faulty use of "set" command when running against the xen hypervisor

[Dave Anderson <anderson@xxxxxxxxxx>]

Cai Quan bumped into another problem when running against the xen hypervisor,
where entering the "set" command alone with no arguments generates a SIGSEGV.

I also note that "set -c #" and "set -p" options make no sense either:

  # crash --xen_phys_start 3ee00000 xen-syms-2.6.18-92.el5 vmcore

  crash 4.0-7.2
  Copyright (C) 2002, 2003, 2004, 2005, 2006, 2007, 2008  Red Hat, Inc.
  Copyright (C) 2004, 2005, 2006  IBM Corporation
  Copyright (C) 1999-2006  Hewlett-Packard Co
  Copyright (C) 2005, 2006  Fujitsu Limited
  Copyright (C) 2006, 2007  VA Linux Systems Japan K.K.
  Copyright (C) 2005  NEC Corporation
  Copyright (C) 1999, 2002, 2007  Silicon Graphics, Inc.
  Copyright (C) 1999, 2000, 2001, 2002  Mission Critical Linux, Inc.
  This program is free software, covered by the GNU General Public License,
  and you are welcome to change it and/or distribute copies of it under
  certain conditions.  Enter "help copying" to see the conditions.
  This program has absolutely no warranty.  Enter "help warranty" for details.

  GNU gdb 6.1
  Copyright 2004 Free Software Foundation, Inc.
  GDB is free software, covered by the GNU General Public License, and you are
  welcome to change it and/or distribute copies of it under certain conditions.
  Type "show copying" to see the conditions.
  There is absolutely no warranty for GDB.  Type "show warranty" for details.
  This GDB was configured as "x86_64-unknown-linux-gnu"...

     KERNEL: xen-syms-2.6.18-92.el5
  DEBUGINFO: ./xen-syms-2.6.18-92.el5.debug
   DUMPFILE: vmcore
       CPUS: 2
    DOMAINS: 4
     UPTIME: 00:34:18
    MACHINE: Intel(R) Pentium(R) 4 CPU 3.40GHz  (3400 Mhz)
     MEMORY: 1 GB
    PCPU-ID: 1
       PCPU: ffff83003f05ff28
    VCPU-ID: 1
       VCPU: ffff83003eef6080  (VCPU_RUNNING)
  DOMAIN-ID: 0
     DOMAIN: ffff83003eef8080  (DOMAIN_RUNNING)
      STATE: CRASH

  crash> set -c 0
  set: invalid cpu number: system has only 0 cpu
  crash> set -p
  set: no panic task found!
  crash> set
  Segmentation fault

  #

Admittedly it's a nonsensical usages of "set", since there's no concept of
a PID "context" in the hypervisor.

The attached patch changes the behavior to:

  crash> set -c 0
  set: -c option not supported on this architecture or kernel
  crash> set -p
  set: -p option not supported on this architecture or kernel
  crash> set
  set: requires an option with the Xen hypervisor
  crash>

Itsura, are you OK with the attached patch?

Thanks,
  Dave


--- crash/tools.c.orig
+++ crash/tools.c
@@ -1673,6 +1673,9 @@ cmd_set(void)
                 switch(c)
 		{
 		case 'c':
+			if (XEN_HYPER_MODE())
+				option_not_supported(c);
+
 			if (!runtime) {
 				error(INFO, 
 				    "cpu setting not allowed from .%src\n",
@@ -1689,6 +1692,9 @@ cmd_set(void)
 			return;
 
 		case 'p':
+			if (XEN_HYPER_MODE())
+				option_not_supported(c);
+
 			if (!runtime)
 				return;
 
@@ -1723,7 +1729,10 @@ cmd_set(void)
 	}
 
 	if (!args[optind]) {
-		if (runtime)
+		if (XEN_HYPER_MODE())
+			error(INFO, 
+			    "requires an option with the Xen hypervisor\n");
+		else if (runtime)
 			show_context(CURRENT_CONTEXT());
 		return;
 	}

--
Crash-utility mailing list
Crash-utility@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/crash-utility