Hi! On top of the https://gcc.gnu.org/pipermail/gcc-patches/2024-June/655012.html https://gcc.gnu.org/pipermail/gcc-patches/2024-June/655013.html patches here is a new version of the patch which adds another #embed extension, gnu::base64.
As mentioned in the documentation, this extension is primarily intended for use by the preprocessor, so that for the larger (say 32+ or 64+ bytes long embeds it doesn't have to emit tens of thousands or millions of comma separated string literals which would be very expensive to parse again, but can emit #embed "." __gnu__::__base64__( \ "Tm9uIGVyYW0gbsOpc2NpdXMsIEJydXRlLCBjdW0sIHF1w6Ygc3VtbWlzIGluZ8OpbmlpcyBleHF1" \ "aXNpdMOhcXVlIGRvY3Ryw61uYSBwaGlsw7Nzb3BoaSBHcsOmY28gc2VybcOzbmUgdHJhY3RhdsOt" \ "c3NlbnQsIGVhIExhdMOtbmlzIGzDrXR0ZXJpcyBtYW5kYXLDqW11cywgZm9yZSB1dCBoaWMgbm9z" \ "dGVyIGxhYm9yIGluIHbDoXJpYXMgcmVwcmVoZW5zacOzbmVzIGluY8O6cnJlcmV0LiBuYW0gcXVp" \ "YsO6c2RhbSwgZXQgaWlzIHF1aWRlbSBub24gw6FkbW9kdW0gaW5kw7NjdGlzLCB0b3R1bSBob2Mg" \ "ZMOtc3BsaWNldCBwaGlsb3NvcGjDoXJpLiBxdWlkYW0gYXV0ZW0gbm9uIHRhbSBpZCByZXByZWjD" \ "qW5kdW50LCBzaSByZW3DrXNzaXVzIGFnw6F0dXIsIHNlZCB0YW50dW0gc3TDumRpdW0gdGFtcXVl" \ "IG11bHRhbSDDs3BlcmFtIHBvbsOpbmRhbSBpbiBlbyBub24gYXJiaXRyw6FudHVyLiBlcnVudCDD" \ "qXRpYW0sIGV0IGlpIHF1aWRlbSBlcnVkw610aSBHcsOmY2lzIGzDrXR0ZXJpcywgY29udGVtbsOp" \ "bnRlcyBMYXTDrW5hcywgcXVpIHNlIGRpY2FudCBpbiBHcsOmY2lzIGxlZ8OpbmRpcyDDs3BlcmFt" \ "IG1hbGxlIGNvbnPDum1lcmUuIHBvc3Ryw6ltbyDDoWxpcXVvcyBmdXTDunJvcyBzw7pzcGljb3Is" \ "IHF1aSBtZSBhZCDDoWxpYXMgbMOtdHRlcmFzIHZvY2VudCwgZ2VudXMgaG9jIHNjcmliw6luZGks" \ "IGV0c2kgc2l0IGVsw6lnYW5zLCBwZXJzw7Nuw6YgdGFtZW4gZXQgZGlnbml0w6F0aXMgZXNzZSBu" \ "ZWdlbnQu") with the meaning don't actually load some file, instead base64 decode (RFC4648 with A-Za-z0-9+/ chars and = padding, no newlines in between) the string and use that as data. This is chosen because it should be -pedantic-errors clean, fairly cheap to decode and then in optimizing compiler could be handled as similar binary blob to normal #embed, while the data isn't left somewhere on the disk, so distcc/ccache etc. can move the preprocessed source without issues. It makes no sense to support limit and gnu::offset parameters together with it IMHO, why would somebody waste providing full data and then threw some away? prefix/suffix/if_empty are normally supported though, but not intended to be used by the preprocessor. This patch adds just the extension side, not the actual emitting of this during -E or -E -fdirectives-only for now, that will be included in the upcoming patch. Compared to the earlier posted version of this extension, this patch allows the string concatenation in the parameter argument (but still doesn't allow escapes in the string, why would anyone use them when only A-Za-z0-9+/= are valid). The patch also adds support for parsing this even in -fpreprocessed compilation. Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? 2024-07-11 Jakub Jelinek <ja...@redhat.com> libcpp/ * internal.h (struct cpp_embed_params): Add base64 member. (_cpp_free_embed_params_tokens): Declare. * directives.cc (DIRECTIVE_TABLE): Add IN_I flag to T_EMBED. (save_token_for_embed, _cpp_free_embed_params_tokens): New functions. (_cpp_parse_embed_params): Parse gnu::base64 parameter. If -fpreprocessed without -fdirectives-only, require #embed to have gnu::base64 parameter. Diagnose conflict between gnu::base64 and limit or gnu::offset parameters. (do_embed): Use _cpp_free_embed_params_tokens. * files.cc (finish_embed, base64_dec_fn): New functions. (base64_dec): New array. (B64D0, B64D1, B64D2, B64D3): Define. (finish_base64_embed): New function. (_cpp_stack_embed): Use finish_embed. Handle params->base64 using finish_base64_embed. * macro.cc (builtin_has_embed): Call _cpp_free_embed_params_tokens. gcc/ * doc/cpp.texi (Binary Resource Inclusion): Document gnu::base64 parameter. gcc/testsuite/ * c-c++-common/cpp/embed-17.c: New test. * c-c++-common/cpp/embed-18.c: New test. * c-c++-common/cpp/embed-19.c: New test. * gcc.dg/cpp/embed-6.c: New test. * gcc.dg/cpp/embed-7.c: New test. --- libcpp/internal.h.jj 2024-07-03 14:51:56.151028840 +0200 +++ libcpp/internal.h 2024-07-03 14:52:12.230817498 +0200 @@ -631,7 +631,7 @@ struct cpp_embed_params location_t loc; bool has_embed; cpp_num_part limit, offset; - cpp_embed_params_tokens prefix, suffix, if_empty; + cpp_embed_params_tokens prefix, suffix, if_empty, base64; }; /* Character classes. Based on the more primitive macros in safe-ctype.h. @@ -805,6 +805,7 @@ extern void _cpp_restore_pragma_names (c extern int _cpp_do__Pragma (cpp_reader *, location_t); extern void _cpp_init_directives (cpp_reader *); extern void _cpp_init_internal_pragmas (cpp_reader *); +extern void _cpp_free_embed_params_tokens (cpp_embed_params_tokens *); extern bool _cpp_parse_embed_params (cpp_reader *, struct cpp_embed_params *); extern void _cpp_do_file_change (cpp_reader *, enum lc_reason, const char *, linenum_type, unsigned int); --- libcpp/directives.cc.jj 2024-07-03 14:51:56.151028840 +0200 +++ libcpp/directives.cc 2024-07-11 13:17:53.898269972 +0200 @@ -159,7 +159,7 @@ static void cpp_pop_definition (cpp_read D(error, T_ERROR, STDC89, 0) \ D(pragma, T_PRAGMA, STDC89, IN_I) \ D(warning, T_WARNING, STDC23, 0) \ - D(embed, T_EMBED, STDC23, INCL | EXPAND) \ + D(embed, T_EMBED, STDC23, IN_I | INCL | EXPAND) \ D(include_next, T_INCLUDE_NEXT, EXTENSION, INCL | EXPAND) \ D(ident, T_IDENT, EXTENSION, IN_I) \ D(import, T_IMPORT, EXTENSION, INCL | EXPAND) /* ObjC */ \ @@ -932,6 +932,50 @@ do_include_next (cpp_reader *pfile) do_include_common (pfile, type); } +/* Helper function for skip_balanced_token_seq and _cpp_parse_embed_params. + Save one token *TOKEN into *SAVE. */ + +static void +save_token_for_embed (cpp_embed_params_tokens *save, const cpp_token *token) +{ + if (save->count == 0) + { + _cpp_init_tokenrun (&save->base_run, 4); + save->cur_run = &save->base_run; + save->cur_token = save->base_run.base; + } + else if (save->cur_token == save->cur_run->limit) + { + save->cur_run->next = XNEW (tokenrun); + save->cur_run->next->prev = save->cur_run; + _cpp_init_tokenrun (save->cur_run->next, 4); + save->cur_run = save->cur_run->next; + save->cur_token = save->cur_run->base; + } + *save->cur_token = *token; + save->cur_token->flags |= NO_EXPAND; + save->cur_token++; + save->count++; +} + +/* Free memory associated with saved tokens in *SAVE. */ + +void +_cpp_free_embed_params_tokens (cpp_embed_params_tokens *save) +{ + if (save->count == 0) + return; + tokenrun *n; + for (tokenrun *t = &save->base_run; t; t = n) + { + n = t->next; + XDELETEVEC (t->base); + if (t != &save->base_run) + XDELETE (t); + } + save->count = 0; +} + /* Skip over balanced preprocessing tokens until END is found. If SAVE is non-NULL, remember the parsed tokens in it. */ @@ -959,26 +1003,7 @@ skip_balanced_token_seq (cpp_reader *pfi if (save && (token->type != CPP_PADDING || save->count) && (token->type != end || nested)) - { - if (save->count == 0) - { - _cpp_init_tokenrun (&save->base_run, 4); - save->cur_run = &save->base_run; - save->cur_token = save->base_run.base; - } - else if (save->cur_token == save->cur_run->limit) - { - save->cur_run->next = XNEW (tokenrun); - save->cur_run->next->prev = save->cur_run; - _cpp_init_tokenrun (save->cur_run->next, 4); - save->cur_run = save->cur_run->next; - save->cur_token = save->cur_run->base; - } - *save->cur_token = *token; - save->cur_token->flags |= NO_EXPAND; - save->cur_token++; - save->count++; - } + save_token_for_embed (save, token); if (token->type == end) return; switch (token->type) @@ -1032,12 +1057,31 @@ _cpp_parse_embed_params (cpp_reader *pfi cpp_error (pfile, CPP_DL_ERROR, "expected ')'"); return false; } - return ret; } - else if (token->type == CPP_CLOSE_PAREN && params->has_embed) - return ret; - cpp_error (pfile, CPP_DL_ERROR, "expected parameter name"); - return false; + else if (token->type != CPP_CLOSE_PAREN || !params->has_embed) + { + cpp_error (pfile, CPP_DL_ERROR, "expected parameter name"); + return false; + } + if (params->base64.count && (seen & ((1 << 0) | (1 << 5))) != 0) + { + ret = false; + if (!params->has_embed) + cpp_error_with_line (pfile, CPP_DL_ERROR, + params->base64.base_run.base->src_loc, 0, + "'gnu::base64' parameter conflicts with " + "'limit' or 'gnu::offset' parameters"); + } + else if (params->base64.count == 0 + && CPP_OPTION (pfile, preprocessed)) + { + ret = false; + if (!params->has_embed) + cpp_error_with_line (pfile, CPP_DL_ERROR, params->loc, 0, + "'gnu::base64' parameter required in " + "preprocessed source"); + } + return ret; } param_name = NODE_NAME (token->val.node.spelling); param_name_len = NODE_LEN (token->val.node.spelling); @@ -1113,7 +1157,7 @@ _cpp_parse_embed_params (cpp_reader *pfi else if (param_prefix_len == 3 && memcmp (param_prefix, "gnu", 3) == 0) { struct { int len; const char *name; } gnu_params[] = { - { 6, "offset" } + { 6, "base64" }, { 6, "offset" } }; for (size_t i = 0; i < sizeof (gnu_params) / sizeof (gnu_params[0]); ++i) @@ -1153,7 +1197,7 @@ _cpp_parse_embed_params (cpp_reader *pfi if (param_kind != (size_t) -1 && token->type != CPP_OPEN_PAREN) cpp_error_with_line (pfile, CPP_DL_ERROR, loc, 0, "expected '('"); - else if (param_kind == 0 || param_kind == 4) + else if (param_kind == 0 || param_kind == 5) { if (params->has_embed && pfile->op_stack == NULL) _cpp_expand_op_stack (pfile); @@ -1170,6 +1214,53 @@ _cpp_parse_embed_params (cpp_reader *pfi } token = _cpp_get_token_no_padding (pfile); } + else if (param_kind == 4) + { + token = _cpp_get_token_no_padding (pfile); + while (token->type == CPP_OTHER + && CPP_OPTION (pfile, preprocessed) + && !CPP_OPTION (pfile, directives_only) + && token->val.str.len == 1 + && token->val.str.text[0] == '\\') + { + /* Allow backslash newline inside of gnu::base64 argument + for -fpreprocessed, so that it doesn't have to be + megabytes long line. */ + pfile->state.in_directive = 0; + token = _cpp_get_token_no_padding (pfile); + pfile->state.in_directive = 3; + } + if (token->type == CPP_STRING) + { + do + { + save_token_for_embed (¶ms->base64, token); + token = _cpp_get_token_no_padding (pfile); + while (token->type == CPP_OTHER + && CPP_OPTION (pfile, preprocessed) + && !CPP_OPTION (pfile, directives_only) + && token->val.str.len == 1 + && token->val.str.text[0] == '\\') + { + pfile->state.in_directive = 0; + token = _cpp_get_token_no_padding (pfile); + pfile->state.in_directive = 3; + } + } + while (token->type == CPP_STRING); + if (token->type != CPP_CLOSE_PAREN) + cpp_error_with_line (pfile, CPP_DL_ERROR, token->src_loc, 0, + "expected ')'"); + } + else + { + cpp_error_with_line (pfile, CPP_DL_ERROR, token->src_loc, 0, + "expected character string literal"); + if (token->type != CPP_CLOSE_PAREN) + token = _cpp_get_token_no_padding (pfile); + } + token = _cpp_get_token_no_padding (pfile); + } else if (token->type == CPP_OPEN_PAREN) { cpp_embed_params_tokens *save = NULL; @@ -1250,26 +1341,10 @@ do_embed (cpp_reader *pfile) if (ok) _cpp_stack_embed (pfile, fname, angle_brackets, ¶ms); - for (int i = 0; i < 3; ++i) - { - cpp_embed_params_tokens *p; - if (i == 0) - p = ¶ms.prefix; - else if (i == 1) - p = ¶ms.suffix; - else - p = ¶ms.if_empty; - if (p->count == 0) - continue; - tokenrun *n; - for (tokenrun *t = &p->base_run; t; t = n) - { - n = t->next; - XDELETEVEC (t->base); - if (t != &p->base_run) - XDELETE (t); - } - } + _cpp_free_embed_params_tokens (¶ms.prefix); + _cpp_free_embed_params_tokens (¶ms.suffix); + _cpp_free_embed_params_tokens (¶ms.if_empty); + _cpp_free_embed_params_tokens (¶ms.base64); done: XDELETEVEC (fname); --- libcpp/files.cc.jj 2024-07-03 14:51:56.152028827 +0200 +++ libcpp/files.cc 2024-07-11 14:19:09.744438223 +0200 @@ -1221,6 +1221,320 @@ cpp_probe_header_unit (cpp_reader *pfile return nullptr; } +/* Helper function for _cpp_stack_embed. Finish #embed/__has_embed processing + after a file is found and data loaded into buffer. */ + +static int +finish_embed (cpp_reader *pfile, _cpp_file *file, + struct cpp_embed_params *params) +{ + const uchar *buffer = file->buffer; + size_t limit = file->limit; + if (params->offset - file->offset > limit) + limit = 0; + else + { + buffer += params->offset - file->offset; + limit -= params->offset - file->offset; + } + if (params->limit < limit) + limit = params->limit; + + /* For sizes larger than say 64 bytes, this is just a temporary + solution, we should emit a single new token which the FEs will + handle as an optimization. */ + size_t max = INTTYPE_MAXIMUM (size_t) / sizeof (cpp_token); + if (limit > max / 2 + || (limit + ? (params->prefix.count > max + || params->suffix.count > max + || (limit * 2 - 1 + params->prefix.count + + params->suffix.count > max)) + : params->if_empty.count > max)) + { + cpp_error_at (pfile, CPP_DL_ERROR, params->loc, + "%s is too large", file->path); + return 0; + } + + size_t len = 0; + for (size_t i = 0; i < limit; ++i) + { + if (buffer[i] < 10) + len += 2; + else if (buffer[i] < 100) + len += 3; +#if UCHAR_MAX == 255 + else + len += 4; +#else + else if (buffer[i] < 1000) + len += 4; + else + { + char buf[64]; + len += sprintf (buf, "%d", buffer[i]) + 1; + } +#endif + if (len > INTTYPE_MAXIMUM (ssize_t)) + { + cpp_error_at (pfile, CPP_DL_ERROR, params->loc, + "%s is too large", file->path); + return 0; + } + } + uchar *s = len ? _cpp_unaligned_alloc (pfile, len) : NULL; + _cpp_buff *tok_buff = NULL; + cpp_token *toks = NULL, *tok = &pfile->directive_result; + size_t count = 0; + if (limit) + count = (params->prefix.count + limit * 2 - 1 + + params->suffix.count) - 1; + else if (params->if_empty.count) + count = params->if_empty.count - 1; + if (count) + { + tok_buff = _cpp_get_buff (pfile, count * sizeof (cpp_token)); + toks = (cpp_token *) tok_buff->base; + } + cpp_embed_params_tokens *prefix + = limit ? ¶ms->prefix : ¶ms->if_empty; + if (prefix->count) + { + *tok = *prefix->base_run.base; + tok = toks; + tokenrun *cur_run = &prefix->base_run; + while (cur_run) + { + size_t cnt = (cur_run->next ? cur_run->limit + : prefix->cur_token) - cur_run->base; + cpp_token *t = cur_run->base; + if (cur_run == &prefix->base_run) + { + t++; + cnt--; + } + memcpy (tok, t, cnt * sizeof (cpp_token)); + tok += cnt; + cur_run = cur_run->next; + } + } + for (size_t i = 0; i < limit; ++i) + { + tok->src_loc = params->loc; + tok->type = CPP_NUMBER; + tok->flags = NO_EXPAND; + if (i == 0) + tok->flags |= PREV_WHITE; + tok->val.str.text = s; + tok->val.str.len = sprintf ((char *) s, "%d", buffer[i]); + s += tok->val.str.len + 1; + if (tok == &pfile->directive_result) + tok = toks; + else + tok++; + if (i < limit - 1) + { + tok->src_loc = params->loc; + tok->type = CPP_COMMA; + tok->flags = NO_EXPAND; + tok++; + } + } + if (limit && params->suffix.count) + { + tokenrun *cur_run = ¶ms->suffix.base_run; + cpp_token *orig_tok = tok; + while (cur_run) + { + size_t cnt = (cur_run->next ? cur_run->limit + : params->suffix.cur_token) - cur_run->base; + cpp_token *t = cur_run->base; + memcpy (tok, t, cnt * sizeof (cpp_token)); + tok += cnt; + cur_run = cur_run->next; + } + orig_tok->flags |= PREV_WHITE; + } + pfile->directive_result.flags |= PREV_WHITE; + if (count) + { + _cpp_push_token_context (pfile, NULL, toks, count); + pfile->context->buff = tok_buff; + } + return limit ? 1 : 2; +} + +/* Helper function for initialization of base64_dec table. + Can't rely on ASCII compatibility, so check each letter + separately. */ + +constexpr signed char +base64_dec_fn (unsigned char c) +{ + return (c == 'A' ? 0 : c == 'B' ? 1 : c == 'C' ? 2 : c == 'D' ? 3 + : c == 'E' ? 4 : c == 'F' ? 5 : c == 'G' ? 6 : c == 'H' ? 7 + : c == 'I' ? 8 : c == 'J' ? 9 : c == 'K' ? 10 : c == 'L' ? 11 + : c == 'M' ? 12 : c == 'N' ? 13 : c == 'O' ? 14 : c == 'P' ? 15 + : c == 'Q' ? 16 : c == 'R' ? 17 : c == 'S' ? 18 : c == 'T' ? 19 + : c == 'U' ? 20 : c == 'V' ? 21 : c == 'W' ? 22 : c == 'X' ? 23 + : c == 'Y' ? 24 : c == 'Z' ? 25 + : c == 'a' ? 26 : c == 'b' ? 27 : c == 'c' ? 28 : c == 'd' ? 29 + : c == 'e' ? 30 : c == 'f' ? 31 : c == 'g' ? 32 : c == 'h' ? 33 + : c == 'i' ? 34 : c == 'j' ? 35 : c == 'k' ? 36 : c == 'l' ? 37 + : c == 'm' ? 38 : c == 'n' ? 39 : c == 'o' ? 40 : c == 'p' ? 41 + : c == 'q' ? 42 : c == 'r' ? 43 : c == 's' ? 44 : c == 't' ? 45 + : c == 'u' ? 46 : c == 'v' ? 47 : c == 'w' ? 48 : c == 'x' ? 49 + : c == 'y' ? 50 : c == 'z' ? 51 + : c == '0' ? 52 : c == '1' ? 53 : c == '2' ? 54 : c == '3' ? 55 + : c == '4' ? 56 : c == '5' ? 57 : c == '6' ? 58 : c == '7' ? 59 + : c == '8' ? 60 : c == '9' ? 61 : c == '+' ? 62 : c == '/' ? 63 + : -1); +} + +/* base64 decoding table. */ + +static constexpr signed char base64_dec[] = { +#define B64D0(x) base64_dec_fn (x) +#define B64D1(x) B64D0 (x), B64D0 (x + 1), B64D0 (x + 2), B64D0 (x + 3) +#define B64D2(x) B64D1 (x), B64D1 (x + 4), B64D1 (x + 8), B64D1 (x + 12) +#define B64D3(x) B64D2 (x), B64D2 (x + 16), B64D2 (x + 32), B64D2 (x + 48) + B64D3 (0), B64D3 (64), B64D3 (128), B64D3 (192) +}; + +/* Helper function for _cpp_stack_embed. Handle #embed/__has_embed with + gnu::base64 parameter. */ + +static int +finish_base64_embed (cpp_reader *pfile, const char *fname, bool angle, + struct cpp_embed_params *params) +{ + size_t len, end, i, j, base64_len = 0, cnt; + uchar *buf = NULL, *q, pbuf[4], qbuf[3]; + const uchar *base64_str; + if (angle || strcmp (fname, ".")) + { + if (!params->has_embed) + cpp_error_at (pfile, CPP_DL_ERROR, params->loc, + "'gnu::base64' parameter can be only used with \".\""); + return 0; + } + tokenrun *cur_run = ¶ms->base64.base_run; + cpp_token *tend, *tok; + while (cur_run) + { + tend = cur_run->next ? cur_run->limit : params->base64.cur_token; + for (tok = cur_run->base; tok < tend; ++tok) + { + if (tok->val.str.len < 2 + || tok->val.str.text[0] != '"' + || tok->val.str.text[tok->val.str.len - 1] != '"') + { + fail: + cpp_error_at (pfile, CPP_DL_ERROR, params->loc, + "'gnu::base64' argument not valid base64 " + "encoded string"); + free (buf); + return 0; + } + if (tok->val.str.len - 2 > (~(size_t) 0) - base64_len) + goto fail; + base64_len += tok->val.str.len - 2; + } + cur_run = cur_run->next; + } + if ((base64_len & 3) != 0) + goto fail; + len = base64_len / 4 * 3; + end = len; + + if (params->has_embed) + q = qbuf; + else + { + buf = XNEWVEC (uchar, len ? len : 1); + q = buf; + } + cur_run = ¶ms->base64.base_run; + tend = cur_run->next ? cur_run->limit : params->base64.cur_token; + tok = cur_run->base; + base64_str = tok->val.str.text + 1; + cnt = tok->val.str.len - 2; + ++tok; + for (i = 0; i < end; i += 3) + { + for (j = 0; j < 4; ++j) + { + while (cnt == 0) + { + if (tok == tend) + { + cur_run = cur_run->next; + tend = (cur_run->next ? cur_run->limit + : params->base64.cur_token); + tok = cur_run->base; + } + base64_str = tok->val.str.text + 1; + cnt = tok->val.str.len - 2; + ++tok; + } + pbuf[j] = *base64_str; + base64_str++; + --cnt; + } + if (pbuf[3] == '=' && i + 3 >= end) + { + end = len - 3; + --len; + if (pbuf[2] == '=') + --len; + break; + } + int a = base64_dec[pbuf[0]]; + int b = base64_dec[pbuf[1]]; + int c = base64_dec[pbuf[2]]; + int d = base64_dec[pbuf[3]]; + if (a == -1 || b == -1 || c == -1 || d == -1) + goto fail; + q[0] = (a << 2) | (b >> 4); + q[1] = (b << 4) | (c >> 2); + q[2] = (c << 6) | d; + if (!params->has_embed) + q += 3; + } + if (len != end) + { + int a = base64_dec[pbuf[0]]; + int b = base64_dec[pbuf[1]]; + if (a == -1 || b == -1) + goto fail; + q[0] = (a << 2) | (b >> 4); + if (len - end == 2) + { + int c = base64_dec[pbuf[2]]; + if (c == -1) + goto fail; + q[1] = (b << 4) | (c >> 2); + if ((c & 3) != 0) + goto fail; + } + else if ((b & 15) != 0) + goto fail; + } + if (params->has_embed) + return len ? 1 : 2; + _cpp_file *file = make_cpp_file (NULL, ""); + file->embed = 1; + file->next_file = pfile->all_files; + pfile->all_files = file; + params->limit = -1; + params->offset = 0; + file->limit = len; + file->buffer = buf; + file->path = xstrdup ("<base64>"); + return finish_embed (pfile, file, params); +} + /* Try to load FNAME with #embed/__has_embed parameters PARAMS. If !PARAMS->has_embed, return new token in pfile->directive_result (first token) and rest in a pushed non-macro context. @@ -1231,6 +1545,8 @@ int _cpp_stack_embed (cpp_reader *pfile, const char *fname, bool angle, struct cpp_embed_params *params) { + if (params->base64.count) + return finish_base64_embed (pfile, fname, angle, params); cpp_dir *dir = search_path_head (pfile, fname, angle, IT_EMBED, params->has_embed); if (!dir) @@ -1450,141 +1766,7 @@ _cpp_stack_embed (cpp_reader *pfile, con return limit && params->limit ? 1 : 2; } - const uchar *buffer = file->buffer; - size_t limit = file->limit; - if (params->offset - file->offset > limit) - limit = 0; - else - { - buffer += params->offset - file->offset; - limit -= params->offset - file->offset; - } - if (params->limit < limit) - limit = params->limit; - - /* For sizes larger than say 64 bytes, this is just a temporary - solution, we should emit a single new token which the FEs will - handle as an optimization. */ - size_t max = INTTYPE_MAXIMUM (size_t) / sizeof (cpp_token); - if (limit > max / 2 - || (limit - ? (params->prefix.count > max - || params->suffix.count > max - || (limit * 2 + params->prefix.count - + params->suffix.count > max)) - : params->if_empty.count > max)) - { - cpp_error_at (pfile, CPP_DL_ERROR, params->loc, - "%s is too large", file->path); - return 0; - } - - size_t len = 0; - for (size_t i = 0; i < limit; ++i) - { - if (buffer[i] < 10) - len += 2; - else if (buffer[i] < 100) - len += 3; -#if UCHAR_MAX == 255 - else - len += 4; -#else - else if (buffer[i] < 1000) - len += 4; - else - { - char buf[64]; - len += sprintf (buf, "%d", buffer[i]) + 1; - } -#endif - if (len > INTTYPE_MAXIMUM (ssize_t)) - { - cpp_error_at (pfile, CPP_DL_ERROR, params->loc, - "%s is too large", file->path); - return 0; - } - } - uchar *s = len ? _cpp_unaligned_alloc (pfile, len) : NULL; - _cpp_buff *tok_buff = NULL; - cpp_token *toks = NULL, *tok = &pfile->directive_result; - size_t count = 0; - if (limit) - count = (params->prefix.count + limit * 2 - 1 - + params->suffix.count) - 1; - else if (params->if_empty.count) - count = params->if_empty.count - 1; - if (count) - { - tok_buff = _cpp_get_buff (pfile, count * sizeof (cpp_token)); - toks = (cpp_token *) tok_buff->base; - } - cpp_embed_params_tokens *prefix - = limit ? ¶ms->prefix : ¶ms->if_empty; - if (prefix->count) - { - *tok = *prefix->base_run.base; - tok = toks; - tokenrun *cur_run = &prefix->base_run; - while (cur_run) - { - size_t cnt = (cur_run->next ? cur_run->limit - : prefix->cur_token) - cur_run->base; - cpp_token *t = cur_run->base; - if (cur_run == &prefix->base_run) - { - t++; - cnt--; - } - memcpy (tok, t, cnt * sizeof (cpp_token)); - tok += cnt; - cur_run = cur_run->next; - } - } - for (size_t i = 0; i < limit; ++i) - { - tok->src_loc = params->loc; - tok->type = CPP_NUMBER; - tok->flags = NO_EXPAND; - if (i == 0) - tok->flags |= PREV_WHITE; - tok->val.str.text = s; - tok->val.str.len = sprintf ((char *) s, "%d", buffer[i]); - s += tok->val.str.len + 1; - if (tok == &pfile->directive_result) - tok = toks; - else - tok++; - if (i < limit - 1) - { - tok->src_loc = params->loc; - tok->type = CPP_COMMA; - tok->flags = NO_EXPAND; - tok++; - } - } - if (limit && params->suffix.count) - { - tokenrun *cur_run = ¶ms->suffix.base_run; - cpp_token *orig_tok = tok; - while (cur_run) - { - size_t cnt = (cur_run->next ? cur_run->limit - : params->suffix.cur_token) - cur_run->base; - cpp_token *t = cur_run->base; - memcpy (tok, t, cnt * sizeof (cpp_token)); - tok += cnt; - cur_run = cur_run->next; - } - orig_tok->flags |= PREV_WHITE; - } - pfile->directive_result.flags |= PREV_WHITE; - if (count) - { - _cpp_push_token_context (pfile, NULL, toks, count); - pfile->context->buff = tok_buff; - } - return limit ? 1 : 2; + return finish_embed (pfile, file, params); } /* Retrofit the just-entered main file asif it was an include. This --- libcpp/macro.cc.jj 2024-07-03 14:51:47.630140833 +0200 +++ libcpp/macro.cc 2024-07-03 14:52:12.231817485 +0200 @@ -505,6 +505,8 @@ builtin_has_embed (cpp_reader *pfile) if (ok && !pfile->state.skip_eval) result = _cpp_stack_embed (pfile, fname, bracket, ¶ms); + _cpp_free_embed_params_tokens (¶ms.base64); + XDELETEVEC (fname); } else if (paren) --- gcc/doc/cpp.texi.jj 2024-07-03 14:51:56.153028814 +0200 +++ gcc/doc/cpp.texi 2024-07-03 14:52:12.232817472 +0200 @@ -3967,7 +3967,8 @@ with currently supported standard parame @code{suffix} and @code{if_empty}, or implementation defined parameters specified by a unique vendor prefix followed by @code{::} followed by name of the parameter. GCC uses the @code{gnu} prefix for vendor -parameters and currently supports the @code{gnu::offset} parameter. +parameters and currently supports the @code{gnu::offset} and +@code{gnu::base64} parameters. The @code{limit} parameter argument is a constant expression which specifies the maximum number of bytes included by the directive, @@ -3981,6 +3982,17 @@ The @code{gnu::offset} parameter argumen which specifies how many bytes to skip from the start of the resource. @code{limit} is then counted from that position. +The @code{gnu::base64} parameter argument is a possibly concatenated +character string literal with base64 encoded data. See +@uref{https://datatracker.ietf.org/doc/html/rfc4648#section-4}. There +should be no newlines in the string literal and because this parameter +is meant namely for use by the preprocessor itself, there is no support +for any escape sequences in the string literal argument. If @code{gnu::base64} +parameter is specified, the @code{limit} and @code{gnu::offset} parameters +should not be specified and the filename should be always @code{"."}. +Instead of reading a file the directive will decode the base64 encoded +data and use that as the data to include. + The @code{#embed} directive is not supported in the Traditional Mode (@pxref{Traditional Mode}). --- gcc/testsuite/c-c++-common/cpp/embed-17.c.jj 2024-07-03 14:52:00.722968749 +0200 +++ gcc/testsuite/c-c++-common/cpp/embed-17.c 2024-07-03 14:52:12.233817459 +0200 @@ -0,0 +1,116 @@ +/* { dg-do run } */ +/* { dg-options "--embed-dir=${srcdir}/c-c++-common/cpp/embed-dir" } */ +/* { dg-additional-options "-std=gnu99" { target c } } */ + +#if __has_embed ("." gnu::base64 ("")) != __STDC_EMBED_EMPTY__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." gnu::base64 ("SA==")) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." prefix(-) suffix (-) if_empty (-) __gnu__::__base64__ ("SA==")) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." gnu::__base64__ ("SGU=")) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." gnu::__base64__ ("SGVs")) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." __gnu__::base64 ("SGVsbG8=")) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +/* M. Tulli Ciceronis De Finibus Bonorum et Malorum. Liber Primus. */ +/* echo "Tm9u....bnQu" | fmt -s -w 76 | base64 -d to decode. */ +#define BONORUM_ET_MALORUM \ +"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" +#if __has_embed ("." gnu::base64 (BONORUM_ET_MALORUM)) != __STDC_EMBED_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("foo" gnu::base64 ("SGU=")) != __STDC_EMBED_NOT_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed (<foo> gnu::base64 ("SGU=")) != __STDC_EMBED_NOT_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed (<.> gnu::base64 ("SGU=")) != __STDC_EMBED_NOT_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." gnu::base64 ("SGU=") limit(5)) != __STDC_EMBED_NOT_FOUND__ +#error "__has_embed fail" +#endif + +#if __has_embed ("." gnu::base64 ("SGU=") gnu::offset(2)) != __STDC_EMBED_NOT_FOUND__ +#error "__has_embed fail" +#endif + +#embed "." gnu::base64 ("") if_empty (int a = 42;) prefix(+ +) suffix (+ +) +#embed "." __gnu__::__base64__ ("SA==") prefix (int b = ) suffix (;) if_empty (+ +) +const unsigned char c[] = { + #embed "." gnu::base64("SGU=") +}; +const unsigned char d[] = { + #embed "." gnu::base64 ("SGVs") +}; +const unsigned char e[] = { + #embed "." gnu::base64 ("SGVsbG8=") +}; +const unsigned char f[] = { +#ifdef __cplusplus + #embed "." gnu::base64 (BONORUM_ET_MALORUM) prefix (' ', ) +#else + #embed "." gnu::base64 (BONORUM_ET_MALORUM) prefix ([1] = ) suffix(, [0] = ' ') +#endif +}; +#if __has_embed ("." gnu::base64("TG9yZW0gaXBzdW0gZG9sb3Igc2l0IGFtZXQsIGNvbnNlY3RldHVyIGFkaXBpc2NpbmcgZWxpdCwg" \ +"c2VkIGRvIGVpdXNtb2QgdGVtcG9yIGluY2lkaWR1bnQgdXQgbGFib3JlIGV0IGRvbG9yZSBtYWdu" \ +"YSBhbGlxdWEuCg==")) == __STDC_EMBED_FOUND__ +const unsigned char g[] = { +#embed "." gnu::base64("" \ +"T" "G9" "yZW" \ +"0gaX" \ +"BzdW0gZG9sb3Igc2l0IGFtZXQsIGNvbnNlY3RldHVyIGFkaXBpc2NpbmcgZWxpdCwg" \ +"c2VkIGRvIGVpdXNtb2QgdGVtcG9yIGluY2lkaWR1bnQgdXQgbGFib3JlIGV0IGRvbG9yZSBtYWdu" \ +"YSBhbGlxdWEuCg==") +}; +#endif + +#ifdef __cplusplus +#define C "C" +#else +#define C +#endif +extern C void abort (void); +extern C int memcmp (const void *, const void *, __SIZE_TYPE__); + +int +main () +{ + if (a != 42 || b != 'H') + abort (); + if (sizeof (c) != 2 || c[0] != 'H' || c[1] != 'e') + abort (); + if (sizeof (d) != 3 || d[0] != 'H' || d[1] != 'e' || d[2] != 'l') + abort (); + if (sizeof (e) != 5 || memcmp (e, "Hello", 5)) + abort (); + if (sizeof (f) != 1 + 747 || memcmp (f, " Non eram néscius, Brute", + sizeof (" Non eram néscius, Brute") - 1)) + abort (); + const char ge[] + = "Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua."; + if (sizeof (g) != sizeof (ge) + || memcmp (g, ge, sizeof (ge) - 1) + || g[sizeof (ge) - 1] != '\n') + abort (); +} --- gcc/testsuite/c-c++-common/cpp/embed-18.c.jj 2024-07-03 14:52:00.722968749 +0200 +++ gcc/testsuite/c-c++-common/cpp/embed-18.c 2024-07-03 14:52:12.233817459 +0200 @@ -0,0 +1,33 @@ +/* { dg-do preprocess } */ +/* { dg-options "" } */ + +#embed "." gnu::base64("") __gnu__::__base64__("") /* { dg-error "duplicate embed parameter 'gnu::base64'" } */ +#embed __FILE__ gnu::base64 prefix() suffix() /* { dg-error "expected '\\\('" } */ +#embed __FILE__ gnu::base64(1) prefix() suffix() /* { dg-error "expected character string literal" } */ +#embed __FILE__ gnu::base64() prefix() suffix() /* { dg-error "expected character string literal" } */ +#embed "." prefix() suffix() gnu::base64("" /* { dg-error "expected '\\\)'" } */ +#embed "." gnu::base64("a") /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#embed "." gnu::base64("----") /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#embed "." gnu::base64("a===") /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#embed "." gnu::base64("TG9yZW0gaXBzdW0gZG9sb3Igc2l0IGFtZXQsIGNvbnNlY3RldHVyIGFkaXBpc2NpbmcgZWxpdCwg\nc2VkIGRvIGVpdXNtb2QgdGVtcG9yIGluY2lkaWR1bnQgdXQgbGFib3JlIGV0IGRvbG9yZSBtYWdu\nYSBhbGlxdWEuCg==") /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#embed "embed-18.c" gnu::base64("SA==") /* { dg-error "'gnu::base64' parameter can be only used with \\\".\\\"" } */ +#embed <embed-18.c> gnu::base64("SA==") /* { dg-error "'gnu::base64' parameter can be only used with \\\".\\\"" } */ +#embed <.> gnu::base64("SA==") /* { dg-error "'gnu::base64' parameter can be only used with \\\".\\\"" } */ +#embed "." gnu::base64("SA==") limit(3) /* { dg-error "'gnu::base64' parameter conflicts with 'limit' or 'gnu::offset' parameters" } */ +#embed "." gnu::base64("SA==") gnu::offset(1) /* { dg-error "'gnu::base64' parameter conflicts with 'limit' or 'gnu::offset' parameters" } */ +#if 1 + __has_embed ("." gnu::base64("") __gnu__::__base64__("")) /* { dg-error "duplicate embed parameter 'gnu::base64'" } */ +#endif +#if 1 + __has_embed (__FILE__ __gnu__::__base64__ prefix() suffix()) /* { dg-error "expected '\\\('" } */ +#endif +#if 1 + __has_embed (__FILE__ __gnu__::__base64__(1) prefix() suffix()) /* { dg-error "expected character string literal" } */ +#endif +#if 1 + __has_embed (__FILE__ gnu::base64() prefix() suffix()) /* { dg-error "expected character string literal" } */ +#endif +#if 1 + __has_embed ("." gnu::base64("a")) /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#endif +#if 1 + __has_embed ("." gnu::base64("----")) /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#endif +#if 1 + __has_embed ("." gnu::base64("a===")) /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#endif +#if 1 + __has_embed ("." gnu::base64("TG9yZW0gaXBzdW0gZG9sb3Igc2l0IGFtZXQsIGNvbnNlY3RldHVyIGFkaXBpc2NpbmcgZWxpdCwg\nc2VkIGRvIGVpdXNtb2QgdGVtcG9yIGluY2lkaWR1bnQgdXQgbGFib3JlIGV0IGRvbG9yZSBtYWdu\nYSBhbGlxdWEuCg==")) /* { dg-error "'gnu::base64' argument not valid base64 encoded string" } */ +#endif --- gcc/testsuite/c-c++-common/cpp/embed-19.c.jj 2024-07-11 13:10:40.644780967 +0200 +++ gcc/testsuite/c-c++-common/cpp/embed-19.c 2024-07-11 13:10:48.661679148 +0200 @@ -0,0 +1,5 @@ +/* { dg-do run } */ +/* { dg-options "--embed-dir=${srcdir}/c-c++-common/cpp/embed-dir -save-temps -fdirectives-only" } */ +/* { dg-additional-options "-std=gnu99" { target c } } */ + +#include "embed-1.c" --- gcc/testsuite/gcc.dg/cpp/embed-6.c.jj 2024-07-11 13:09:59.896298507 +0200 +++ gcc/testsuite/gcc.dg/cpp/embed-6.c 2024-07-11 13:19:37.307954440 +0200 @@ -0,0 +1,6 @@ +/* { dg-do compile } */ +/* { dg-options "-fpreprocessed" } */ + +const unsigned char c[] = { +#embed "embed-6.c" limit (64) /* { dg-error "'gnu::base64' parameter required in preprocessed source" } */ +}; --- gcc/testsuite/gcc.dg/cpp/embed-7.c.jj 2024-07-11 13:19:44.743859845 +0200 +++ gcc/testsuite/gcc.dg/cpp/embed-7.c 2024-07-11 13:19:53.434749284 +0200 @@ -0,0 +1,6 @@ +/* { dg-do compile } */ +/* { dg-options "-fpreprocessed -fdirectives-only" } */ + +const unsigned char c[] = { +#embed "embed-7.c" limit (64) /* { dg-error "'gnu::base64' parameter required in preprocessed source" } */ +}; Jakub