Paul Durrant
2010-Jul-28 14:15 UTC
[Xen-devel] [PATCH] Add tool to attempt to crash HVM guests
# HG changeset patch
# User Paul Durrant <paul.durrant@citrix.com>
# Date 1280326501 -3600
# Node ID c28224c0f0657aa29c1385f5711cccafbb2024d2
# Parent 4207549948a4c34b9e192635ec5b3b75a09dbf30
Add tool to attempt to crash HVM guests
This tool reads the CPU save records, overwrites RIP with a bogus
value, and then restores them. This is, of course, not guaranteed
to crash the guest (since the CPUs may not be executing in kernel
at the time) but it''s good for breaking into some tight loops that
would be hard to debug otherwise.
signed-off-by: Paul Durrant <paul.durrant@citrix.com>
diff -r 4207549948a4 -r c28224c0f065 tools/misc/Makefile
--- a/tools/misc/Makefile Wed Jul 28 08:32:01 2010 +0100
+++ b/tools/misc/Makefile Wed Jul 28 15:15:01 2010 +0100
@@ -11,7 +11,7 @@ HDRS = $(wildcard *.h)
HDRS = $(wildcard *.h)
TARGETS-y := xenperf xenpm xen-tmem-list-parse gtraceview gtracestat
xenlockprof xenwatchdogd
-TARGETS-$(CONFIG_X86) += xen-detect xen-hvmctx
+TARGETS-$(CONFIG_X86) += xen-detect xen-hvmctx xen-hvmcrash
TARGETS-$(CONFIG_MIGRATE) += xen-hptool
TARGETS := $(TARGETS-y)
@@ -24,7 +24,7 @@ INSTALL_BIN := $(INSTALL_BIN-y)
INSTALL_BIN := $(INSTALL_BIN-y)
INSTALL_SBIN-y := xm xen-bugtool xen-python-path xend xenperf xsview xenpm
xen-tmem-list-parse gtraceview gtracestat xenlockprof xenwatchdogd
-INSTALL_SBIN-$(CONFIG_X86) += xen-hvmctx
+INSTALL_SBIN-$(CONFIG_X86) += xen-hvmctx xen-hvmcrash
INSTALL_SBIN-$(CONFIG_MIGRATE) += xen-hptool
INSTALL_SBIN := $(INSTALL_SBIN-y)
@@ -51,7 +51,7 @@ clean:
%.o: %.c $(HDRS) Makefile
$(CC) -c $(CFLAGS) -o $@ $<
-xen-hvmctx xenperf xenpm gtracestat xenlockprof xen-hptool xenwatchdogd: %: %.o
Makefile
+xen-hvmctx xen-hvmcrash xenperf xenpm gtracestat xenlockprof xen-hptool
xenwatchdogd: %: %.o Makefile
$(CC) $(CFLAGS) -o $@ $< $(LDFLAGS) $(LDFLAGS_libxenctrl)
$(LDFLAGS_libxenguest) $(LDFLAGS_libxenstore)
gtraceview: %: %.o Makefile
diff -r 4207549948a4 -r c28224c0f065 tools/misc/xen-hvmcrash.c
--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/tools/misc/xen-hvmcrash.c Wed Jul 28 15:15:01 2010 +0100
@@ -0,0 +1,143 @@
+/*
+ * xen-hvmcrash.c
+ *
+ * Attempt to crash an HVM guest by overwriting RIP/EIP with a bogus value
+ *
+ * Copyright (c) 2010 Citrix Systems, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the
"Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+#include <inttypes.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stddef.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <string.h>
+#include <errno.h>
+#include <limits.h>
+
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <arpa/inet.h>
+
+#include <xenctrl.h>
+#include <xen/xen.h>
+#include <xen/domctl.h>
+#include <xen/hvm/save.h>
+
+int
+main(int argc, char **argv)
+{
+ int domid;
+ xc_interface *xch;
+ xc_dominfo_t dominfo;
+ int ret;
+ uint32_t len;
+ uint8_t *buf;
+ uint32_t off;
+ struct hvm_save_descriptor *descriptor;
+
+ if (argc != 2 || !argv[1] || (domid = atoi(argv[1])) < 0) {
+ fprintf(stderr, "usage: %s <domid>\n", argv[0]);
+ exit(1);
+ }
+
+ xch = xc_interface_open(0, 0, 0);
+ if (!xch) {
+ fprintf(stderr, "error: can''t open libxc handle\n");
+ exit(1);
+ }
+
+ ret = xc_domain_getinfo(xch, domid, 1, &dominfo);
+ if (ret < 0) {
+ perror("xc_domain_getinfo");
+ exit(1);
+ }
+
+ if (!dominfo.hvm) {
+ fprintf(stderr, "domain %d is not HVM\n", domid);
+ exit(1);
+ }
+
+ ret = xc_domain_pause(xch, domid);
+ if (ret < 0) {
+ perror("xc_domain_pause");
+ exit(-1);
+ }
+
+ /*
+ * Calling with zero buffer length should return the buffer length
+ * required.
+ */
+ ret = xc_domain_hvm_getcontext(xch, domid, 0, 0);
+ if (ret < 0) {
+ perror("xc_domain_hvm_getcontext");
+ exit(1);
+ }
+
+ len = ret;
+ buf = malloc(len);
+ if (buf == NULL) {
+ perror("malloc");
+ exit(1);
+ }
+
+ ret = xc_domain_hvm_getcontext(xch, domid, buf, len);
+ if (ret < 0) {
+ perror("xc_domain_hvm_getcontext");
+ exit(1);
+ }
+
+ off = 0;
+
+ while (off < len) {
+ descriptor = (struct hvm_save_descriptor *)(buf + off);
+
+ off += sizeof (struct hvm_save_descriptor);
+
+ if (descriptor->typecode == HVM_SAVE_CODE(CPU)) {
+ HVM_SAVE_TYPE(CPU) *cpu;
+
+ /* Overwrite EIP/RIP with some recognisable but bogus value */
+ cpu = (HVM_SAVE_TYPE(CPU) *)(buf + off);
+ printf("CPU[%d]: RIP = %" PRIx64 "\n",
descriptor->instance, cpu->rip);
+ cpu->rip = 0xf001;
+ } else if (descriptor->typecode == HVM_SAVE_CODE(END)) {
+ break;
+ }
+
+ off += descriptor->length;
+ }
+
+ ret = xc_domain_hvm_setcontext(xch, domid, buf, len);
+ if (ret < 0) {
+ perror("xc_domain_hvm_setcontext");
+ exit(1);
+ }
+
+ ret = xc_domain_unpause(xch, domid);
+ if (ret < 0) {
+ perror("xc_domain_unpause");
+ exit(1);
+ }
+
+ return 0;
+}
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
Ian Jackson
2010-Jul-29 15:51 UTC
Re: [Xen-devel] [PATCH] Add tool to attempt to crash HVM guests
Paul Durrant writes ("[Xen-devel] [PATCH] Add tool to attempt to crash HVM
guests"):> This tool reads the CPU save records, overwrites RIP with a bogus
> value, and then restores them. This is, of course, not guaranteed
> to crash the guest (since the CPUs may not be executing in kernel
> at the time) but it''s good for breaking into some tight loops that
> would be hard to debug otherwise.
Surely there must be a better way of doing this! Isn''t it sufficient
to pause the guest and then inspect it ? Or do we need a new domctl
to crash the guest explicitly ?
Ian.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
Stefano Stabellini
2010-Jul-29 15:57 UTC
Re: [Xen-devel] [PATCH] Add tool to attempt to crash HVM guests
On Thu, 29 Jul 2010, Ian Jackson wrote:> Paul Durrant writes ("[Xen-devel] [PATCH] Add tool to attempt to crash HVM guests"): > > This tool reads the CPU save records, overwrites RIP with a bogus > > value, and then restores them. This is, of course, not guaranteed > > to crash the guest (since the CPUs may not be executing in kernel > > at the time) but it''s good for breaking into some tight loops that > > would be hard to debug otherwise. > > Surely there must be a better way of doing this! Isn''t it sufficient > to pause the guest and then inspect it ? Or do we need a new domctl > to crash the guest explicitly ? >Explicitly crashing the guest can be very useful for testing, I had to use this tool myself few times... _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Paul Durrant
2010-Jul-29 16:12 UTC
RE: [Xen-devel] [PATCH] Add tool to attempt to crash HVM guests
> -----Original Message----- > From: Ian Jackson > Sent: 29 July 2010 16:51 > To: Paul Durrant > Cc: xen-devel@lists.xensource.com > Subject: Re: [Xen-devel] [PATCH] Add tool to attempt to crash HVM > guests > > Paul Durrant writes ("[Xen-devel] [PATCH] Add tool to attempt to > crash HVM guests"): > > This tool reads the CPU save records, overwrites RIP with a bogus > > value, and then restores them. This is, of course, not guaranteed > > to crash the guest (since the CPUs may not be executing in kernel > > at the time) but it''s good for breaking into some tight loops that > > would be hard to debug otherwise. > > Surely there must be a better way of doing this! Isn''t it > sufficient > to pause the guest and then inspect it ? Or do we need a new domctl > to crash the guest explicitly ? >Ian, Inspecting the guest from dom0 is of limited use since we have nothing running there to attach windbg too. Causing it to fallover in the way this tool does usually unlocks the kernel debugger polling loop and hence windbg can get a lookin over the emulated serial device. By ''crash the guest explicitly'' what were you thinking? Paul _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Ian Jackson
2010-Jul-29 16:28 UTC
RE: [Xen-devel] [PATCH] Add tool to attempt to crash HVM guests
Paul Durrant writes ("RE: [Xen-devel] [PATCH] Add tool to attempt to crash
HVM guests"):
...> Inspecting the guest from dom0 is of limited use since we have
> nothing running there to attach windbg too. Causing it to fallover
> in the way this tool does usually unlocks the kernel debugger
> polling loop and hence windbg can get a lookin over the emulated
> serial device.
Oh, I see what you''re trying to do. This makes more sense now.
Ian.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel