|
For some reason when we try to ever use the builtins, even with the
symbol there as a fallback, something goes horribly wrong somewhere
around here:
| (gdb) bt
| #0 strcmp (s1=0x7d492359 "MD5", s2=0x7d492359 "MD5") at include/system/string.h:57
| #1 0x000000007d460419 in getrn (lh=lh@entry=0x7e081318, data=data@entry=0x7e084398, rhash=rhash@entry=0x7f7c9268) at crypto/lhash/lhash.c:415
| #2 0x000000007d46076e in lh_insert (lh=0x7e081318, data=data@entry=0x7e084398) at crypto/lhash/lhash.c:188
| #3 0x000000007d43e027 in OBJ_NAME_add (name=name@entry=0x7d492359 "MD5", type=type@entry=1, data=data@entry=0x7d4ad3a0 <md5_md> "\004") at crypto/objects/o_names.c:202
As much as I love a Sisyphean challenge, in the interest of not having
bugs or time, this patch changes it to just not use them for anything
other than guaranteeing our implementations have the exact same types as
you would expect.
Signed-off-by: Peter Jones <pjones@redhat.com>
|
