mir/bind
2
0
Fork 0
mirror of https://gitlab.isc.org/isc-projects/bind9 synced 2025-08-28 21:17:54 +00:00
Code Issues Releases Wiki Activity

While searching in dns_rbt_findnode(), cope with the possibility that

Browse Source 
a bitstring label might be founding things from any label of a
multilabel search_name.
This commit is contained in:
David Lawrence 2001-05-31 11:03:33 +00:00
parent ea70ca20b2
commit b79e66b06d
1 changed files with 37 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

61
lib/dns/rbt.c
View File

@ -15,7 +15,7 @@
* WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
/* $Id: rbt.c,v 1.111 2001/04/12 21:18:14 tale Exp $ */
/* $Id: rbt.c,v 1.112 2001/05/31 11:03:33 tale Exp $ */
/* Principal Authors: DCL */
@ -858,6 +858,7 @@ dns_rbt_findnode(dns_rbt_t *rbt, dns_name_t *name, dns_name_t *foundname,
unsigned int nlabels;
unsigned int tlabels = 1;
unsigned int hash;
isc_boolean_t has_bitstring = ISC_FALSE;
if (rbt->hashtable == NULL)
goto nohash;
@ -916,24 +917,37 @@ dns_rbt_findnode(dns_rbt_t *rbt, dns_name_t *name, dns_name_t *foundname,
}
}
if (tlabels++ < nlabels)
if (tlabels++ < nlabels) {
/*
* XXXDCL Bitstring labels complicate things,
* as usual. Checking for the situation could
* be done up by the dns_name_getlabelsequence
* so that they could still use the hashing
* code, but it would be messy to repeatedly
* try various bitstring lengths. Instead
* just notice when a bitstring label is
* involved and then punt to the traditional
* binary search if no hash node is found
* after all of the labels are tried.
*/
if (has_bitstring == ISC_FALSE &&
hash_name.ndata[0] ==
DNS_LABELTYPE_BITSTRING)
has_bitstring = ISC_TRUE;
goto hashagain;
}
/*
* XXXDCL Bitstring labels complicate things, as usual.
* Checking for the situation could be done up by
* the dns_name_getlabelsequence so that they could
* still use the hashing code, but it would be messy
* to repeatedly try various bitstring lengths.
* Best to keep the issue out of the way down here for
* now by punting to the traditional binary search.
*
* Yes, accessing the name structure data directly
* is evil. This function gets used to much to slow
* it down with the name.h function call APIs.
* All of the labels have been tried against the hash
* table. If there wasn't a bitstring label involved,
* the name isn't in the table. If there was, fall
* through to the traditional search algorithm.
*/
if (search_name->ndata[search_name->offsets[nlabels-1]]
!= DNS_LABELTYPE_BITSTRING) {
if (! has_bitstring) {
/*
* Done with the search.
*/
current = NULL;
continue;
}
@ -1323,13 +1337,13 @@ dns_rbt_deletename(dns_rbt_t *rbt, dns_name_t *name, isc_boolean_t recurse) {
result = dns_rbt_findnode(rbt, name, NULL, &node, NULL,
DNS_RBTFIND_NOOPTIONS, NULL, NULL);
if (result == ISC_R_SUCCESS)
if (result == ISC_R_SUCCESS) {
if (DATA(node) != NULL)
result = dns_rbt_deletenode(rbt, node, recurse);
else
result = ISC_R_NOTFOUND;
else if (result == DNS_R_PARTIALMATCH)
} else if (result == DNS_R_PARTIALMATCH)
result = ISC_R_NOTFOUND;
return (result);
@ -1357,20 +1371,19 @@ dns_rbt_deletename(dns_rbt_t *rbt, dns_name_t *name, isc_boolean_t recurse) {
* "b.c" is left hanging around without data or children. This condition
* is actually pretty easy to detect, but ... should it really be removed?
* Is a chain pointing to it? An iterator? Who knows! (Note that the
* references structure member because it is private to rbtdb.) This is
* ugly and makes me unhappy, but after hours of trying to make it more
* aesthetically proper and getting nowhere, this is the way it is going
* to stay until such time as it proves to be a *real* problem.
* references structure member cannot be looked at because it is private to
* rbtdb.) This is ugly and makes me unhappy, but after hours of trying to
* make it more aesthetically proper and getting nowhere, this is the way it
* is going to stay until such time as it proves to be a *real* problem.
*
* Finally, for reference, note that the original routine that did node
* joining was called join_nodes(). It has been excised, living now only
* in the CVS histroy, but comments have been left behind that point to it just
* in the CVS history, but comments have been left behind that point to it just
* in case someone wants to muck with this some more.
*
* The one positive aspect of all of this is that joining used to have a
* case where it might fail. Without trying to join, now this function always
* succeeds. It still returns isc_result_t, though, so the API wouldn't change.
*/
* succeeds. It still returns isc_result_t, though, so the API wouldn't change. */
isc_result_t
dns_rbt_deletenode(dns_rbt_t *rbt, dns_rbtnode_t *node, isc_boolean_t recurse)
{