Immortal data should go in an immortal bucket, not a pool bucket. It would

still work in a pool bucket, but only by coincidence.  At the very least, a
pool bucket is way more overhead than this poor unsuspecting little CRLF
needs to get by.

git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/trunk@91160 13f79535-47bb-0310-9956-ffa450edef68

diff --git a/modules/proxy/proxy_http.c b/modules/proxy/proxy_http.c
index 6540933cc16e347d8f0aca0fbbe4cb9866da6728..4ebff5be95f8d8d382d3db64017194dba577a110 100644 (file)
--- a/modules/proxy/proxy_http.c
+++ b/modules/proxy/proxy_http.c
@@ -588,7 +588,7 @@ apr_status_t ap_proxy_http_request(apr_pool_t *p, request_rec *r,
     }
 
     /* add empty line at the end of the headers */
-    e = apr_bucket_pool_create(CRLF, strlen(CRLF), p);
+    e = apr_bucket_immortal_create(CRLF, sizeof(CRLF)-1);
     APR_BRIGADE_INSERT_TAIL(bb, e);
     e = apr_bucket_flush_create();
     APR_BRIGADE_INSERT_TAIL(bb, e);