Oops.. Sorry. I wanted to rewrite the piece of code before submitting it.
Here's a quickfix, incorporating Aaron's and Doug's comments.
However, at a later part, I was thinking of having the realloc logic
something like :
if (requested_size < current_size) {
split the current block
free the second part
return
}
else if (requested_size > current_size) {
if (do_we_have_enough_memory to allocate) {
allocate memory
copy from old to new
free old
return
}
else {
if (do we have free blocks around this memory so that a bigger chunk can
be formed to satisfy the request) {
re-adjust the header block
return
}
}
}
any comments. (both patch and the above logic)
-Madhu
Index: apr_rmm.c
===================================================================
RCS file: /home/cvspublic/apr-util/misc/apr_rmm.c,v
retrieving revision 1.10
diff -u -r1.10 apr_rmm.c
--- apr_rmm.c 14 Mar 2002 23:46:29 -0000 1.10
+++ apr_rmm.c 15 Mar 2002 00:17:21 -0000
@@ -383,13 +383,12 @@
old = apr_rmm_offset_get(rmm, entity);
if ((this = apr_rmm_malloc(rmm, reqsize)) < 0) {
- return rv;
+ return this;
}
if (old >= 0) {
memcpy(apr_rmm_addr_get(rmm, this),
apr_rmm_addr_get(rmm, old), reqsize);
- move_block(rmm, old, 1);
apr_rmm_free(rmm, old);
}
-----Original Message-----
From: Doug MacEachern [mailto:[EMAIL PROTECTED]
Sent: Thursday, March 14, 2002 4:02 PM
To: Aaron Bannert
Cc: [email protected]
Subject: Re: cvs commit: apr-util/misc apr_rmm.c
just noticed another problem, rv never gets set here:
apr_status_t rv;
...
if ((this = apr_rmm_malloc(rmm, reqsize)) < 0) {
return rv;
}
thinking it should be APR_EINVAL, but the return type of apr_rmm_realloc
is apr_rmm_off_t. then again, both apr_rmm_{calloc,malloc} have the same
return type but return an apr_status_t if APR_ANYLOCK_LOCK fails.
what's up with that?
