diff options
author | dgp <dgp@users.sourceforge.net> | 2013-07-18 20:27:01 (GMT) |
---|---|---|
committer | dgp <dgp@users.sourceforge.net> | 2013-07-18 20:27:01 (GMT) |
commit | 17ebc3cd20794e0246b5c6b5f86b128cf1e7b69b (patch) | |
tree | 054d9ea09ad9e162f640d47306181917d0f14df2 | |
parent | 81a1a91361cc615d3e9bf5b9016bd12cd258aa8d (diff) | |
parent | 5e9e003379a72c3459d97b1931c3e8b8627541f8 (diff) | |
download | tcl-17ebc3cd20794e0246b5c6b5f86b128cf1e7b69b.zip tcl-17ebc3cd20794e0246b5c6b5f86b128cf1e7b69b.tar.gz tcl-17ebc3cd20794e0246b5c6b5f86b128cf1e7b69b.tar.bz2 |
Rewrite of the TclCompileScript() routine.
Primarily this breaks that large, rather convoluted routine into many
smaller more single-purpose routines for greater clarity. This also
permits more sharing of function components. Ensemble compiles now share
the same low level routines as top-level command compiles. This refactoring
should also have the benefit of easing merging between branches by better
locating patches according to what function or data structure they tweak.
Also in this checkin:
Minimized calls to Tcl_ResetResult() during compile.
Several commands compile their detected syntax errors into
bytecode that reports the syntax error.
Significant reform to the TIP 280 machinery, including elimination
of the litInfo hash table. Its function of finding the index
of a compiled command in a script from a pc value has been redone
using the codeDelta information already stored in ByteCode.
-rw-r--r-- | generic/tclAssembly.c | 30 | ||||
-rw-r--r-- | generic/tclBasic.c | 122 | ||||
-rw-r--r-- | generic/tclCompCmds.c | 5 | ||||
-rw-r--r-- | generic/tclCompCmdsGR.c | 1 | ||||
-rw-r--r-- | generic/tclCompCmdsSZ.c | 3 | ||||
-rw-r--r-- | generic/tclCompile.c | 521 | ||||
-rw-r--r-- | generic/tclCompile.h | 13 | ||||
-rw-r--r-- | generic/tclEnsemble.c | 122 | ||||
-rw-r--r-- | generic/tclExecute.c | 62 | ||||
-rw-r--r-- | generic/tclInt.h | 2 |
10 files changed, 654 insertions, 227 deletions
diff --git a/generic/tclAssembly.c b/generic/tclAssembly.c index 416ee10..9b9b6f8 100644 --- a/generic/tclAssembly.c +++ b/generic/tclAssembly.c @@ -930,6 +930,12 @@ TclCompileAssembleCmd( { Tcl_Token *tokenPtr; /* Token in the input script */ +#if 1 + int numCommands = envPtr->numCommands; + int offset = envPtr->codeNext - envPtr->codeStart; + int depth = envPtr->currStackDepth; +#endif + /* * Make sure that the command has a single arg that is a simple word. */ @@ -943,10 +949,28 @@ TclCompileAssembleCmd( } /* - * Compile the code and return any error from the compilation. + * Compile the code and convert any error from the compilation into + * bytecode reporting the error; */ - return TclAssembleCode(envPtr, tokenPtr[1].start, tokenPtr[1].size, 0); + if (TCL_ERROR == TclAssembleCode(envPtr, tokenPtr[1].start, + tokenPtr[1].size, TCL_EVAL_DIRECT)) { + +#if 1 + Tcl_AppendObjToErrorInfo(interp, Tcl_ObjPrintf( + "\n (\"%.*s\" body, line %d)", + parsePtr->tokenPtr->size, parsePtr->tokenPtr->start, + Tcl_GetErrorLine(interp))); + envPtr->numCommands = numCommands; + envPtr->codeNext = envPtr->codeStart + offset; + envPtr->currStackDepth = depth; + TclCompileSyntaxError(interp, envPtr); +#else + Tcl_ResetResult(interp); + return TCL_ERROR; +#endif + } + return TCL_OK; } /* @@ -1108,7 +1132,7 @@ NewAssemblyEnv( assemEnvPtr->envPtr = envPtr; assemEnvPtr->parsePtr = parsePtr; - assemEnvPtr->cmdLine = envPtr->line; + assemEnvPtr->cmdLine = 1; assemEnvPtr->clNext = envPtr->clNext; /* diff --git a/generic/tclBasic.c b/generic/tclBasic.c index 1d35034..82affb0 100644 --- a/generic/tclBasic.c +++ b/generic/tclBasic.c @@ -1609,8 +1609,6 @@ DeleteInterpProc( ckfree(eclPtr->loc); } - Tcl_DeleteHashTable(&eclPtr->litInfo); - ckfree(eclPtr); Tcl_DeleteHashEntry(hPtr); } @@ -5617,76 +5615,88 @@ TclArgumentBCEnter( int objc, void *codePtr, CmdFrame *cfPtr, + int cmd, int pc) { + ExtCmdLoc *eclPtr; + int word; + ECL *ePtr; + CFWordBC *lastPtr = NULL; Interp *iPtr = (Interp *) interp; Tcl_HashEntry *hePtr = Tcl_FindHashEntry(iPtr->lineBCPtr, (char *) codePtr); - ExtCmdLoc *eclPtr; if (!hePtr) { return; } eclPtr = Tcl_GetHashValue(hePtr); - hePtr = Tcl_FindHashEntry(&eclPtr->litInfo, INT2PTR(pc)); - if (hePtr) { - int word; - int cmd = PTR2INT(Tcl_GetHashValue(hePtr)); - ECL *ePtr = &eclPtr->loc[cmd]; - CFWordBC *lastPtr = NULL; + ePtr = &eclPtr->loc[cmd]; - /* - * A few truths ... - * (1) ePtr->nline == objc - * (2) (ePtr->line[word] < 0) => !literal, for all words - * (3) (word == 0) => !literal - * - * Item (2) is why we can use objv to get the literals, and do not - * have to save them at compile time. - */ + /* + * ePtr->nline is the number of words originally parsed. + * + * objc is the number of elements getting invoked. + * + * If they are not the same, we arrived here by compiling an + * ensemble dispatch. Ensemble subcommands that lead to script + * evaluation are not supposed to get compiled, because a command + * such as [info level] in the script can expose some of the dispatch + * shenanigans. This means that we don't have to tend to the + * housekeeping, and can escape now. + */ + + if (ePtr->nline != objc) { + return; + } - if (ePtr->nline != objc) { - Tcl_Panic ("TIP 280 data structure inconsistency"); - } + /* + * Having disposed of the ensemble cases, we can state... + * A few truths ... + * (1) ePtr->nline == objc + * (2) (ePtr->line[word] < 0) => !literal, for all words + * (3) (word == 0) => !literal + * + * Item (2) is why we can use objv to get the literals, and do not + * have to save them at compile time. + */ - for (word = 1; word < objc; word++) { - if (ePtr->line[word] >= 0) { - int isnew; - Tcl_HashEntry *hPtr = Tcl_CreateHashEntry(iPtr->lineLABCPtr, - objv[word], &isnew); - CFWordBC *cfwPtr = ckalloc(sizeof(CFWordBC)); - - cfwPtr->framePtr = cfPtr; - cfwPtr->obj = objv[word]; - cfwPtr->pc = pc; - cfwPtr->word = word; - cfwPtr->nextPtr = lastPtr; - lastPtr = cfwPtr; - - if (isnew) { - /* - * The word is not on the stack yet, remember the current - * location and initialize references. - */ - - cfwPtr->prevPtr = NULL; - } else { - /* - * The object is already on the stack, however it may have - * a different location now (literal sharing may map - * multiple location to a single Tcl_Obj*. Save the old - * information in the new structure. - */ - - cfwPtr->prevPtr = Tcl_GetHashValue(hPtr); - } + for (word = 1; word < objc; word++) { + if (ePtr->line[word] >= 0) { + int isnew; + Tcl_HashEntry *hPtr = Tcl_CreateHashEntry(iPtr->lineLABCPtr, + objv[word], &isnew); + CFWordBC *cfwPtr = ckalloc(sizeof(CFWordBC)); + + cfwPtr->framePtr = cfPtr; + cfwPtr->obj = objv[word]; + cfwPtr->pc = pc; + cfwPtr->word = word; + cfwPtr->nextPtr = lastPtr; + lastPtr = cfwPtr; + + if (isnew) { + /* + * The word is not on the stack yet, remember the current + * location and initialize references. + */ - Tcl_SetHashValue(hPtr, cfwPtr); + cfwPtr->prevPtr = NULL; + } else { + /* + * The object is already on the stack, however it may have + * a different location now (literal sharing may map + * multiple location to a single Tcl_Obj*. Save the old + * information in the new structure. + */ + + cfwPtr->prevPtr = Tcl_GetHashValue(hPtr); } - } /* for */ - cfPtr->litarg = lastPtr; - } /* if */ + Tcl_SetHashValue(hPtr, cfwPtr); + } + } /* for */ + + cfPtr->litarg = lastPtr; } /* diff --git a/generic/tclCompCmds.c b/generic/tclCompCmds.c index 37ce335..7fe0728 100644 --- a/generic/tclCompCmds.c +++ b/generic/tclCompCmds.c @@ -2536,7 +2536,7 @@ CompileEachloopCmd( Tcl_DStringInit(&varList); TclDStringAppendToken(&varList, &tokenPtr[1]); - code = Tcl_SplitList(interp, Tcl_DStringValue(&varList), + code = Tcl_SplitList(NULL, Tcl_DStringValue(&varList), &varcList[loopIndex], &varvList[loopIndex]); Tcl_DStringFree(&varList); if (code != TCL_OK) { @@ -2978,7 +2978,8 @@ TclCompileFormatCmd( ckfree(objv); Tcl_DecrRefCount(formatObj); if (tmpObj == NULL) { - return TCL_ERROR; + TclCompileSyntaxError(interp, envPtr); + return TCL_OK; } /* diff --git a/generic/tclCompCmdsGR.c b/generic/tclCompCmdsGR.c index 2c71dc5..fc68509 100644 --- a/generic/tclCompCmdsGR.c +++ b/generic/tclCompCmdsGR.c @@ -2531,6 +2531,7 @@ TclCompileSyntaxError( TclEmitPush(TclRegisterNewLiteral(envPtr, bytes, numBytes), envPtr); CompileReturnInternal(envPtr, INST_SYNTAX, TCL_ERROR, 0, TclNoErrorStack(interp, Tcl_GetReturnOptions(interp, TCL_ERROR))); + Tcl_ResetResult(interp); } /* diff --git a/generic/tclCompCmdsSZ.c b/generic/tclCompCmdsSZ.c index 3a91c83..320ed57 100644 --- a/generic/tclCompCmdsSZ.c +++ b/generic/tclCompCmdsSZ.c @@ -743,6 +743,9 @@ TclSubstCompile( Tcl_InterpState state = NULL; TclSubstParse(interp, bytes, numBytes, flags, &parse, &state); + if (state != NULL) { + Tcl_ResetResult(interp); + } /* * Tricky point! If the first token does not result in a *guaranteed* push diff --git a/generic/tclCompile.c b/generic/tclCompile.c index 6a2a318..a52ad3e 100644 --- a/generic/tclCompile.c +++ b/generic/tclCompile.c @@ -16,6 +16,8 @@ #include "tclCompile.h" #include <assert.h> +#define REWRITE + /* * Table of all AuxData types. */ @@ -562,8 +564,10 @@ static void EnterCmdExtentData(CompileEnv *envPtr, int cmdNumber, int numSrcBytes, int numCodeBytes); static void EnterCmdStartData(CompileEnv *envPtr, int cmdNumber, int srcOffset, int codeOffset); +#ifndef REWRITE static Command * FindCompiledCommandFromToken(Tcl_Interp *interp, Tcl_Token *tokenPtr); +#endif static void FreeByteCodeInternalRep(Tcl_Obj *objPtr); static void FreeSubstCodeInternalRep(Tcl_Obj *objPtr); static int GetCmdLocEncodingSize(CompileEnv *envPtr); @@ -1292,8 +1296,6 @@ ReleaseCmdWordData( ckfree((char *) eclPtr->loc); } - Tcl_DeleteHashTable (&eclPtr->litInfo); - ckfree((char *) eclPtr); } @@ -1378,7 +1380,6 @@ TclInitCompileEnv( envPtr->extCmdMapPtr->nloc = 0; envPtr->extCmdMapPtr->nuloc = 0; envPtr->extCmdMapPtr->path = NULL; - Tcl_InitHashTable(&envPtr->extCmdMapPtr->litInfo, TCL_ONE_WORD_KEYS); if ((invoker == NULL) || (invoker->type == TCL_LOCATION_EVAL_LIST)) { /* @@ -1671,6 +1672,7 @@ TclWordKnownAtCompileTime( return 1; } +#ifndef REWRITE /* * --------------------------------------------------------------------- * @@ -1718,6 +1720,7 @@ FindCompiledCommandFromToken( Tcl_DStringFree(&ds); return cmdPtr; } +#endif /* *---------------------------------------------------------------------- @@ -1737,6 +1740,339 @@ FindCompiledCommandFromToken( *---------------------------------------------------------------------- */ +#ifdef REWRITE + +static int +ExpandRequested( + Tcl_Token *tokenPtr, + int numWords) +{ + /* Determine whether any words of the command require expansion */ + while (numWords--) { + if (tokenPtr->type == TCL_TOKEN_EXPAND_WORD) { + return 1; + } + tokenPtr = TokenAfter(tokenPtr); + } + return 0; +} + +static void +CompileCmdLiteral( + Tcl_Interp *interp, + Tcl_Obj *cmdObj, + CompileEnv *envPtr) +{ + int numBytes; + const char *bytes = Tcl_GetStringFromObj(cmdObj, &numBytes); + int cmdLitIdx = TclRegisterNewCmdLiteral(envPtr, bytes, numBytes); + Command *cmdPtr = (Command *) Tcl_GetCommandFromObj(interp, cmdObj); + + if (cmdPtr) { + TclSetCmdNameObj(interp, TclFetchLiteral(envPtr, cmdLitIdx), cmdPtr); + } + TclEmitPush(cmdLitIdx, envPtr); +} + +void +TclCompileInvocation( + Tcl_Interp *interp, + Tcl_Token *tokenPtr, + Tcl_Obj *cmdObj, + int numWords, + CompileEnv *envPtr) +{ + int wordIdx = 0; + DefineLineInformation; + + if (cmdObj) { + CompileCmdLiteral(interp, cmdObj, envPtr); + wordIdx = 1; + tokenPtr = TokenAfter(tokenPtr); + } + + for (; wordIdx < numWords; wordIdx++, tokenPtr = TokenAfter(tokenPtr)) { + int objIdx; + + SetLineInformation(wordIdx); + + if (tokenPtr->type != TCL_TOKEN_SIMPLE_WORD) { + CompileTokens(envPtr, tokenPtr, interp); + continue; + } + + objIdx = TclRegisterNewLiteral(envPtr, + tokenPtr[1].start, tokenPtr[1].size); + if (envPtr->clNext) { + TclContinuationsEnterDerived(TclFetchLiteral(envPtr, objIdx), + tokenPtr[1].start - envPtr->source, envPtr->clNext); + } + TclEmitPush(objIdx, envPtr); + } + + if (wordIdx <= 255) { + TclEmitInstInt1(INST_INVOKE_STK1, wordIdx, envPtr); + } else { + TclEmitInstInt4(INST_INVOKE_STK4, wordIdx, envPtr); + } +} + +static void +CompileExpanded( + Tcl_Interp *interp, + Tcl_Token *tokenPtr, + Tcl_Obj *cmdObj, + int numWords, + CompileEnv *envPtr) +{ + int wordIdx = 0; + DefineLineInformation; + + + StartExpanding(envPtr); + if (cmdObj) { + CompileCmdLiteral(interp, cmdObj, envPtr); + wordIdx = 1; + tokenPtr = TokenAfter(tokenPtr); + } + + for (; wordIdx < numWords; wordIdx++, tokenPtr = TokenAfter(tokenPtr)) { + int objIdx; + + SetLineInformation(wordIdx); + + if (tokenPtr->type != TCL_TOKEN_SIMPLE_WORD) { + CompileTokens(envPtr, tokenPtr, interp); + if (tokenPtr->type == TCL_TOKEN_EXPAND_WORD) { + TclEmitInstInt4(INST_EXPAND_STKTOP, + envPtr->currStackDepth, envPtr); + } + continue; + } + + objIdx = TclRegisterNewLiteral(envPtr, + tokenPtr[1].start, tokenPtr[1].size); + if (envPtr->clNext) { + TclContinuationsEnterDerived(TclFetchLiteral(envPtr, objIdx), + tokenPtr[1].start - envPtr->source, envPtr->clNext); + } + TclEmitPush(objIdx, envPtr); + } + + /* + * The stack depth during argument expansion can only be + * managed at runtime, as the number of elements in the + * expanded lists is not known at compile time. We adjust here + * the stack depth estimate so that it is correct after the + * command with expanded arguments returns. + * + * The end effect of this command's invocation is that all the + * words of the command are popped from the stack, and the + * result is pushed: the stack top changes by (1-wordIdx). + * + * Note that the estimates are not correct while the command + * is being prepared and run, INST_EXPAND_STKTOP is not + * stack-neutral in general. + */ + + TclEmitOpcode(INST_INVOKE_EXPANDED, envPtr); + envPtr->expandCount--; + TclAdjustStackDepth(1 - wordIdx, envPtr); +} + +static int +CompileCmdCompileProc( + Tcl_Interp *interp, + Tcl_Parse *parsePtr, + Command *cmdPtr, + CompileEnv *envPtr) +{ + int unwind = 0, incrOffset = -1; + DefineLineInformation; + + /* + * Emit of the INST_START_CMD instruction is controlled by + * the value of envPtr->atCmdStart: + * + * atCmdStart == 2 : We are not using the INST_START_CMD instruction. + * atCmdStart == 1 : INST_START_CMD was the last instruction emitted. + * : We do not need to emit another. Instead we + * : increment the number of cmds started at it (except + * : for the special case at the start of a script.) + * atCmdStart == 0 : The last instruction was something else. We need + * : to emit INST_START_CMD here. + */ + + switch (envPtr->atCmdStart) { + case 0: + unwind = tclInstructionTable[INST_START_CMD].numBytes; + TclEmitInstInt4(INST_START_CMD, 0, envPtr); + incrOffset = envPtr->codeNext - envPtr->codeStart; + TclEmitInt4(0, envPtr); + break; + case 1: + if (envPtr->codeNext > envPtr->codeStart) { + incrOffset = envPtr->codeNext - 4 - envPtr->codeStart; + } + break; + case 2: + /* Nothing to do */ + ; + } + + if (TCL_OK == TclAttemptCompileProc(interp, parsePtr, 1, cmdPtr, envPtr)) { + if (incrOffset >= 0) { + /* + * We successfully compiled a command. Increment the number + * of commands that start at the currently active INST_START_CMD. + */ + unsigned char *incrPtr = envPtr->codeStart + incrOffset; + unsigned char *startPtr = incrPtr - 5; + + TclIncrUInt4AtPtr(incrPtr, 1); + if (unwind) { + /* We started the INST_START_CMD. Record the code length. */ + TclStoreInt4AtPtr(envPtr->codeNext - startPtr, startPtr + 1); + } + } + return TCL_OK; + } + + envPtr->codeNext -= unwind; /* Unwind INST_START_CMD */ + + /* + * Throw out any line information generated by the failed + * compile attempt. + */ + while (mapPtr->nuloc - 1 > eclIndex) { + mapPtr->nuloc--; + ckfree(mapPtr->loc[mapPtr->nuloc].line); + mapPtr->loc[mapPtr->nuloc].line = NULL; + } + + /* + * Reset the index of next command. + * Toss out any from failed nested partial compiles. + */ + envPtr->numCommands = mapPtr->nuloc; + + return TCL_ERROR; +} + +static int +CompileCommandTokens( + Tcl_Interp *interp, + Tcl_Parse *parsePtr, + CompileEnv *envPtr) +{ + Interp *iPtr = (Interp *) interp; + Tcl_Token *tokenPtr = parsePtr->tokenPtr; + ExtCmdLoc *eclPtr = envPtr->extCmdMapPtr; + Tcl_Obj *cmdObj = Tcl_NewObj(); + Command *cmdPtr = NULL; + int code = TCL_ERROR; + int cmdKnown, expand = -1; + int *wlines, wlineat; + int cmdLine = envPtr->line; + int *clNext = envPtr->clNext; + int cmdIdx = envPtr->numCommands; + int startCodeOffset = envPtr->codeNext - envPtr->codeStart; + + assert (parsePtr->numWords > 0); + + /* Pre-Compile */ + + envPtr->numCommands++; + EnterCmdStartData(envPtr, cmdIdx, + parsePtr->commandStart - envPtr->source, startCodeOffset); + + /* + * TIP #280. Scan the words and compute the extended location + * information. The map first contain full per-word line + * information for use by the compiler. This is later replaced by + * a reduced form which signals non-literal words, stored in + * 'wlines'. + */ + + EnterCmdWordData(eclPtr, parsePtr->commandStart - envPtr->source, + parsePtr->tokenPtr, parsePtr->commandStart, + parsePtr->commandSize, parsePtr->numWords, cmdLine, + clNext, &wlines, envPtr); + wlineat = eclPtr->nuloc - 1; + + envPtr->line = eclPtr->loc[wlineat].line[0]; + envPtr->clNext = eclPtr->loc[wlineat].next[0]; + + /* Do we know the command word? */ + Tcl_IncrRefCount(cmdObj); + tokenPtr = parsePtr->tokenPtr; + cmdKnown = TclWordKnownAtCompileTime(tokenPtr, cmdObj); + + /* Is this a command we should (try to) compile with a compileProc ? */ + if (cmdKnown && !(iPtr->flags & DONT_COMPILE_CMDS_INLINE)) { + cmdPtr = (Command *) Tcl_GetCommandFromObj(interp, cmdObj); + if (cmdPtr) { + /* + * Found a command. Test the ways we can be told + * not to attempt to compile it. + */ + if ((cmdPtr->compileProc == NULL) + || (cmdPtr->nsPtr->flags & NS_SUPPRESS_COMPILATION) + || (cmdPtr->flags & CMD_HAS_EXEC_TRACES)) { + cmdPtr = NULL; + } + } + if (cmdPtr && !(cmdPtr->flags & CMD_COMPILES_EXPANDED)) { + expand = ExpandRequested(parsePtr->tokenPtr, parsePtr->numWords); + if (expand) { + /* We need to expand, but compileProc cannot. */ + cmdPtr = NULL; + } + } + } + + /* If cmdPtr != NULL, we will try to call cmdPtr->compileProc */ + if (cmdPtr) { + code = CompileCmdCompileProc(interp, parsePtr, cmdPtr, envPtr); + } + + if (code == TCL_ERROR) { + if (expand < 0) { + expand = ExpandRequested(parsePtr->tokenPtr, parsePtr->numWords); + } + + if (expand) { + CompileExpanded(interp, parsePtr->tokenPtr, + cmdKnown ? cmdObj : NULL, parsePtr->numWords, envPtr); + } else { + TclCompileInvocation(interp, parsePtr->tokenPtr, + cmdKnown ? cmdObj : NULL, parsePtr->numWords, envPtr); + } + } + + Tcl_DecrRefCount(cmdObj); + + TclEmitOpcode(INST_POP, envPtr); + EnterCmdExtentData(envPtr, cmdIdx, + parsePtr->term - parsePtr->commandStart, + (envPtr->codeNext-envPtr->codeStart) - startCodeOffset); + + /* + * TIP #280: Free full form of per-word line data and insert the + * reduced form now + */ + + envPtr->line = cmdLine; + envPtr->clNext = clNext; + ckfree(eclPtr->loc[wlineat].line); + ckfree(eclPtr->loc[wlineat].next); + eclPtr->loc[wlineat].line = wlines; + eclPtr->loc[wlineat].next = NULL; + + return cmdIdx; +} +#endif + void TclCompileScript( Tcl_Interp *interp, /* Used for error and status reporting. Also @@ -1748,6 +2084,121 @@ TclCompileScript( * first null character. */ CompileEnv *envPtr) /* Holds resulting instructions. */ { +#ifdef REWRITE + int lastCmdIdx = -1; /* Index into envPtr->cmdMapPtr of the last + * command this routine compiles into bytecode. + * Initial value of -1 indicates this routine + * has not yet generated any bytecode. */ + const char *p = script; /* Where we are in our compile. */ + + if (envPtr->iPtr == NULL) { + Tcl_Panic("TclCompileScript() called on uninitialized CompileEnv"); + } + + /* Each iteration compiles one command from the script. */ + + while (numBytes > 0) { + Tcl_Parse parse; + const char *next; + + if (TCL_OK != Tcl_ParseCommand(interp, p, numBytes, 0, &parse)) { + /* + * Compile bytecodes to report the parse error at runtime. + */ + + Tcl_LogCommandInfo(interp, script, parse.commandStart, + parse.term + 1 - parse.commandStart); + TclCompileSyntaxError(interp, envPtr); + return; + } + +#ifdef TCL_COMPILE_DEBUG + /* + * If tracing, print a line for each top level command compiled. + */ + + if ((tclTraceCompile >= 1) && (envPtr->procPtr == NULL)) { + int commandLength = parse.term - parse.commandStart; + fprintf(stdout, " Compiling: "); + TclPrintSource(stdout, parse.commandStart, + TclMin(commandLength, 55)); + fprintf(stdout, "\n"); + } +#endif + + /* + * TIP #280: Count newlines before the command start. + * (See test info-30.33). + */ + + TclAdvanceLines(&envPtr->line, p, parse.commandStart); + TclAdvanceContinuations(&envPtr->line, &envPtr->clNext, + parse.commandStart - envPtr->source); + + /* + * Advance parser to the next command in the script. + */ + + next = parse.commandStart + parse.commandSize; + numBytes -= next - p; + p = next; + + if (parse.numWords == 0) { + /* + * The "command" parsed has no words. In this case + * we can skip the rest of the loop body. With no words, + * clearly CompileCommandTokens() has nothing to do. Since + * the parser aggressively sucks up leading comment and white + * space, including newlines, parse.commandStart must be + * pointing at either the end of script, or a command-terminating + * semi-colon. In either case, the TclAdvance*() calls have + * nothing to do. Finally, when no words are parsed, no + * tokens have been allocated at parse.tokenPtr so there's + * also nothing for Tcl_FreeParse() to do. + * + * The advantage of this shortcut is that CompileCommandTokens() + * can be written with an assumption that parse.numWords > 0, + * with the implication the CCT() always generates bytecode. + */ + continue; + } + + lastCmdIdx = CompileCommandTokens(interp, &parse, envPtr); + + /* + * TIP #280: Track lines in the just compiled command. + */ + + TclAdvanceLines(&envPtr->line, parse.commandStart, p); + TclAdvanceContinuations(&envPtr->line, &envPtr->clNext, + p - envPtr->source); + Tcl_FreeParse(&parse); + } + + if (lastCmdIdx == -1) { + /* + * Compiling the script yielded no bytecode. The script must be + * all whitespace, comments, and empty commands. Such scripts + * are defined to successfully produce the empty string result, + * so we emit the simple bytecode that makes that happen. + */ + PushStringLiteral(envPtr, ""); + } else { + /* + * We compiled at least one command to bytecode. The routine + * CompileCommandTokens() follows the bytecode of each compiled + * command with an INST_POP, so that stack balance is maintained + * when several commands are in sequence. (The result of each + * command is thrown away before moving on to the next command). + * For the last command compiled, we need to undo that INST_POP + * so that the result of the last command becomes the result of + * the script. The code here removes that trailing INST_POP. + */ + envPtr->cmdMapPtr[lastCmdIdx].numCodeBytes--; + envPtr->codeNext--; + envPtr->currStackDepth++; + } +#else int lastTopLevelCmdIndex = -1; /* Index of most recent toplevel command in * the command location table. Initialized to @@ -2189,6 +2640,7 @@ TclCompileScript( if (envPtr->codeNext == entryCodeNext) { PushStringLiteral(envPtr, ""); } +#endif } /* @@ -3959,69 +4411,6 @@ TclFixupForwardJump( } } - /* - * TIP #280: Adjust the mapping from PC values to the per-command - * information about arguments and their line numbers. - * - * Note: We cannot simply remove an out-of-date entry and then reinsert - * with the proper PC, because then we might overwrite another entry which - * was at that location. Therefore we pull (copy + delete) all effected - * entries (beyond the fixed PC) into an array, update them there, and at - * last reinsert them all. - */ - - { - ExtCmdLoc* eclPtr = envPtr->extCmdMapPtr; - - /* A helper structure */ - - typedef struct { - int pc; - int cmd; - } MAP; - - /* - * And the helper array. At most the whole hashtable is placed into - * this. - */ - - MAP *map = (MAP*) ckalloc (sizeof(MAP) * eclPtr->litInfo.numEntries); - - Tcl_HashSearch hSearch; - Tcl_HashEntry* hPtr; - int n, k, isnew; - - /* - * Phase I: Locate the affected entries, and save them in adjusted - * form to the array. This removes them from the hash. - */ - - for (n = 0, hPtr = Tcl_FirstHashEntry(&eclPtr->litInfo, &hSearch); - hPtr != NULL; - hPtr = Tcl_NextHashEntry(&hSearch)) { - - map [n].cmd = PTR2INT(Tcl_GetHashValue(hPtr)); - map [n].pc = PTR2INT(Tcl_GetHashKey (&eclPtr->litInfo,hPtr)); - - if (map[n].pc >= (jumpFixupPtr->codeOffset + 2)) { - Tcl_DeleteHashEntry(hPtr); - map [n].pc += 3; - n++; - } - } - - /* - * Phase II: Re-insert the modified entries into the hash. - */ - - for (k=0;k<n;k++) { - hPtr = Tcl_CreateHashEntry(&eclPtr->litInfo, INT2PTR(map[k].pc), &isnew); - Tcl_SetHashValue(hPtr, INT2PTR(map[k].cmd)); - } - - ckfree (map); - } - return 1; /* the jump was grown */ } diff --git a/generic/tclCompile.h b/generic/tclCompile.h index cf8475d..56315db 100644 --- a/generic/tclCompile.h +++ b/generic/tclCompile.h @@ -193,13 +193,6 @@ typedef struct ExtCmdLoc { ECL *loc; /* Command word locations (lines). */ int nloc; /* Number of allocated entries in 'loc'. */ int nuloc; /* Number of used entries in 'loc'. */ - Tcl_HashTable litInfo; /* Indexed by bytecode 'PC', to have the - * information accessible per command and - * argument, not per whole bytecode. Value is - * index of command in 'loc', giving us the - * literals to associate with line information - * as command argument, see - * TclArgumentBCEnter() */ } ExtCmdLoc; /* @@ -991,6 +984,9 @@ MODULE_SCOPE ByteCode * TclCompileObj(Tcl_Interp *interp, Tcl_Obj *objPtr, *---------------------------------------------------------------- */ +MODULE_SCOPE int TclAttemptCompileProc(Tcl_Interp *interp, + Tcl_Parse *parsePtr, int depth, Command *cmdPtr, + CompileEnv *envPtr); MODULE_SCOPE void TclCleanupByteCode(ByteCode *codePtr); MODULE_SCOPE void TclCleanupStackForBreakContinue(CompileEnv *envPtr, ExceptionAux *auxPtr); @@ -1002,6 +998,9 @@ MODULE_SCOPE void TclCompileExpr(Tcl_Interp *interp, const char *script, MODULE_SCOPE void TclCompileExprWords(Tcl_Interp *interp, Tcl_Token *tokenPtr, int numWords, CompileEnv *envPtr); +MODULE_SCOPE void TclCompileInvocation(Tcl_Interp *interp, + Tcl_Token *tokenPtr, Tcl_Obj *cmdObj, int numWords, + CompileEnv *envPtr); MODULE_SCOPE void TclCompileScript(Tcl_Interp *interp, const char *script, int numBytes, CompileEnv *envPtr); diff --git a/generic/tclEnsemble.c b/generic/tclEnsemble.c index 680ab45d..bab63c9 100644 --- a/generic/tclEnsemble.c +++ b/generic/tclEnsemble.c @@ -35,9 +35,6 @@ static void MakeCachedEnsembleCommand(Tcl_Obj *objPtr, static void FreeEnsembleCmdRep(Tcl_Obj *objPtr); static void DupEnsembleCmdRep(Tcl_Obj *objPtr, Tcl_Obj *copyPtr); static void StringOfEnsembleCmdRep(Tcl_Obj *objPtr); -static int CompileToCompiledCommand(Tcl_Interp *interp, - Tcl_Parse *parsePtr, int depth, Command *cmdPtr, - CompileEnv *envPtr); static void CompileToInvokedCommand(Tcl_Interp *interp, Tcl_Parse *parsePtr, Tcl_Obj *replacements, Command *cmdPtr, CompileEnv *envPtr); @@ -2994,8 +2991,8 @@ TclCompileEnsemble( */ invokeAnyway = 1; - if (CompileToCompiledCommand(interp, parsePtr, depth, cmdPtr, - envPtr) == TCL_OK) { + if (TCL_OK == TclAttemptCompileProc(interp, parsePtr, depth, cmdPtr, + envPtr)) { ourResult = TCL_OK; goto cleanup; } @@ -3029,24 +3026,16 @@ TclCompileEnsemble( return ourResult; } -/* - * How to compile a subcommand using its own command compiler. To do that, we - * have to perform some trickery to rewrite the arguments, as compilers *must* - * have parse tokens that refer to addresses in the original script. - */ - -static int -CompileToCompiledCommand( +int +TclAttemptCompileProc( Tcl_Interp *interp, Tcl_Parse *parsePtr, int depth, Command *cmdPtr, CompileEnv *envPtr) /* Holds resulting instructions. */ { - Tcl_Parse synthetic; - Tcl_Token *tokenPtr; int result, i; - int savedNumCmds = envPtr->numCommands; + Tcl_Token *saveTokenPtr = parsePtr->tokenPtr; int savedStackDepth = envPtr->currStackDepth; unsigned savedCodeNext = envPtr->codeNext - envPtr->codeStart; DefineLineInformation; @@ -3055,47 +3044,17 @@ CompileToCompiledCommand( return TCL_ERROR; } - TclParseInit(interp, NULL, 0, &synthetic); - synthetic.numWords = parsePtr->numWords - depth + 1; - TclGrowParseTokenArray(&synthetic, 2); - synthetic.numTokens = 2; - - /* - * Now we have the space to work in, install something rewritten. The - * first word will "officially" be the bytes of the structured ensemble - * name. That's technically wrong, but nobody will care; we just need - * *something* here... - */ - - synthetic.tokenPtr[0].type = TCL_TOKEN_SIMPLE_WORD; - synthetic.tokenPtr[0].start = parsePtr->tokenPtr[0].start; - synthetic.tokenPtr[0].numComponents = 1; - synthetic.tokenPtr[1].type = TCL_TOKEN_TEXT; - synthetic.tokenPtr[1].start = parsePtr->tokenPtr[0].start; - synthetic.tokenPtr[1].numComponents = 0; - for (i=0,tokenPtr=parsePtr->tokenPtr ; i<depth ; i++) { - int sclen = (tokenPtr->start - synthetic.tokenPtr[0].start) - + tokenPtr->size; - - synthetic.tokenPtr[0].size = sclen; - synthetic.tokenPtr[1].size = sclen; - tokenPtr = TokenAfter(tokenPtr); - } - /* - * Copy over the real argument tokens. + * Advance parsePtr->tokenPtr so that it points at the last subcommand. + * This will be wrong, but it will not matter, and it will put the + * tokens for the arguments in the right place without the needed to + * allocate a synthetic Tcl_Parse struct, or copy tokens around. */ - for (i=1; i<synthetic.numWords; i++) { - int toCopy; - - toCopy = tokenPtr->numComponents + 1; - TclGrowParseTokenArray(&synthetic, toCopy); - memcpy(synthetic.tokenPtr + synthetic.numTokens, tokenPtr, - sizeof(Tcl_Token) * toCopy); - synthetic.numTokens += toCopy; - tokenPtr = TokenAfter(tokenPtr); + for (i = 0; i < depth - 1; i++) { + parsePtr->tokenPtr = TokenAfter(parsePtr->tokenPtr); } + parsePtr->numWords -= (depth - 1); /* * Shift the line information arrays to account for different word @@ -3109,7 +3068,7 @@ CompileToCompiledCommand( * Hand off compilation to the subcommand compiler. At last! */ - result = cmdPtr->compileProc(interp, &synthetic, cmdPtr, envPtr); + result = cmdPtr->compileProc(interp, parsePtr, cmdPtr, envPtr); /* * Undo the shift. @@ -3118,22 +3077,37 @@ CompileToCompiledCommand( mapPtr->loc[eclIndex].line -= (depth - 1); mapPtr->loc[eclIndex].next -= (depth - 1); + parsePtr->numWords += (depth - 1); + parsePtr->tokenPtr = saveTokenPtr; + /* - * If our target fails to compile, revert the number of commands and the - * pointer to the place to issue the next instruction. [Bug 3600328] + * If our target failed to compile, revert any data from failed partial + * compiles. Note that envPtr->numCommands need not be checked because + * we avoid compiling subcommands that recursively call TclCompileScript(). */ if (result != TCL_OK) { - envPtr->numCommands = savedNumCmds; envPtr->currStackDepth = savedStackDepth; envPtr->codeNext = envPtr->codeStart + savedCodeNext; - } +#ifdef TCL_COMPILE_DEBUG + } else { + /* + * Confirm that the command compiler generated a single value on + * the stack as its result. This is only done in debugging mode, + * as it *should* be correct and normal users have no reasonable + * way to fix it anyway. + */ - /* - * Clean up if necessary. - */ + int diff = envPtr->currStackDepth - savedStackDepth; + + if (diff != 1) { + Tcl_Panic("bad stack adjustment when compiling" + " %.*s (was %d instead of 1)", parsePtr->tokenPtr->size, + parsePtr->tokenPtr->start, diff); + } +#endif + } - Tcl_FreeParse(&synthetic); return result; } @@ -3163,12 +3137,16 @@ CompileToInvokedCommand( */ Tcl_ListObjGetElements(NULL, replacements, &numWords, &words); - for (i=0,tokPtr=parsePtr->tokenPtr ; i<parsePtr->numWords ; i++) { + for (i = 0, tokPtr = parsePtr->tokenPtr; i < parsePtr->numWords; + i++, tokPtr = TokenAfter(tokPtr)) { if (i > 0 && i < numWords+1) { bytes = Tcl_GetStringFromObj(words[i-1], &length); PushLiteral(envPtr, bytes, length); - } else if (tokPtr->type == TCL_TOKEN_SIMPLE_WORD) { - /* TODO: Check about registering Cmd Literals here */ + continue; + } + + SetLineInformation(i); + if (tokPtr->type == TCL_TOKEN_SIMPLE_WORD) { int literal = TclRegisterNewLiteral(envPtr, tokPtr[1].start, tokPtr[1].size); @@ -3176,14 +3154,12 @@ CompileToInvokedCommand( TclContinuationsEnterDerived( TclFetchLiteral(envPtr, literal), tokPtr[1].start - envPtr->source, - mapPtr->loc[eclIndex].next[i]); + envPtr->clNext); } TclEmitPush(literal, envPtr); } else { - SetLineInformation(i); CompileTokens(envPtr, tokPtr, interp); } - tokPtr = TokenAfter(tokPtr); } /* @@ -3229,6 +3205,15 @@ CompileBasicNArgCommand( * compiled. */ CompileEnv *envPtr) /* Holds resulting instructions. */ { +#if 1 + Tcl_Obj *objPtr = Tcl_NewObj(); + + Tcl_IncrRefCount(objPtr); + Tcl_GetCommandFullName(interp, (Tcl_Command) cmdPtr, objPtr); + TclCompileInvocation(interp, parsePtr->tokenPtr, objPtr, + parsePtr->numWords, envPtr); + Tcl_DecrRefCount(objPtr); +#else Tcl_Token *tokenPtr; Tcl_Obj *objPtr; char *bytes; @@ -3272,6 +3257,7 @@ CompileBasicNArgCommand( } else { TclEmitInstInt4(INST_INVOKE_STK4, i, envPtr); } +#endif return TCL_OK; } diff --git a/generic/tclExecute.c b/generic/tclExecute.c index d3a0d32..f8ed667 100644 --- a/generic/tclExecute.c +++ b/generic/tclExecute.c @@ -721,7 +721,7 @@ static ExceptionRange * GetExceptRangeForPc(const unsigned char *pc, int catchOnly, ByteCode *codePtr); static const char * GetSrcInfoForPc(const unsigned char *pc, ByteCode *codePtr, int *lengthPtr, - const unsigned char **pcBeg); + const unsigned char **pcBeg, int *cmdIdxPtr); static Tcl_Obj ** GrowEvaluationStack(ExecEnv *eePtr, int growth, int move); static void IllegalExprOperandType(Tcl_Interp *interp, @@ -1426,17 +1426,12 @@ Tcl_NRExprObj( Tcl_Obj *resultPtr) { ByteCode *codePtr; + Tcl_InterpState state = Tcl_SaveInterpState(interp, TCL_OK); - /* TODO: consider saving whole state? */ - Tcl_Obj *saveObjPtr = Tcl_GetObjResult(interp); - - Tcl_IncrRefCount(saveObjPtr); - + Tcl_ResetResult(interp); codePtr = CompileExprObj(interp, objPtr); - /* TODO: Confirm reset not required? */ - /*Tcl_ResetResult(interp);*/ - Tcl_NRAddCallback(interp, ExprObjCallback, saveObjPtr, resultPtr, + Tcl_NRAddCallback(interp, ExprObjCallback, state, resultPtr, NULL, NULL); return TclNRExecuteByteCode(interp, codePtr); } @@ -1447,14 +1442,15 @@ ExprObjCallback( Tcl_Interp *interp, int result) { - Tcl_Obj *saveObjPtr = data[0]; + Tcl_InterpState state = data[0]; Tcl_Obj *resultPtr = data[1]; if (result == TCL_OK) { TclSetDuplicateObj(resultPtr, Tcl_GetObjResult(interp)); - Tcl_SetObjResult(interp, saveObjPtr); + (void) Tcl_RestoreInterpState(interp, state); + } else { + Tcl_DiscardInterpState(state); } - TclDecrRefCount(saveObjPtr); return result; } @@ -2435,8 +2431,11 @@ TEBCresume( iPtr->cmdFramePtr = bcFramePtr; if (iPtr->flags & INTERP_DEBUG_FRAME) { - TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, - codePtr, bcFramePtr, pc - codePtr->codeStart); + int cmd; + if (GetSrcInfoForPc(pc, codePtr, NULL, NULL, &cmd)) { + TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, + codePtr, bcFramePtr, cmd, pc - codePtr->codeStart); + } } pc++; @@ -2889,8 +2888,11 @@ TEBCresume( iPtr->cmdFramePtr = bcFramePtr; if (iPtr->flags & INTERP_DEBUG_FRAME) { - TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, - codePtr, bcFramePtr, pc - codePtr->codeStart); + int cmd; + if (GetSrcInfoForPc(pc, codePtr, NULL, NULL, &cmd)) { + TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, + codePtr, bcFramePtr, cmd, pc - codePtr->codeStart); + } } DECACHE_STACK_INFO(); @@ -3035,8 +3037,11 @@ TEBCresume( bcFramePtr->data.tebc.pc = (char *) pc; iPtr->cmdFramePtr = bcFramePtr; if (iPtr->flags & INTERP_DEBUG_FRAME) { - TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, - codePtr, bcFramePtr, pc - codePtr->codeStart); + int cmd; + if (GetSrcInfoForPc(pc, codePtr, NULL, NULL, &cmd)) { + TclArgumentBCEnter((Tcl_Interp *) iPtr, objv, objc, + codePtr, bcFramePtr, cmd, pc - codePtr->codeStart); + } } iPtr->ensembleRewrite.sourceObjs = objv; iPtr->ensembleRewrite.numRemovedObjs = opnd; @@ -6971,7 +6976,7 @@ TEBCresume( if ((result == TCL_ERROR) && !(iPtr->flags & ERR_ALREADY_LOGGED)) { const unsigned char *pcBeg; - bytes = GetSrcInfoForPc(pc, codePtr, &length, &pcBeg); + bytes = GetSrcInfoForPc(pc, codePtr, &length, &pcBeg, NULL); DECACHE_STACK_INFO(); TclLogCommandInfo(interp, codePtr->source, bytes, bytes ? length : 0, pcBeg, tosPtr); @@ -7153,7 +7158,7 @@ TEBCresume( } codePtr->flags |= TCL_BYTECODE_RECOMPILE; - bytes = GetSrcInfoForPc(pc, codePtr, &length, NULL); + bytes = GetSrcInfoForPc(pc, codePtr, &length, NULL, NULL); opnd = TclGetUInt4AtPtr(pc+1); pc += (opnd-1); PUSH_OBJECT(Tcl_NewStringObj(bytes, length)); @@ -8646,7 +8651,7 @@ ValidatePcAndStackTop( if (checkStack && ((stackTop < 0) || (stackTop > stackUpperBound))) { int numChars; - const char *cmd = GetSrcInfoForPc(pc, codePtr, &numChars, NULL); + const char *cmd = GetSrcInfoForPc(pc, codePtr, &numChars, NULL, NULL); fprintf(stderr, "\nBad stack top %d at pc %u in TclNRExecuteByteCode (min 0, max %i)", stackTop, relativePc, stackUpperBound); @@ -8760,7 +8765,7 @@ TclGetSrcInfoForCmd( ByteCode *codePtr = (ByteCode *) cfPtr->data.tebc.codePtr; return GetSrcInfoForPc((unsigned char *) cfPtr->data.tebc.pc, - codePtr, lenPtr, NULL); + codePtr, lenPtr, NULL, NULL); } void @@ -8772,7 +8777,7 @@ TclGetSrcInfoForPc( if (cfPtr->cmd.str.cmd == NULL) { cfPtr->cmd.str.cmd = GetSrcInfoForPc( (unsigned char *) cfPtr->data.tebc.pc, codePtr, - &cfPtr->cmd.str.len, NULL); + &cfPtr->cmd.str.len, NULL, NULL); } if (cfPtr->cmd.str.cmd != NULL) { @@ -8832,9 +8837,12 @@ GetSrcInfoForPc( int *lengthPtr, /* If non-NULL, the location where the length * of the command's source should be stored. * If NULL, no length is stored. */ - const unsigned char **pcBeg)/* If non-NULL, the bytecode location + const unsigned char **pcBeg,/* If non-NULL, the bytecode location * where the current instruction starts. * If NULL; no pointer is stored. */ + int *cmdIdxPtr) /* If non-NULL, the location where the index + * of the command containing the pc should + * be stored. */ { register int pcOffset = (pc - codePtr->codeStart); int numCmds = codePtr->numCommands; @@ -8844,6 +8852,7 @@ GetSrcInfoForPc( int bestDist = INT_MAX; /* Distance of pc to best cmd's start pc. */ int bestSrcOffset = -1; /* Initialized to avoid compiler warning. */ int bestSrcLength = -1; /* Initialized to avoid compiler warning. */ + int bestCmdIdx = -1; if ((pcOffset < 0) || (pcOffset >= codePtr->numCodeBytes)) { if (pcBeg != NULL) *pcBeg = NULL; @@ -8911,6 +8920,7 @@ GetSrcInfoForPc( bestDist = dist; bestSrcOffset = srcOffset; bestSrcLength = srcLen; + bestCmdIdx = i; } } } @@ -8940,6 +8950,10 @@ GetSrcInfoForPc( *lengthPtr = bestSrcLength; } + if (cmdIdxPtr != NULL) { + *cmdIdxPtr = bestCmdIdx; + } + return (codePtr->source + bestSrcOffset); } diff --git a/generic/tclInt.h b/generic/tclInt.h index b940225..da09366 100644 --- a/generic/tclInt.h +++ b/generic/tclInt.h @@ -2828,7 +2828,7 @@ MODULE_SCOPE void TclArgumentRelease(Tcl_Interp *interp, Tcl_Obj *objv[], int objc); MODULE_SCOPE void TclArgumentBCEnter(Tcl_Interp *interp, Tcl_Obj *objv[], int objc, - void *codePtr, CmdFrame *cfPtr, int pc); + void *codePtr, CmdFrame *cfPtr, int cmd, int pc); MODULE_SCOPE void TclArgumentBCRelease(Tcl_Interp *interp, CmdFrame *cfPtr); MODULE_SCOPE void TclArgumentGet(Tcl_Interp *interp, Tcl_Obj *obj, |