diff options
Diffstat (limited to 'bfd/hash.c')
-rw-r--r-- | bfd/hash.c | 289 |
1 files changed, 282 insertions, 7 deletions
@@ -16,7 +16,7 @@ GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software -Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ +Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #include "bfd.h" #include "sysdep.h" @@ -234,8 +234,12 @@ EXAMPLE . {* Allocate the structure if it has not already been allocated by a . derived class. *} . if (ret == (@var{entry_type} *) NULL) -. ret = ((@var{entry_type} *) -. bfd_hash_allocate (table, sizeof (@var{entry_type}))); +. { +. ret = ((@var{entry_type} *) +. bfd_hash_allocate (table, sizeof (@var{entry_type}))); +. if (ret == (@var{entry_type} *) NULL) +. return NULL; +. } . . {* Call the allocation method of the base class. *} . ret = ((@var{entry_type} *) @@ -311,14 +315,14 @@ bfd_hash_table_init_n (table, newfunc, size) alloc = size * sizeof (struct bfd_hash_entry *); if (!obstack_begin (&table->memory, alloc)) { - bfd_error = no_memory; + bfd_set_error (bfd_error_no_memory); return false; } table->table = ((struct bfd_hash_entry **) obstack_alloc (&table->memory, alloc)); if (!table->table) { - bfd_error = no_memory; + bfd_set_error (bfd_error_no_memory); return false; } memset ((PTR) table->table, 0, alloc); @@ -399,7 +403,7 @@ bfd_hash_lookup (table, string, create, copy) new = (char *) obstack_alloc (&table->memory, len + 1); if (!new) { - bfd_error = no_memory; + bfd_set_error (bfd_error_no_memory); return (struct bfd_hash_entry *) NULL; } strcpy (new, string); @@ -413,6 +417,32 @@ bfd_hash_lookup (table, string, create, copy) return hashp; } +/* Replace an entry in a hash table. */ + +void +bfd_hash_replace (table, old, nw) + struct bfd_hash_table *table; + struct bfd_hash_entry *old; + struct bfd_hash_entry *nw; +{ + unsigned int index; + struct bfd_hash_entry **pph; + + index = old->hash % table->size; + for (pph = &table->table[index]; + (*pph) != (struct bfd_hash_entry *) NULL; + pph = &(*pph)->next) + { + if (*pph == old) + { + *pph = nw; + return; + } + } + + abort (); +} + /* Base method for creating a new hash table entry. */ /*ARGSUSED*/ @@ -435,7 +465,12 @@ bfd_hash_allocate (table, size) struct bfd_hash_table *table; unsigned int size; { - return obstack_alloc (&table->memory, size); + PTR ret; + + ret = obstack_alloc (&table->memory, size); + if (ret == NULL && size != 0) + bfd_set_error (bfd_error_no_memory); + return ret; } /* Traverse a hash table. */ @@ -459,3 +494,243 @@ bfd_hash_traverse (table, func, info) } } } + +/* A few different object file formats (a.out, COFF, ELF) use a string + table. These functions support adding strings to a string table, + returning the byte offset, and writing out the table. + + Possible improvements: + + look for strings matching trailing substrings of other strings + + better data structures? balanced trees? + + look at reducing memory use elsewhere -- maybe if we didn't have + to construct the entire symbol table at once, we could get by + with smaller amounts of VM? (What effect does that have on the + string table reductions?) */ + +/* An entry in the strtab hash table. */ + +struct strtab_hash_entry +{ + struct bfd_hash_entry root; + /* Index in string table. */ + bfd_size_type index; + /* Next string in strtab. */ + struct strtab_hash_entry *next; +}; + +/* The strtab hash table. */ + +struct bfd_strtab_hash +{ + struct bfd_hash_table table; + /* Size of strtab--also next available index. */ + bfd_size_type size; + /* First string in strtab. */ + struct strtab_hash_entry *first; + /* Last string in strtab. */ + struct strtab_hash_entry *last; + /* Whether to precede strings with a two byte length, as in the + XCOFF .debug section. */ + boolean xcoff; +}; + +static struct bfd_hash_entry *strtab_hash_newfunc + PARAMS ((struct bfd_hash_entry *, struct bfd_hash_table *, const char *)); + +/* Routine to create an entry in a strtab. */ + +static struct bfd_hash_entry * +strtab_hash_newfunc (entry, table, string) + struct bfd_hash_entry *entry; + struct bfd_hash_table *table; + const char *string; +{ + struct strtab_hash_entry *ret = (struct strtab_hash_entry *) entry; + + /* Allocate the structure if it has not already been allocated by a + subclass. */ + if (ret == (struct strtab_hash_entry *) NULL) + ret = ((struct strtab_hash_entry *) + bfd_hash_allocate (table, sizeof (struct strtab_hash_entry))); + if (ret == (struct strtab_hash_entry *) NULL) + return NULL; + + /* Call the allocation method of the superclass. */ + ret = ((struct strtab_hash_entry *) + bfd_hash_newfunc ((struct bfd_hash_entry *) ret, table, string)); + + if (ret) + { + /* Initialize the local fields. */ + ret->index = (bfd_size_type) -1; + ret->next = NULL; + } + + return (struct bfd_hash_entry *) ret; +} + +/* Look up an entry in an strtab. */ + +#define strtab_hash_lookup(t, string, create, copy) \ + ((struct strtab_hash_entry *) \ + bfd_hash_lookup (&(t)->table, (string), (create), (copy))) + +/* Create a new strtab. */ + +struct bfd_strtab_hash * +_bfd_stringtab_init () +{ + struct bfd_strtab_hash *table; + + table = (struct bfd_strtab_hash *) malloc (sizeof (struct bfd_strtab_hash)); + if (table == NULL) + { + bfd_set_error (bfd_error_no_memory); + return NULL; + } + + if (! bfd_hash_table_init (&table->table, strtab_hash_newfunc)) + { + free (table); + return NULL; + } + + table->size = 0; + table->first = NULL; + table->last = NULL; + table->xcoff = false; + + return table; +} + +/* Create a new strtab in which the strings are output in the format + used in the XCOFF .debug section: a two byte length precedes each + string. */ + +struct bfd_strtab_hash * +_bfd_xcoff_stringtab_init () +{ + struct bfd_strtab_hash *ret; + + ret = _bfd_stringtab_init (); + if (ret != NULL) + ret->xcoff = true; + return ret; +} + +/* Free a strtab. */ + +void +_bfd_stringtab_free (table) + struct bfd_strtab_hash *table; +{ + bfd_hash_table_free (&table->table); + free (table); +} + +/* Get the index of a string in a strtab, adding it if it is not + already present. If HASH is false, we don't really use the hash + table, and we don't eliminate duplicate strings. */ + +bfd_size_type +_bfd_stringtab_add (tab, str, hash, copy) + struct bfd_strtab_hash *tab; + const char *str; + boolean hash; + boolean copy; +{ + register struct strtab_hash_entry *entry; + + if (hash) + { + entry = strtab_hash_lookup (tab, str, true, copy); + if (entry == NULL) + return (bfd_size_type) -1; + } + else + { + entry = ((struct strtab_hash_entry *) + bfd_hash_allocate (&tab->table, + sizeof (struct strtab_hash_entry))); + if (entry == NULL) + return (bfd_size_type) -1; + if (! copy) + entry->root.string = str; + else + { + char *n; + + n = (char *) bfd_hash_allocate (&tab->table, strlen (str) + 1); + if (n == NULL) + return (bfd_size_type) -1; + entry->root.string = n; + } + entry->index = (bfd_size_type) -1; + entry->next = NULL; + } + + if (entry->index == (bfd_size_type) -1) + { + entry->index = tab->size; + tab->size += strlen (str) + 1; + if (tab->xcoff) + { + entry->index += 2; + tab->size += 2; + } + if (tab->first == NULL) + tab->first = entry; + else + tab->last->next = entry; + tab->last = entry; + } + + return entry->index; +} + +/* Get the number of bytes in a strtab. */ + +bfd_size_type +_bfd_stringtab_size (tab) + struct bfd_strtab_hash *tab; +{ + return tab->size; +} + +/* Write out a strtab. ABFD must already be at the right location in + the file. */ + +boolean +_bfd_stringtab_emit (abfd, tab) + register bfd *abfd; + struct bfd_strtab_hash *tab; +{ + register boolean xcoff; + register struct strtab_hash_entry *entry; + + xcoff = tab->xcoff; + + for (entry = tab->first; entry != NULL; entry = entry->next) + { + register const char *str; + register size_t len; + + str = entry->root.string; + len = strlen (str) + 1; + + if (xcoff) + { + bfd_byte buf[2]; + + /* The output length includes the null byte. */ + bfd_put_16 (abfd, len, buf); + if (bfd_write ((PTR) buf, 1, 2, abfd) != 2) + return false; + } + + if (bfd_write ((PTR) str, 1, len, abfd) != len) + return false; + } + + return true; +} |