I'm in my beach-house and pretty offline right now, but I reviewed the
change Kirk did to addaliasu() and found at least one scenario where
it wouldn't do what he expected:

If bdevvp() is called more than once with the same dev_t we would
create more than one anonymous vnode for that dev_t and addaliasu()
would only get one of these in its yarn.  This patch makes bdevvp()
only produce max one anonymous vnode per dev_t.

I don't know if the new rootfs-locating code calls bdevvp() more
than once, if it doesn't this patch is a harmless bit of overhead.

Please test, and if it doesn't do any harm I will commit it.

(This does not fix the panic everybody is talking about btw)


Index: vfs_subr.c
RCS file: /home/ncvs/src/sys/kern/vfs_subr.c,v
retrieving revision 1.267
diff -u -r1.267 vfs_subr.c
--- vfs_subr.c  2000/07/24 05:28:29     1.267
+++ vfs_subr.c  2000/07/28 19:08:36
@@ -1276,6 +1276,8 @@
                *vpp = NULLVP;
                return (ENXIO);
+       if (vfinddev(dev, VCHR, vpp))
+               return (0);
        error = getnewvnode(VT_NON, (struct mount *)0, spec_vnodeop_p, &nvp);
        if (error) {
                *vpp = NULLVP;
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
[EMAIL PROTECTED]         | TCP/IP since RFC 956
FreeBSD coreteam member | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to