Nasm源代码解析 预处理器--expand_smacro(二)

函数流程

/*
 * Expand all single-line macro calls made in the given line.
 * Return the expanded version of the line. The original is deemed
 * to be destroyed in the process. (In reality we'll just move
 * Tokens from input to output a lot of the time, rather than
 * actually bothering to destroy and replicate.)
 */
static Token *expand_smacro(Token * tline)
{
    Token *t, *tt, *mstart, **tail, *thead;
    SMacro *head = NULL, *m;
    Token **params;
    int *paramsize;
    int nparam, sparam, brackets, rescan;
    Token *org_tline = tline;
    Context *ctx;
    char *mname;

    /*
     * Trick: we should avoid changing the start token pointer since it can
     * be contained in "next" field of other token. Because of this
     * we allocate a copy of first token and work with it; at the end of
     * routine we copy it back
     */
    if (org_tline) {
        tline = new_Token(org_tline->next, org_tline->type, org_tline->text, 0);
        tline->mac = org_tline->mac;
        nasm_free(org_tline->text);
        org_tline->text = NULL;
    }

  again:
    tail = &thead;
    thead = NULL;

    while (tline)
    { /* main token loop */
        if ((mname = tline->text))
        {
            /* 获取与tline->text相关的宏 */
            /* if this token is a local macro, look in local context */
            if (tline->type == TOK_ID || tline->type == TOK_PREPROC_ID)
                ctx = get_ctx(mname, TRUE);
            else
                ctx = NULL;
            if (!ctx)
                head = smacros[hash(mname)];
            else
                head = ctx->localmac;
            /*
             * We've hit an identifier. As in is_mmacro below, we first
             * check whether the identifier is a single-line macro at
             * all, then think about checking for parameters if
             * necessary.
             */
            for (m = head; m; m = m->next)
            {
                if (!mstrcmp(m->name, mname, m->casesense))
                    break;
            }

            /* 如果存在与tline->text相关的宏 */
            if (m)
            {
                mstart = tline;
                params = NULL;
                paramsize = NULL;
                if (m->nparam == 0)
                {
                    /*
                     * Simple case: the macro is parameterless. Discard the
                     * one token that the macro call took, and push the
                     * expansion back on the to-do stack.
                     */
                    if (!m->expansion)
                    {
                        /* 处理Nasm预先定义的宏__FILE__ */
                        if (!strcmp("__FILE__", m->name))
                        {
                            long num = 0;
                            src_get(&num, &(tline->text));
                            nasm_quote(&(tline->text));
                            tline->type = TOK_STRING;
                            continue;
                        }
                        /* 处理Nasm预先定义的宏__LINE__ */
                        if (!strcmp("__LINE__", m->name))
                        {
                            nasm_free(tline->text);
                            make_tok_num(tline, src_get_linnum());
                            continue;
                        }
                        /* 空宏 */
                        tline = delete_Token(tline);
                        continue;
                    }
                }
                else
                {
                    /*
                     * Complicated case: at least one macro with this name
                     * exists and takes parameters. We must find the
                     * parameters in the call, count them, find the SMacro
                     * that corresponds to that form of the macro call, and
                     * substitute for the parameters when we expand. What a
                     * pain.
                     */
                    /* tline = tline->next;
                       skip_white_(tline); */
                    do {
                        t = tline->next;
                        /* ????? */
                        while (tok_type_(t, TOK_SMAC_END)) {
                            t->mac->in_progress = FALSE;
                            t->text = NULL;
                            t = tline->next = delete_Token(t);
                        }
                        tline = t;
                    } while (tok_type_(tline, TOK_WHITESPACE));
                    if (!tok_is_(tline, "("))
                    {
                        /*
                         * This macro wasn't called with parameters: ignore
                         * the call. (Behaviour borrowed from gnu cpp.)
                         */
                        tline = mstart;
                        m = NULL;
                    }
                    else /* '(',处理宏参数 */
                    {
                        int paren = 0;
                        int white = 0;
                        brackets = 0;
                        nparam = 0;
                        sparam = PARAM_DELTA;
                        params = nasm_malloc(sparam * sizeof(Token *));
                        params[0] = tline->next;
                        paramsize = nasm_malloc(sparam * sizeof(int));
                        paramsize[0] = 0;
                        /* 正式处理宏参数，将宏参数放入params中，并且计算宏参数的个数 */
                        while (TRUE) { /* parameter loop */
                            /*
                             * For some unusual expansions
                             * which concatenates function call
                             */
                            t = tline->next;
                            /* ????? */
                            while (tok_type_(t, TOK_SMAC_END)) {
                                t->mac->in_progress = FALSE;
                                t->text = NULL;
                                t = tline->next = delete_Token(t);
                            }
                            tline = t;

                            if (!tline) {
                                error(ERR_NONFATAL, "macro call expects terminating `)'");
                                break;
                            }
                            if (tline->type == TOK_WHITESPACE && brackets <= 0) {
                                if (paramsize[nparam])
                                    white++;
                                else
                                    params[nparam] = tline->next;
                                continue; /* parameter loop */
                            }
                            if (tline->type == TOK_OTHER && tline->text[1] == 0) /* 判断是否是单字符 */
                            {
                                char ch = tline->text[0];
                                if (ch == ',' && !paren && brackets <= 0) { /* 处理完一个参数 */
                                    if (++nparam >= sparam) { /* 如果原有的数据不够，重新分配缓存 */
                                        sparam += PARAM_DELTA;
                                        params = nasm_realloc(params, sparam * sizeof(Token *));
                                        paramsize = nasm_realloc(paramsize, sparam * sizeof(int));
                                    }
                                    params[nparam] = tline->next;
                                    paramsize[nparam] = 0;
                                    white = 0;
                                    continue; /* parameter loop */
                                }
                                if (ch == '{' && (brackets > 0 || (brackets == 0 && !paramsize[nparam])))
                                {
                                    if (!(brackets++)) {
                                        params[nparam] = tline->next; /* 去掉'{' */
                                        continue; /* parameter loop */
                                    }
                                }
                                if (ch == '}' && brackets > 0)
                                    if (--brackets == 0) {
                                        brackets = -1; /* 单行宏参数中不能使用'{' 和 '}',去掉 '}' */
                                        continue; /* parameter loop */
                                    }
                                if (ch == '(' && !brackets)
                                    paren++;
                                if (ch == ')' && brackets <= 0)
                                    if (--paren < 0)
                                        break;
                            }
                            if (brackets < 0) {
                                brackets = 0;
                                error(ERR_NONFATAL, "braces do not enclose all of macro parameter");
                            }
                            paramsize[nparam] += white + 1;
                            white = 0;
                        } /* parameter loop */
                        nparam++;
                        while (m && (m->nparam != nparam || mstrcmp(m->name, mname, m->casesense))) /* 处理同名宏不同参数的情况，找到相应的宏 */
                            m = m->next;
                        if (!m)
                            error(ERR_WARNING | ERR_WARN_MNP, "macro `%s' exists, but not taking %d parameters", mstart->text, nparam);
                    }
                }
                if (m && m->in_progress)
                    m = NULL;
                if (!m) { /* in progess or didn't find '(' or wrong nparam */
                    /*
                     * Design question: should we handle !tline, which
                     * indicates missing ')' here, or expand those
                     * macros anyway, which requires the (t) test a few
                     * lines down?
                     */
                    nasm_free(params);
                    nasm_free(paramsize);
                    tline = mstart;
                } else {
                    /*
                     * Expand the macro: we are placed on the last token of the
                     * call, so that we can easily split the call from the
                     * following tokens. We also start by pushing an SMAC_END
                     * token for the cycle removal.
                     */
                    t = tline;
                    if (t) {
                        tline = t->next;
                        t->next = NULL;
                    }
                    tt = new_Token(tline, TOK_SMAC_END, NULL, 0);
                    tt->mac = m;
                    m->in_progress = TRUE;
                    tline = tt;
                    /* 真实的参数扩展 */
                    for (t = m->expansion; t; t = t->next) {
                        if (t->type >= TOK_SMAC_PARAM) {
                            Token *pcopy = tline, **ptail = &pcopy;
                            Token *ttt, *pt;
                            int i;

                            ttt = params[t->type - TOK_SMAC_PARAM]; /* 获取实参 */
                            for (i = paramsize[t->type - TOK_SMAC_PARAM]; --i >= 0;) {
                                pt = *ptail = new_Token(tline, ttt->type, ttt->text, 0);
                                ptail = &pt->next;
                                ttt = ttt->next;
                            }
                            tline = pcopy;
                        } else {
                            tt = new_Token(tline, t->type, t->text, 0);
                            tline = tt;
                        }
                    }

                    /*
                     * Having done that, get rid of the macro call, and clean
                     * up the parameters.
                     */
                    nasm_free(params);
                    nasm_free(paramsize);
                    free_tlist(mstart);
                    continue; /* main token loop */
                }
            }
        }

        if (tline->type == TOK_SMAC_END) {
            tline->mac->in_progress = FALSE;
            tline = delete_Token(tline);
        } else {
            t = *tail = tline;
            tline = tline->next;
            t->mac = NULL;
            t->next = NULL;
            tail = &t->next;
        }
    }

    /*
     * Now scan the entire line and look for successive TOK_IDs that resulted
     * after expansion (they can't be produced by tokenise()). The successive
     * TOK_IDs should be concatenated.
     * Also we look for %+ tokens and concatenate the tokens before and after
     * them (without white spaces in between).
     */
    t = thead;
    rescan = 0;
    while (t) {
        while (t && t->type != TOK_ID && t->type != TOK_PREPROC_ID)
            t = t->next;
        if (!t || !t->next)
            break;
        /*
         * %define ADD(a,b) ((a) + (b))
         * %define ADD(a,b,c) ((a %+ DD(1,2)) + (b) + (c))
         *
         * [bits 32]
         * _start:
         * mov eax,ADD(1,3)
         * mov eax,ADD(A,2,3)
         */
        if (t->next->type == TOK_ID ||
            t->next->type == TOK_PREPROC_ID ||
            t->next->type == TOK_NUMBER) {
            char *p = nasm_strcat(t->text, t->next->text);
            nasm_free(t->text);
            t->next = delete_Token(t->next);
            t->text = p;
            rescan = 1;
        } else if (t->next->type == TOK_WHITESPACE && t->next->next &&
                   t->next->next->type == TOK_PREPROC_ID &&
                   strcmp(t->next->next->text, "%+") == 0) {
            /* free the next whitespace, the %+ token and next whitespace */
            int i;
            for (i = 1; i <= 3; i++) {
                if (!t->next
                    || (i != 2 && t->next->type != TOK_WHITESPACE))
                    break;
                t->next = delete_Token(t->next);
            } /* endfor */
        } else
            t = t->next;
    }
    /* If we concatenaded something, re-scan the line for macros */
    if (rescan) {
        tline = thead;
        goto again;
    }

    if (org_tline) {
        if (thead) {
            *org_tline = *thead;
            /* since we just gave text to org_line, don't free it */
            thead->text = NULL;
            delete_Token(thead);
        } else {
            /* the expression expanded to empty line;
               we can't return NULL for some reasons
               we just set the line to a single WHITESPACE token. */
            memset(org_tline, 0, sizeof(*org_tline));
            org_tline->text = NULL;
            org_tline->type = TOK_WHITESPACE;
        }
        thead = org_tline;
    }

    return thead;
}