Well, actually this seemed to be a bug in generic code, that

was being exercised. running with a 8k fs blk size and 4 k page
size was a problem. Now running with fs blk size == page size works
much better. The fix checked in a week or to back probably would fix
this too.

diff --git a/sys/arch/powerpc/patches b/sys/arch/powerpc/patches
index c869028..3c44004 100644 (file)
--- a/sys/arch/powerpc/patches
+++ b/sys/arch/powerpc/patches
@@ -131,17 +131,3 @@ diff -r1.9 fdisk.c
 ---
 >      if (getshort(&mboot.signature) != BOOT_MAGIC) {
 >              warnx("invalid fdisk partition table found!\n");
-Index: src/sys/kern/vfs_cache.c
-# Probably a bug in the port, but this makes the system MANY TIMES
-# more stablility.
-===================================================================
-RCS file: /cvs/src/sys/kern/vfs_cache.c,v
-retrieving revision 1.2
-diff -r1.2 vfs_cache.c
-74a75,78
-> #ifdef __powerpc__
-> /* WRONG, hack for stability for now -- XXX */
-> int doingcache = 0;                   /* 1 => enable the cache */
-> #else
-75a80
-> #endif