summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorHugo Landau <hlandau@openssl.org>2023-12-08 10:14:27 +0000
committerTomas Mraz <tomas@openssl.org>2023-12-15 12:46:59 +0100
commit1a18769d245367545c8bb82b83c49845d9e43907 (patch)
tree875eba0b53b1d2e52c120b5a5d3dc2de65079343
parentd802bfbf80bab00123a4a6209f255852b3a10207 (diff)
LHASH: Fix documentation for doall-delete hazards
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com> Reviewed-by: Neil Horman <nhorman@openssl.org> (Merged from https://github.com/openssl/openssl/pull/23032) (cherry picked from commit e91f21193c1b11de775e65b4e209a9fed67a41c1)
-rw-r--r--doc/man3/OPENSSL_LH_COMPFUNC.pod26
1 files changed, 17 insertions, 9 deletions
diff --git a/doc/man3/OPENSSL_LH_COMPFUNC.pod b/doc/man3/OPENSSL_LH_COMPFUNC.pod
index d3bb272c4a..aabd13c522 100644
--- a/doc/man3/OPENSSL_LH_COMPFUNC.pod
+++ b/doc/man3/OPENSSL_LH_COMPFUNC.pod
@@ -145,15 +145,6 @@ For example:
/* Then the hash table itself can be deallocated */
lh_TYPE_free(hashtable);
-When doing this, be careful if you delete entries from the hash table
-in your callbacks: the table may decrease in size, moving the item
-that you are currently on down lower in the hash table - this could
-cause some entries to be skipped during the iteration. The second
-best solution to this problem is to set hash-E<gt>down_load=0 before
-you start (which will stop the hash table ever decreasing in size).
-The best solution is probably to avoid deleting items from the hash
-table inside a "doall" callback!
-
B<lh_I<TYPE>_doall_arg>() is the same as B<lh_I<TYPE>_doall>() except that
I<func> will be called with I<arg> as the second argument and I<func>
should be of type B<LHASH_DOALL_ARG_FN>(B<I<TYPE>>) (a callback prototype
@@ -175,6 +166,23 @@ that is provided by the caller):
lh_TYPE_doall_arg(hashtable, LHASH_DOALL_ARG_FN(TYPE_print), BIO,
logging_bio);
+Note that it is by default B<not> safe to use B<lh_I<TYPE>_delete>() inside a
+callback passed to B<lh_I<TYPE>_doall>() or B<lh_I<TYPE>_doall_arg>(). The
+reason for this is that deleting an item from the hash table may result in the
+hash table being contracted to a smaller size and rehashed.
+B<lh_I<TYPE>_doall>() and B<lh_I<TYPE>_doall_arg>() are unsafe and will exhibit
+undefined behaviour under these conditions, as these functions assume the hash
+table size and bucket pointers do not change during the call.
+
+If it is desired to use B<lh_I<TYPE>_doall>() or B<lh_I<TYPE>_doall_arg>() with
+B<lh_I<TYPE>_delete>(), it is essential that you call
+B<lh_I<TYPE>_set_down_load>() with a I<down_load> argument of 0 first. This
+disables hash table contraction and guarantees that it will be safe to delete
+items from a hash table during a call to B<lh_I<TYPE>_doall>() or
+B<lh_I<TYPE>_doall_arg>().
+
+It is never safe to call B<lh_I<TYPE>_insert>() during a call to
+B<lh_I<TYPE>_doall>() or B<lh_I<TYPE>_doall_arg>().
B<lh_I<TYPE>_error>() can be used to determine if an error occurred in the last
operation.