Restore the x509_hash_name_alpgorithm_change patch.

openssl/crypto/x509/by_dir.c

Index: openssl/crypto/x509/by_dir.c
===================================================================
--- openssl/crypto/x509/by_dir.c	(revision 189610)
+++ openssl/crypto/x509/by_dir.c	(working copy)
@@ -287,6 +287,8 @@
 	int ok=0;
 	int i,j,k;
 	unsigned long h;
+	unsigned long hash_array[2];
+	int hash_index;
 	BUF_MEM *b=NULL;
 	X509_OBJECT stmp,*tmp;
 	const char *postfix="";
@@ -323,6 +325,11 @@
 	ctx=(BY_DIR *)xl->method_data;
 
 	h=X509_NAME_hash(name);
+	hash_array[0]=h;

[Ryan Sleevi, 2013/03/21 19:56:01]

Why structured like this?

Why not
hash_array[0]=X509_NAME_hash(name);

since you assign h on line 332

[digit1, 2013/03/21 21:23:31]

That's not my code :-) I'll try to fix it nonetheless.

+	hash_array[1]=X509_NAME_hash_old(name);
+	for (hash_index=0; hash_index < 2; hash_index++)

[Ryan Sleevi, 2013/03/21 19:56:01]

++hash_index

[wtc, 2013/03/21 22:41:49]

Our C++ Style Guide only recommends using the
pre-increment operator on iterators.

In C code it is common to post-increment for loop index
variables, which seems to be the local preference in this
file (see line 333).

http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Preincrement_a...

[agl, 2013/03/22 13:52:36]

Long ago, post incrementing STL iterators was less efficient and so a style
guide clause was born:

"Use prefix form (++i) of the increment and decrement operators with iterators
and other template objects."

It only ever applied to iterators and the like, and it's long been the case that
C++ compilers will generate exactly the same code for both i++ and ++i, even
then.

This aspect of the style guide is now an anachronism.

+		{
+		h=hash_array[hash_index];
 	for (i=0; i < sk_BY_DIR_ENTRY_num(ctx->dirs); i++)
 		{
 		BY_DIR_ENTRY *ent;
@@ -476,6 +483,7 @@
 			goto finish;
 			}
 		}
+		}
 finish:
 	if (b != NULL) BUF_MEM_free(b);
 	return(ok);