OK, this basically adds a function that allows us to create

apr_lock_t's using an apr_sms_t for the memory.  It's a very small first
step, and at present is only intended to be used internally in APR,
hence the position of the function definitions in the locks.h file.

Given that we don't want to duplicate code where we don't have to, I've
added some macros that allow us to do memory allocations regardless of
whether we have a pool or sms.  This also highlighted that we haven't
yet managed to change our member names from cntxt to pool everywhere!

The surprise comes in just how far reaching the apr_pool_t goes.

As apr_lock.h uses ap_sms_t and apr_sms.h uses apr_lock_t I've moved
the tyepdef's into apr.h.in, but this may be bogus...

Hopefully this will allow us to get locking into sms and so start
moving forward again, and at the same time it starts to throw up
the problems for changing our memory system throughout APR.


git-svn-id: http://svn.apache.org/repos/asf/apr/apr/trunk@61698 13f79535-47bb-0310-9956-ffa450edef68

1 files changed, 2 insertions, 0 deletions

diff --git a/include/apr.h.in b/include/apr.h.in
index 63136726c..c72981e43 100644
--- a/include/apr.h.in
+++ b/include/apr.h.in
@@ -174,6 +174,8 @@ typedef  @ssize_t_value@         apr_ssize_t;
 typedef  @off_t_value@           apr_off_t;
 typedef  @socklen_t_value@       apr_socklen_t;
 
+typedef struct apr_lock_t        apr_lock_t;
+typedef struct apr_sms_t         apr_sms_t;
 
 /* Mechanisms to properly type numeric literals */
 @int64_literal@