The dcache in pre-patch 2.1.45-3 is much improved -- it seems to be
releasing some inodes now, enough so that I can copy kernel source trees
and run diff -ru in a 6M machine. A nasty bug triggered by removing
busy files also has disappeared.
One minor problem I'm having -- after I mount and umount a volume,
there's still an inode held on the volume, so I can't mount it again.
Is there a call to invalidate the dcache for a device?
I needed one trivial patch to get things to compile, as attached.
Regards,
Bill
--------------A176206276F9E06A51C71D9A
Content-Type: text/plain; charset=us-ascii; name="binfmt_misc-patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline; filename="binfmt_misc-patch"
--- fs/binfmt_misc.c.old Wed Jul 9 07:18:25 1997
+++ fs/binfmt_misc.c Sat Jul 12 19:51:18 1997
@@ -197,7 +197,7 @@
}
bprm->filename = iname; /* for binfmt_script */
- if ((retval = open_namei(iname, 0, 0, &bprm->inode, NULL)))
+ if ((retval = open_namei(iname, 0, 0, &bprm->inode)))
goto _ret;
bprm->dont_iput = 0;
--------------A176206276F9E06A51C71D9A--