: */ /* a. In all of these, is a string that is something like */ /* "A(i+5*k)" or "structure.field", giving the high-level */ /* construct that is being loaded or stored. */ /* */ /* b. In all of these, refers to an instruction offset, */ /* expressed as a 32-bit signed integer. It is assumed */ /* that any prefetches will be within this range of the */ /* load/store they are prefetching for. */ CCM_MPREFETCH=0x08000, /* Prefetch of inserted */ /* [This message has a lineno for the source, */ /* but no instaddr for the disassembly.] */ CCM_MPREFETCH_LD, /* Prefetch of inserted for load at */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MPREFETCH_ST, /* Prefetch of inserted for store at */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MPREFETCH_FB, /* Prefetch of inserted based on feedback data */ /* [This message has a lineno for the source, */ /* but no instaddr for the disassembly.] */ CCM_MPREFETCH_FB_LD, /* Prefetch of inserted for load at based */ /* on feedback data */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MPREFETCH_FB_ST, /* Prefetch of inserted for store at based */ /* on feedback data */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MLOAD, /* Load below refers to */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MSTORE, /* Store below refers to */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MLOAD_P, /* Load below refers to , and was prefetched */ /* at */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ CCM_MSTORE_P, /* Store below refers to , and was prefetched */ /* at */ /* [This message has lineno = -1, */ /* and is for disassembly only] */ /* Group: Front-end messages [all compilers] */ /* Group: F95 Front-end Messages */ CCM_COPYIN=0x10000, /* Parameter caused a copyin in the following */ /* call */ CCM_COPYOUT, /* Parameter caused a copyout in the following */ /* call */ CCM_COPYINOUT, /* Parameter caused both a copyin and copyout */ /* in the following call */ CCM_PADDING, /* Padding of bytes inserted before */ /* array */ CCM_PADCOMMON, /* Padding of bytes inserted before */ /* array in common block */ CCM_ALIGN_EQ, /* Variable/array can not be double-aligned, */ /* because it is equivalenced */ CCM_ALIGN_PERF, /* Alignment of variables in common block may cause */ /* performance degradation */ CCM_ALIGN_STRUCT, /* Alignment of component in numeric sequence */ /* structure may cause performance degradation */ CCM_TMP_COPY, /* Argument copied to a temporary */ CCM_TMP_COPYM, /* Argument might be copied to a temporary; */ /* runtime decision made */ CCM_PROC_MISMATCH, /* Argument to subprogram differs from */ /* reference on line */ CCM_PROC_MISMATCH2, /* Scalar argument to subprogram is */ /* referred to as an array on line */ CCM_PROC_MISMATCH3, /* Return type/rank from subprogram differs */ /* from return on line */ CCM_DO_EXPR, /* DO statement bounds lead to no executions of the */ /* loop */ CCM_AUTO_BND, /* The bounds for automatic variable are not */ /* available at all entry points; zero-length */ /* variable might be allocated */ CCM_LIT_PAD, /* The character string literal padded */ /* to the length specified for the dummy argument */ CCM_ARRAY_LOOP, /* Array statement below generated a loop */ CCM_ARRAY_LOOPNEST, /* Array statement below generated nested loops */ CCM_ALIGN_PERF2, /* Alignment of variable in common block */ /* may cause a performance degradation */ CCM_ALIGN_PERF3, /* Alignment of variable in blank common may */ /* cause a performance degradation */ CCM_IO_LOOP_ARRAY, /* I/O implied do item below generated an array */ /* section */ /* Group: C++ Front-end Messages */ CCM_TMPCONST, /* Implicit invocation of class constructor for */ /* temporary */ CCM_TMPDEST, /* Implicit invocation of class destructor for */ /* temporary */ CCM_DBL_CONST, /* Double constant used in float expression */ CCM_MINLINE, /* Function inlined from source file by */ /* front-end */ /* [This refers to front-end inlining, */ /* not the backend inlining above.] */ CCM_MINLINE2, /* Function from source file inlined into */ /* inline copy of method by front-end */ /* [This refers to front-end inlining, */ /* not the backend inlining above.] */ CCM_MINLINE3, /* Function not inlined because it uses keyword */ /* */ CCM_MINLINE4, /* Function not inlined because it is too */ /* complex */ CCM_TMP_COPYOUT, /* Argument copied from a temporary */ CCM_TMP_COPYOUTM, /* Argument might be copied from a temporary; */ /* runtime decision made */ CCM_TMP_COPYINOUT, /* Argument copied in and out of a temporary */ CCM_TMP_COPYINOUTM, /* Argument might be copied in and out of */ /* a temporary; runtime decision made */ /* Group: C Front-end Messages */ /* Group: NJC Front-end Messages */ /* Group: Updated F95 Front-end Messages */ CCM_ARRAY_LOOP_2, /* Array statement below generated loop */ CCM_ARRAY_LOOPNEST_2, /* Array statement below generated nested */ /* loops: , , ... */ /* [The number of parameters will determine how many */ /* names appear, and the formatter will get the */ /* commas right.] */ CCM_IO_LOOP_ARRAY_2, /* I/O implied do item below generated an array */ /* section: */ CCM_USER_LOOP, /* Source loop below has tag */ CCM_FOUND_LOOP, /* Discovered loop below has tag */ CCM_MFUNCTION_LOOP, /* Copy in M-function of loop below has tag */ /* Group: Code-generator Messages */ CCM_FSIMPLE=0x20000, /* Transformations for fsimple= applied */ CCM_STACK, /* Function requires Mbytes of stack */ /* storage */ CCM_TAILRECUR, /* Recursive tail call in optimized to jump to */ /* entry point */ CCM_TAILCALL, /* Call to function was tail-call optimized */ CCM_NI_EXIT_OR_PSEUDO, /* Template could not be early inlined because it */ /* contains the pseudo instruction */ CCM_NI_BAD_UNARY_OPC, /* Template could not be early inlined because it */ /* contains the instruction opcode */ CCM_NI_INT_LDD_ON_V9, /* Template could not be early inlined because it */ /* contains integer ldd instructions, which are */ /* deprecated in the v9 architecture */ CCM_NI_LATE_INL_OPC, /* Template could not be early inlined because it */ /* contains the instruction opcode */ CCM_NI_BAD_IMM_OP, /* Template could not be early inlined because the */ /* relocation or immediate operand is not well */ /* understood by the optimizer */ CCM_NI_BAD_STATELEAF, /* Template could not be early inlined because it */ /* references the state register */ CCM_NI_BAD_ASR_19, /* Template could not be early inlined because */ /* %asr19 is not supported in pre v8plus code */ CCM_NI_BAD_FSR_USE, /* Template could not be early inlined because */ /* references to %fsr can only be optimized when the */ /* -iaopts flag is used */ CCM_NI_BAD_REGISTER, /* Template could not be early inlined because it */ /* references the register */ CCM_NI_NO_RET_VAL, /* Template could not be early inlined because it */ /* does not return the value declared */ CCM_NI_DELAY, /* Template could not be early inlined because it */ /* contains a non nop delay slot */ CCM_NI_SCALL, /* Template could not be early inlined because it */ /* calls a function which returns a structure */ CCM_CASE_POSITION, /* Case block below was placed at position */ /* based on execution frequency */ CCM_CALL_WITH_CODE, /* Call to replaced with inline code. */ /* loops created: , , ... */ CCM_NI_BAD_SP_ADDR, /* Template could not be early inlined because it */ /* contains a %sp+reg address */ CCM_NI_BAD_SP_USAGE, /* Template could not be early inlined because it */ /* uses/defines the stack pointer in a non-load/store instruction */ CCM_NI_MIXED_REG_TYPES, /* Template could not be early inlined because it */ /* contains register used as both x-register and register pair */ CCM_LAST } COMPMSG_ID; /* * The Message Structure * Each message is a fixed-length structure as follows: */ typedef struct { int64_t instaddr; /* the PC offset, relative to the .o .text section */ int32_t lineno; /* the source line to which it refers */ COMPMSG_ID msg_type; /* the specific message index */ int32_t nparam; /* number of parameters to this message */ int32_t param_index; /* the index of the first parameter */ } compmsg; #if defined(__cplusplus) extern "C" { #endif /* * Initializes the data structures, converts the source name to a string, * and fills in srcname and version in the header */ void compcom_p_open (char *srcname, int32_t version); /* * Finds or enters the string s into the string table, and returns the index * of the string */ int32_t compcom_p_string (char *s); /* * Enter the single message. Any string parameters should have been converted * to int32's by calling compcom_p_string() */ void compcom_p_putmsg (int32_t show_bits, int64_t pcoffset, int32_t lineno, COMPMSG_ID m, int32_t nparams); /* * Whatever is needed to close the section and write it out to the .o */ void compcom_p_finalize (); #if defined(__cplusplus) } #endif #endif /* _COMP_COM_H */