NetBSD Problem Report #45420

From woods@once.weird.com  Tue Oct  4 18:33:55 2011
Return-Path: <woods@once.weird.com>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by www.NetBSD.org (Postfix) with ESMTP id BB0A763B85F
	for <gnats-bugs@gnats.NetBSD.org>; Tue,  4 Oct 2011 18:33:55 +0000 (UTC)
Message-Id: <m1RB9oM-001EBeC@once.weird.com>
Date: Tue, 4 Oct 2011 11:33:50 -0700 (PDT)
From: "Greg A. Woods" <woods@planix.com>
Sender: "Greg A. Woods" <woods@once.weird.com>
Reply-To: "Greg A. Woods" <woods@planix.com>
To: gnats-bugs@gnats.NetBSD.org
Subject: kernel should verify both /dev/constty and /dev/console
X-Send-Pr-Version: 3.95

>Number:         45420
>Category:       kern
>Synopsis:       kernel should verify both /dev/constty and /dev/console
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          change-request
>Submitter-Id:   net
>Arrival-Date:   Tue Oct 04 18:35:00 +0000 2011
>Originator:     Greg A. Woods
>Release:        -current 2011/10/04
>Organization:
Planix, Inc.; Kelowna, BC; Canada
>Environment:
System: NetBSD
Architecture: all
Machine: all
>Description:

	the kernel checks to see if /dev/console exits, but fails to
	also check if /dev/constty exits

	both should exist for constty to be used properly

	also note the updated comment -- is it really that much more
	work to craft up open file descriptors for the active console
	device?

>How-To-Repeat:

>Fix:

--- init_main.c	28 Sep 2011 16:06:27 -0700	1.436
+++ init_main.c	04 Oct 2011 10:56:04 -0700	
@@ -832,6 +832,15 @@
 		printf("warning: no /dev/console\n");
 	else
 		printf("warning: lookup /dev/console: error %d\n", error);
+
+	error = namei_simple_kernel("/dev/constty",
+				NSM_FOLLOW_NOEMULROOT, &vp);
+	if (error == 0)
+		vrele(nd.ni_vp);
+	else if (error == ENOENT)
+		printf("warning: no /dev/constty\n");
+	else
+		printf("warning: lookup /dev/constty: error %d\n", error);
 }

 /*
@@ -881,10 +890,16 @@
 	mutex_exit(proc_lock);

 	/*
-	 * This is not the right way to do this.  We really should
-	 * hand-craft a descriptor onto /dev/console to hand to init,
-	 * but that's a _lot_ more work, and the benefit from this easy
-	 * hack makes up for the "good is the enemy of the best" effect.
+	 * This is not the right way to do this.  We really should hand-craft
+	 * descriptors onto the active console device we're using internally in
+	 * the kernel, just as if they were open on /dev/constty, and pass
+	 * those to init as its STDIO descriptors, but apparently that's a
+	 * _lot_ more work.
+	 *
+	 * This easy hack at least leaves hints on the real console whenever
+	 * there's a potential problem with the /dev/console and/or
+	 * /dev/constty files; and init's attempts to hook on to the console
+	 * will hopefully do for now.
 	 */
 	check_console(l);

Home	PR Database Search