From: Ben Gras <ben@minix3.org>
Date: Tue, 11 Sep 2007 15:52:22 +0000 (+0000)
Subject: Don't truncate read requests based on v_size; v_size can be stale in the
X-Git-Tag: v3.1.4~323
X-Git-Url: http://zhaoyanbai.com/repos/%22http:/www.isc.org/icons/dnssec-revoke.html?a=commitdiff_plain;h=38604e4e3a80ea742001379a3aed867a83ab81c2;p=minix.git

Don't truncate read requests based on v_size; v_size can be stale in the
case of directories extended by subfilesystem. Rely on subfilesystem to
do read size truncating and return actual i/o size. This fixes bug 81 in
gforge, and unbreaks test 23.
---

diff --git a/servers/vfs/read.c b/servers/vfs/read.c
index d3660d6cf..8c874d0b1 100644
--- a/servers/vfs/read.c
+++ b/servers/vfs/read.c
@@ -150,7 +150,9 @@ int rw_flag;			/* READING or WRITING */
       /* Fill in request structure */
       num_of_bytes = m_in.nbytes;
 
-      /* Truncate read request at size (mustn't do this for special files). */
+#if 0  /* Don't truncate read request at size. The filesystem process will
+	* do this itself.
+	*/
       if((rw_flag == READING) &&
 	cmp64ul(add64ul(position, num_of_bytes), vp->v_size) > 0) {
 		/* Position always should fit in an off_t (LONG_MAX). */
@@ -162,6 +164,7 @@ int rw_flag;			/* READING or WRITING */
 		num_of_bytes = vp->v_size - pos32;
 		assert(num_of_bytes >= 0);
       }
+#endif
 
       /* Issue request */
       r = req_readwrite(vp->v_fs_e, vp->v_inode_nr, vp->v_index, position,