dovecot: Make sure we don't try to re-read cache fields when com...

[dovecot at dovecot.org]

details:   http://hg.dovecot.org/dovecot/rev/30904b20782d
changeset: 6295:30904b20782d
user:      Timo Sirainen <tss at iki.fi>
date:      Mon Aug 13 20:16:07 2007 +0300
description:
Make sure we don't try to re-read cache fields when compressing.

diffstat:

1 file changed, 5 insertions(+), 3 deletions(-)
src/lib-index/mail-cache-lookup.c |    8 +++++---

diffs (18 lines):

diff -r 10a5f61d9692 -r 30904b20782d src/lib-index/mail-cache-lookup.c
--- a/src/lib-index/mail-cache-lookup.c	Mon Aug 13 20:05:41 2007 +0300
+++ b/src/lib-index/mail-cache-lookup.c	Mon Aug 13 20:16:07 2007 +0300
@@ -199,9 +199,11 @@ int mail_cache_lookup_iter_next(struct m
 
 	if (file_field >= cache->file_fields_count) {
 		/* new field, have to re-read fields header to figure
-		   out its size */
-		if (mail_cache_header_fields_read(cache) < 0)
-			return -1;
+		   out its size. don't do this if we're compressing. */
+		if (!cache->locked) {
+			if (mail_cache_header_fields_read(cache) < 0)
+				return -1;
+		}
 		if (file_field >= cache->file_fields_count) {
 			mail_cache_set_corrupted(cache,
 				"field index too large (%u >= %u)",