| /* Textual dumping of CTF data. | |
| Copyright (C) 2019-2023 Free Software Foundation, Inc. | |
| This file is part of libctf. | |
| libctf 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; see the file COPYING. If not see | |
| <http://www.gnu.org/licenses/>. */ | |
| #include <ctf-impl.h> | |
| #include <string.h> | |
| #define str_append(s, a) ctf_str_append_noerr (s, a) | |
| /* One item to be dumped, in string form. */ | |
| typedef struct ctf_dump_item | |
| { | |
| ctf_list_t cdi_list; | |
| char *cdi_item; | |
| } ctf_dump_item_t; | |
| /* Cross-call state for dumping. Basically just enough to track the section in | |
| use and a list of return strings. */ | |
| struct ctf_dump_state | |
| { | |
| ctf_sect_names_t cds_sect; | |
| ctf_dict_t *cds_fp; | |
| ctf_dump_item_t *cds_current; | |
| ctf_list_t cds_items; | |
| }; | |
| /* Cross-call state for ctf_dump_member. */ | |
| typedef struct ctf_dump_membstate | |
| { | |
| char **cdm_str; | |
| ctf_dict_t *cdm_fp; | |
| const char *cdm_toplevel_indent; | |
| } ctf_dump_membstate_t; | |
| static int | |
| ctf_dump_append (ctf_dump_state_t *state, char *str) | |
| { | |
| ctf_dump_item_t *cdi; | |
| if ((cdi = malloc (sizeof (struct ctf_dump_item))) == NULL) | |
| return (ctf_set_errno (state->cds_fp, ENOMEM)); | |
| cdi->cdi_item = str; | |
| ctf_list_append (&state->cds_items, cdi); | |
| return 0; | |
| } | |
| static void | |
| ctf_dump_free (ctf_dump_state_t *state) | |
| { | |
| ctf_dump_item_t *cdi, *next_cdi; | |
| if (state == NULL) | |
| return; | |
| for (cdi = ctf_list_next (&state->cds_items); cdi != NULL; | |
| cdi = next_cdi) | |
| { | |
| free (cdi->cdi_item); | |
| next_cdi = ctf_list_next (cdi); | |
| free (cdi); | |
| } | |
| } | |
| /* Return a dump for a single type, without member info: but do optionally show | |
| the type's references. */ | |
| #define CTF_FT_REFS 0x2 /* Print referenced types. */ | |
| #define CTF_FT_BITFIELD 0x4 /* Print :BITS if a bitfield. */ | |
| #define CTF_FT_ID 0x8 /* Print "ID: " in front of type IDs. */ | |
| static char * | |
| ctf_dump_format_type (ctf_dict_t *fp, ctf_id_t id, int flag) | |
| { | |
| ctf_id_t new_id; | |
| char *str = NULL, *bit = NULL, *buf = NULL; | |
| ctf_set_errno (fp, 0); | |
| new_id = id; | |
| do | |
| { | |
| ctf_encoding_t ep; | |
| ctf_arinfo_t ar; | |
| int kind, unsliced_kind; | |
| ssize_t size, align; | |
| const char *nonroot_leader = ""; | |
| const char *nonroot_trailer = ""; | |
| const char *idstr = ""; | |
| id = new_id; | |
| if (flag == CTF_ADD_NONROOT) | |
| { | |
| nonroot_leader = "{"; | |
| nonroot_trailer = "}"; | |
| } | |
| buf = ctf_type_aname (fp, id); | |
| if (!buf) | |
| { | |
| if (id == 0 || ctf_errno (fp) == ECTF_NONREPRESENTABLE) | |
| { | |
| ctf_set_errno (fp, ECTF_NONREPRESENTABLE); | |
| str = str_append (str, " (type not represented in CTF)"); | |
| return str; | |
| } | |
| goto err; | |
| } | |
| if (flag & CTF_FT_ID) | |
| idstr = "ID "; | |
| if (asprintf (&bit, "%s%s0x%lx: (kind %i) ", nonroot_leader, idstr, | |
| id, ctf_type_kind (fp, id)) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| if (buf[0] != '\0') | |
| str = str_append (str, buf); | |
| free (buf); | |
| buf = NULL; | |
| unsliced_kind = ctf_type_kind_unsliced (fp, id); | |
| kind = ctf_type_kind (fp, id); | |
| /* Report encodings of everything with an encoding other than enums: | |
| base-type enums cannot have a nonzero cte_offset or cte_bits value. | |
| (Slices of them can, but they are of kind CTF_K_SLICE.) */ | |
| if (unsliced_kind != CTF_K_ENUM && ctf_type_encoding (fp, id, &ep) == 0) | |
| { | |
| if ((ssize_t) ep.cte_bits != ctf_type_size (fp, id) * CHAR_BIT | |
| && flag & CTF_FT_BITFIELD) | |
| { | |
| if (asprintf (&bit, ":%i", ep.cte_bits) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| } | |
| if ((ssize_t) ep.cte_bits != ctf_type_size (fp, id) * CHAR_BIT | |
| || ep.cte_offset != 0) | |
| { | |
| const char *slice = ""; | |
| if (unsliced_kind == CTF_K_SLICE) | |
| slice = "slice "; | |
| if (asprintf (&bit, " [%s0x%x:0x%x]", | |
| slice, ep.cte_offset, ep.cte_bits) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| } | |
| if (asprintf (&bit, " (format 0x%x)", ep.cte_format) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| } | |
| size = ctf_type_size (fp, id); | |
| if (kind != CTF_K_FUNCTION && size >= 0) | |
| { | |
| if (asprintf (&bit, " (size 0x%lx)", (unsigned long int) size) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| } | |
| align = ctf_type_align (fp, id); | |
| if (align >= 0) | |
| { | |
| if (asprintf (&bit, " (aligned at 0x%lx)", | |
| (unsigned long int) align) < 0) | |
| goto oom; | |
| str = str_append (str, bit); | |
| free (bit); | |
| bit = NULL; | |
| } | |
| if (nonroot_trailer[0] != 0) | |
| str = str_append (str, nonroot_trailer); | |
| /* Just exit after one iteration if we are not showing the types this type | |
| references. */ | |
| if (!(flag & CTF_FT_REFS)) | |
| return str; | |
| /* Keep going as long as this type references another. We consider arrays | |
| to "reference" their element type. */ | |
| if (kind == CTF_K_ARRAY) | |
| { | |
| if (ctf_array_info (fp, id, &ar) < 0) | |
| goto err; | |
| new_id = ar.ctr_contents; | |
| } | |
| else | |
| new_id = ctf_type_reference (fp, id); | |
| if (new_id != CTF_ERR) | |
| str = str_append (str, " -> "); | |
| } | |
| while (new_id != CTF_ERR); | |
| if (ctf_errno (fp) != ECTF_NOTREF) | |
| { | |
| free (str); | |
| return NULL; | |
| } | |
| return str; | |
| oom: | |
| ctf_set_errno (fp, errno); | |
| err: | |
| ctf_err_warn (fp, 1, 0, _("cannot format name dumping type 0x%lx"), id); | |
| free (buf); | |
| free (str); | |
| free (bit); | |
| return NULL; | |
| } | |
| /* Dump one string field from the file header into the cds_items. */ | |
| static int | |
| ctf_dump_header_strfield (ctf_dict_t *fp, ctf_dump_state_t *state, | |
| const char *name, uint32_t value) | |
| { | |
| char *str; | |
| if (value) | |
| { | |
| if (asprintf (&str, "%s: %s\n", name, ctf_strptr (fp, value)) < 0) | |
| goto err; | |
| ctf_dump_append (state, str); | |
| } | |
| return 0; | |
| err: | |
| return (ctf_set_errno (fp, errno)); | |
| } | |
| /* Dump one section-offset field from the file header into the cds_items. */ | |
| static int | |
| ctf_dump_header_sectfield (ctf_dict_t *fp, ctf_dump_state_t *state, | |
| const char *sect, uint32_t off, uint32_t nextoff) | |
| { | |
| char *str; | |
| if (nextoff - off) | |
| { | |
| if (asprintf (&str, "%s:\t0x%lx -- 0x%lx (0x%lx bytes)\n", sect, | |
| (unsigned long) off, (unsigned long) (nextoff - 1), | |
| (unsigned long) (nextoff - off)) < 0) | |
| goto err; | |
| ctf_dump_append (state, str); | |
| } | |
| return 0; | |
| err: | |
| return (ctf_set_errno (fp, errno)); | |
| } | |
| /* Dump the file header into the cds_items. */ | |
| static int | |
| ctf_dump_header (ctf_dict_t *fp, ctf_dump_state_t *state) | |
| { | |
| char *str; | |
| char *flagstr = NULL; | |
| const ctf_header_t *hp = fp->ctf_header; | |
| const char *vertab[] = | |
| { | |
| NULL, "CTF_VERSION_1", | |
| "CTF_VERSION_1_UPGRADED_3 (latest format, version 1 type " | |
| "boundaries)", | |
| "CTF_VERSION_2", | |
| "CTF_VERSION_3", NULL | |
| }; | |
| const char *verstr = NULL; | |
| if (asprintf (&str, "Magic number: 0x%x\n", hp->cth_magic) < 0) | |
| goto err; | |
| ctf_dump_append (state, str); | |
| if (hp->cth_version <= CTF_VERSION) | |
| verstr = vertab[hp->cth_version]; | |
| if (verstr == NULL) | |
| verstr = "(not a valid version)"; | |
| if (asprintf (&str, "Version: %i (%s)\n", hp->cth_version, | |
| verstr) < 0) | |
| goto err; | |
| ctf_dump_append (state, str); | |
| /* Everything else is only printed if present. */ | |
| /* The flags are unusual in that they represent the ctf_dict_t *in memory*: | |
| flags representing compression, etc, are turned off as the file is | |
| decompressed. So we store a copy of the flags before they are changed, for | |
| the dumper. */ | |
| if (fp->ctf_openflags > 0) | |
| { | |
| if (asprintf (&flagstr, "%s%s%s%s%s%s%s", | |
| fp->ctf_openflags & CTF_F_COMPRESS | |
| ? "CTF_F_COMPRESS": "", | |
| (fp->ctf_openflags & CTF_F_COMPRESS) | |
| && (fp->ctf_openflags & ~CTF_F_COMPRESS) | |
| ? ", " : "", | |
| fp->ctf_openflags & CTF_F_NEWFUNCINFO | |
| ? "CTF_F_NEWFUNCINFO" : "", | |
| (fp->ctf_openflags & (CTF_F_COMPRESS | CTF_F_NEWFUNCINFO)) | |
| && (fp->ctf_openflags & ~(CTF_F_COMPRESS | CTF_F_NEWFUNCINFO)) | |
| ? ", " : "", | |
| fp->ctf_openflags & CTF_F_IDXSORTED | |
| ? "CTF_F_IDXSORTED" : "", | |
| fp->ctf_openflags & (CTF_F_COMPRESS | CTF_F_NEWFUNCINFO | |
| | CTF_F_IDXSORTED) | |
| && (fp->ctf_openflags & ~(CTF_F_COMPRESS | CTF_F_NEWFUNCINFO | |
| | CTF_F_IDXSORTED)) | |
| ? ", " : "", | |
| fp->ctf_openflags & CTF_F_DYNSTR | |
| ? "CTF_F_DYNSTR" : "") < 0) | |
| goto err; | |
| if (asprintf (&str, "Flags: 0x%x (%s)", fp->ctf_openflags, flagstr) < 0) | |
| goto err; | |
| ctf_dump_append (state, str); | |
| } | |
| if (ctf_dump_header_strfield (fp, state, "Parent label", | |
| hp->cth_parlabel) < 0) | |
| goto err; | |
| if (ctf_dump_header_strfield (fp, state, "Parent name", hp->cth_parname) < 0) | |
| goto err; | |
| if (ctf_dump_header_strfield (fp, state, "Compilation unit name", | |
| hp->cth_cuname) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Label section", hp->cth_lbloff, | |
| hp->cth_objtoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Data object section", | |
| hp->cth_objtoff, hp->cth_funcoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Function info section", | |
| hp->cth_funcoff, hp->cth_objtidxoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Object index section", | |
| hp->cth_objtidxoff, hp->cth_funcidxoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Function index section", | |
| hp->cth_funcidxoff, hp->cth_varoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Variable section", | |
| hp->cth_varoff, hp->cth_typeoff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "Type section", | |
| hp->cth_typeoff, hp->cth_stroff) < 0) | |
| goto err; | |
| if (ctf_dump_header_sectfield (fp, state, "String section", hp->cth_stroff, | |
| hp->cth_stroff + hp->cth_strlen + 1) < 0) | |
| goto err; | |
| return 0; | |
| err: | |
| free (flagstr); | |
| return (ctf_set_errno (fp, errno)); | |
| } | |
| /* Dump a single label into the cds_items. */ | |
| static int | |
| ctf_dump_label (const char *name, const ctf_lblinfo_t *info, | |
| void *arg) | |
| { | |
| char *str; | |
| char *typestr; | |
| ctf_dump_state_t *state = arg; | |
| if (asprintf (&str, "%s -> ", name) < 0) | |
| return (ctf_set_errno (state->cds_fp, errno)); | |
| if ((typestr = ctf_dump_format_type (state->cds_fp, info->ctb_type, | |
| CTF_ADD_ROOT | CTF_FT_REFS)) == NULL) | |
| { | |
| free (str); | |
| return 0; /* Swallow the error. */ | |
| } | |
| str = str_append (str, typestr); | |
| free (typestr); | |
| ctf_dump_append (state, str); | |
| return 0; | |
| } | |
| /* Dump all the object or function entries into the cds_items. */ | |
| static int | |
| ctf_dump_objts (ctf_dict_t *fp, ctf_dump_state_t *state, int functions) | |
| { | |
| const char *name; | |
| ctf_id_t id; | |
| ctf_next_t *i = NULL; | |
| char *str = NULL; | |
| if ((functions && fp->ctf_funcidx_names) | |
| || (!functions && fp->ctf_objtidx_names)) | |
| str = str_append (str, _("Section is indexed.\n")); | |
| else if (fp->ctf_symtab.cts_data == NULL) | |
| str = str_append (str, _("No symbol table.\n")); | |
| while ((id = ctf_symbol_next (fp, &i, &name, functions)) != CTF_ERR) | |
| { | |
| char *typestr = NULL; | |
| /* Emit the name, if we know it. No trailing space: ctf_dump_format_type | |
| has a leading one. */ | |
| if (name) | |
| { | |
| if (asprintf (&str, "%s -> ", name) < 0) | |
| goto oom; | |
| } | |
| else | |
| str = xstrdup (""); | |
| if ((typestr = ctf_dump_format_type (state->cds_fp, id, | |
| CTF_ADD_ROOT | CTF_FT_REFS)) == NULL) | |
| { | |
| ctf_dump_append (state, str); | |
| continue; /* Swallow the error. */ | |
| } | |
| str = str_append (str, typestr); | |
| free (typestr); | |
| ctf_dump_append (state, str); | |
| continue; | |
| oom: | |
| ctf_set_errno (fp, ENOMEM); | |
| ctf_next_destroy (i); | |
| return -1; | |
| } | |
| return 0; | |
| } | |
| /* Dump a single variable into the cds_items. */ | |
| static int | |
| ctf_dump_var (const char *name, ctf_id_t type, void *arg) | |
| { | |
| char *str; | |
| char *typestr; | |
| ctf_dump_state_t *state = arg; | |
| if (asprintf (&str, "%s -> ", name) < 0) | |
| return (ctf_set_errno (state->cds_fp, errno)); | |
| if ((typestr = ctf_dump_format_type (state->cds_fp, type, | |
| CTF_ADD_ROOT | CTF_FT_REFS)) == NULL) | |
| { | |
| free (str); | |
| return 0; /* Swallow the error. */ | |
| } | |
| str = str_append (str, typestr); | |
| free (typestr); | |
| ctf_dump_append (state, str); | |
| return 0; | |
| } | |
| /* Dump a single struct/union member into the string in the membstate. */ | |
| static int | |
| ctf_dump_member (const char *name, ctf_id_t id, unsigned long offset, | |
| int depth, void *arg) | |
| { | |
| ctf_dump_membstate_t *state = arg; | |
| char *typestr = NULL; | |
| char *bit = NULL; | |
| /* The struct/union itself has already been printed. */ | |
| if (depth == 0) | |
| return 0; | |
| if (asprintf (&bit, "%s%*s", state->cdm_toplevel_indent, (depth-1)*4, "") < 0) | |
| goto oom; | |
| *state->cdm_str = str_append (*state->cdm_str, bit); | |
| free (bit); | |
| if ((typestr = ctf_dump_format_type (state->cdm_fp, id, | |
| CTF_ADD_ROOT | CTF_FT_BITFIELD | |
| | CTF_FT_ID)) == NULL) | |
| return -1; /* errno is set for us. */ | |
| if (asprintf (&bit, "[0x%lx] %s: %s\n", offset, name, typestr) < 0) | |
| goto oom; | |
| *state->cdm_str = str_append (*state->cdm_str, bit); | |
| free (typestr); | |
| free (bit); | |
| typestr = NULL; | |
| bit = NULL; | |
| return 0; | |
| oom: | |
| free (typestr); | |
| free (bit); | |
| return (ctf_set_errno (state->cdm_fp, errno)); | |
| } | |
| /* Report the number of digits in the hexadecimal representation of a type | |
| ID. */ | |
| static int | |
| type_hex_digits (ctf_id_t id) | |
| { | |
| int i = 0; | |
| if (id == 0) | |
| return 1; | |
| for (; id > 0; id >>= 4, i++); | |
| return i; | |
| } | |
| /* Dump a single type into the cds_items. */ | |
| static int | |
| ctf_dump_type (ctf_id_t id, int flag, void *arg) | |
| { | |
| char *str; | |
| char *indent; | |
| ctf_dump_state_t *state = arg; | |
| ctf_dump_membstate_t membstate = { &str, state->cds_fp, NULL }; | |
| /* Indent neatly. */ | |
| if (asprintf (&indent, " %*s", type_hex_digits (id), "") < 0) | |
| return (ctf_set_errno (state->cds_fp, ENOMEM)); | |
| /* Dump the type itself. */ | |
| if ((str = ctf_dump_format_type (state->cds_fp, id, | |
| flag | CTF_FT_REFS)) == NULL) | |
| goto err; | |
| str = str_append (str, "\n"); | |
| membstate.cdm_toplevel_indent = indent; | |
| /* Member dumping for structs, unions... */ | |
| if (ctf_type_kind (state->cds_fp, id) == CTF_K_STRUCT | |
| || ctf_type_kind (state->cds_fp, id) == CTF_K_UNION) | |
| { | |
| if ((ctf_type_visit (state->cds_fp, id, ctf_dump_member, &membstate)) < 0) | |
| { | |
| if (id == 0 || ctf_errno (state->cds_fp) == ECTF_NONREPRESENTABLE) | |
| { | |
| ctf_dump_append (state, str); | |
| return 0; | |
| } | |
| ctf_err_warn (state->cds_fp, 1, ctf_errno (state->cds_fp), | |
| _("cannot visit members dumping type 0x%lx"), id); | |
| goto err; | |
| } | |
| } | |
| /* ... and enums, for which we dump the first and last few members and skip | |
| the ones in the middle. */ | |
| if (ctf_type_kind (state->cds_fp, id) == CTF_K_ENUM) | |
| { | |
| int enum_count = ctf_member_count (state->cds_fp, id); | |
| ctf_next_t *it = NULL; | |
| int i = 0; | |
| const char *enumerand; | |
| char *bit; | |
| int value; | |
| while ((enumerand = ctf_enum_next (state->cds_fp, id, | |
| &it, &value)) != NULL) | |
| { | |
| i++; | |
| if ((i > 5) && (i < enum_count - 4)) | |
| continue; | |
| str = str_append (str, indent); | |
| if (asprintf (&bit, "%s: %i\n", enumerand, value) < 0) | |
| { | |
| ctf_next_destroy (it); | |
| goto oom; | |
| } | |
| str = str_append (str, bit); | |
| free (bit); | |
| if ((i == 5) && (enum_count > 10)) | |
| { | |
| str = str_append (str, indent); | |
| str = str_append (str, "...\n"); | |
| } | |
| } | |
| if (ctf_errno (state->cds_fp) != ECTF_NEXT_END) | |
| { | |
| ctf_err_warn (state->cds_fp, 1, ctf_errno (state->cds_fp), | |
| _("cannot visit enumerands dumping type 0x%lx"), id); | |
| goto err; | |
| } | |
| } | |
| ctf_dump_append (state, str); | |
| free (indent); | |
| return 0; | |
| err: | |
| free (indent); | |
| free (str); | |
| /* Swallow the error: don't cause an error in one type to abort all | |
| type dumping. */ | |
| return 0; | |
| oom: | |
| free (indent); | |
| free (str); | |
| return ctf_set_errno (state->cds_fp, ENOMEM); | |
| } | |
| /* Dump the string table into the cds_items. */ | |
| static int | |
| ctf_dump_str (ctf_dict_t *fp, ctf_dump_state_t *state) | |
| { | |
| const char *s = fp->ctf_str[CTF_STRTAB_0].cts_strs; | |
| for (; s < fp->ctf_str[CTF_STRTAB_0].cts_strs + | |
| fp->ctf_str[CTF_STRTAB_0].cts_len;) | |
| { | |
| char *str; | |
| if (asprintf (&str, "0x%lx: %s", | |
| (unsigned long) (s - fp->ctf_str[CTF_STRTAB_0].cts_strs), | |
| s) < 0) | |
| return (ctf_set_errno (fp, errno)); | |
| ctf_dump_append (state, str); | |
| s += strlen (s) + 1; | |
| } | |
| return 0; | |
| } | |
| /* Dump a particular section of a CTF file, in textual form. Call with a | |
| pointer to a NULL STATE: each call emits a dynamically allocated string | |
| containing a description of one entity in the specified section, in order. | |
| Only the first call (with a NULL state) may vary SECT. Once the CTF section | |
| has been entirely dumped, the call returns NULL and frees and annuls the | |
| STATE, ready for another section to be dumped. The returned textual content | |
| may span multiple lines: between each call the FUNC is called with one | |
| textual line at a time, and should return a suitably decorated line (it can | |
| allocate a new one and return it if it likes). */ | |
| char * | |
| ctf_dump (ctf_dict_t *fp, ctf_dump_state_t **statep, ctf_sect_names_t sect, | |
| ctf_dump_decorate_f *func, void *arg) | |
| { | |
| char *str; | |
| char *line; | |
| ctf_dump_state_t *state = NULL; | |
| if (*statep == NULL) | |
| { | |
| /* Data collection. Transforming a call-at-a-time iterator into a | |
| return-at-a-time iterator in a language without call/cc is annoying. It | |
| is easiest to simply collect everything at once and then return it bit | |
| by bit. The first call will take (much) longer than otherwise, but the | |
| amortized time needed is the same. */ | |
| if ((*statep = malloc (sizeof (struct ctf_dump_state))) == NULL) | |
| { | |
| ctf_set_errno (fp, ENOMEM); | |
| goto end; | |
| } | |
| state = *statep; | |
| memset (state, 0, sizeof (struct ctf_dump_state)); | |
| state->cds_fp = fp; | |
| state->cds_sect = sect; | |
| switch (sect) | |
| { | |
| case CTF_SECT_HEADER: | |
| ctf_dump_header (fp, state); | |
| break; | |
| case CTF_SECT_LABEL: | |
| if (ctf_label_iter (fp, ctf_dump_label, state) < 0) | |
| { | |
| if (ctf_errno (fp) != ECTF_NOLABELDATA) | |
| goto end; /* errno is set for us. */ | |
| ctf_set_errno (fp, 0); | |
| } | |
| break; | |
| case CTF_SECT_OBJT: | |
| if (ctf_dump_objts (fp, state, 0) < 0) | |
| goto end; /* errno is set for us. */ | |
| break; | |
| case CTF_SECT_FUNC: | |
| if (ctf_dump_objts (fp, state, 1) < 0) | |
| goto end; /* errno is set for us. */ | |
| break; | |
| case CTF_SECT_VAR: | |
| if (ctf_variable_iter (fp, ctf_dump_var, state) < 0) | |
| goto end; /* errno is set for us. */ | |
| break; | |
| case CTF_SECT_TYPE: | |
| if (ctf_type_iter_all (fp, ctf_dump_type, state) < 0) | |
| goto end; /* errno is set for us. */ | |
| break; | |
| case CTF_SECT_STR: | |
| ctf_dump_str (fp, state); | |
| break; | |
| default: | |
| ctf_set_errno (fp, ECTF_DUMPSECTUNKNOWN); | |
| goto end; | |
| } | |
| } | |
| else | |
| { | |
| state = *statep; | |
| if (state->cds_sect != sect) | |
| { | |
| ctf_set_errno (fp, ECTF_DUMPSECTCHANGED); | |
| goto end; | |
| } | |
| } | |
| if (state->cds_current == NULL) | |
| state->cds_current = ctf_list_next (&state->cds_items); | |
| else | |
| state->cds_current = ctf_list_next (state->cds_current); | |
| if (state->cds_current == NULL) | |
| goto end; | |
| /* Hookery. There is some extra complexity to preserve linefeeds within each | |
| item while removing linefeeds at the end. */ | |
| if (func) | |
| { | |
| size_t len; | |
| str = NULL; | |
| for (line = state->cds_current->cdi_item; line && *line; ) | |
| { | |
| char *nline = line; | |
| char *ret; | |
| nline = strchr (line, '\n'); | |
| if (nline) | |
| nline[0] = '\0'; | |
| ret = func (sect, line, arg); | |
| str = str_append (str, ret); | |
| str = str_append (str, "\n"); | |
| if (ret != line) | |
| free (ret); | |
| if (nline) | |
| { | |
| nline[0] = '\n'; | |
| nline++; | |
| } | |
| line = nline; | |
| } | |
| len = strlen (str); | |
| if (str[len-1] == '\n') | |
| str[len-1] = '\0'; | |
| } | |
| else | |
| { | |
| str = strdup (state->cds_current->cdi_item); | |
| if (!str) | |
| { | |
| ctf_set_errno (fp, ENOMEM); | |
| return str; | |
| } | |
| } | |
| ctf_set_errno (fp, 0); | |
| return str; | |
| end: | |
| ctf_dump_free (state); | |
| free (state); | |
| ctf_set_errno (fp, 0); | |
| *statep = NULL; | |
| return NULL; | |
| } |