Skip to contentSkip to conversation

virtual TPM test

PATCH In review Stefan Berger 2 participants 6 messages Feb 21, 2006 – Feb 22, 2006

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 virtual TPM test

The attached patch adds 2 tests for the virtual TPM to the test suite (more to come over time). It skips the test if it cannot be run. It builds on top of the previous patch.

xen-unstable.hg/tools/xm-test/configure.ac +1 -0
Index: xen/xen-unstable.hg/tools/xm-test/configure.ac
===================================================================
@@ -93,6 +93,7 @@ AC_CONFIG_FILES([
     tests/unpause/Makefile
     tests/vcpu-pin/Makefile
     tests/vcpu-disable/Makefile
+    tests/vtpm/Makefile
     tests/enforce_dom0_cpus/Makefile
     lib/XmTestReport/xmtest.py
     lib/XmTestLib/config.py
xen-unstable.hg/tools/xm-test/tests/Makefile.am +1 -0
Index: xen/xen-unstable.hg/tools/xm-test/tests/Makefile.am
===================================================================
@@ -23,6 +23,7 @@ SUBDIRS = 	               	\
 		unpause         \
 		vcpu-disable    \
 		vcpu-pin	\
+		vtpm            \
 		enforce_dom0_cpus	\
 		save restore migrate
 
xen-unstable.hg/tools/xm-test/tests/vtpm/Makefile.am new file +22 -0
Index: xen/xen-unstable.hg/tools/xm-test/tests/vtpm/Makefile.am
===================================================================
@@ -0,0 +1,22 @@
+
+SUBDIRS =
+
+TESTS = 01_vtpm-list_pos.test \
+        02_vtpm-cat_pcrs.test
+
+XFAIL_TESTS =
+
+EXTRA_DIST = $(TESTS) $(XFAIL_TESTS)
+
+TESTS_ENVIRONMENT=@TENV@
+
+%.test: %.py
+	cp $< $@
+	chmod +x $@
+
+clean-local: am_config_clean-local
+
+am_config_clean-local:
+	rm -f *test
+	rm -f *log
+	rm -f *~
xen-unstable.hg/tools/xm-test/lib/XmTestLib/XenDomain.py +1 -0
Index: xen/xen-unstable.hg/tools/xm-test/lib/XmTestLib/XenDomain.py
===================================================================
@@ -99,6 +99,7 @@ class XenConfig:
         # These options need to be lists
         self.defaultOpts["disk"] = []
         self.defaultOpts["vif"]  = []
+        self.defaultOpts["vtpm"] = []
 
         self.opts = self.defaultOpts
 
xen-unstable.hg/tools/xm-test/tests/vtpm/01_vtpm-list_pos.py new file +45 -0

45 lines not on this page · raw message

xen-unstable.hg/tools/xm-test/tests/vtpm/02_vtpm-cat_pcrs.py new file +81 -0

81 lines not on this page · raw message

xen-unstable.hg/tools/examples/Makefile +1 -1
Index: xen/xen-unstable.hg/tools/examples/Makefile
===================================================================
@@ -26,7 +26,7 @@ XEN_SCRIPTS += network-route vif-route
 XEN_SCRIPTS += network-nat vif-nat
 XEN_SCRIPTS += block
 XEN_SCRIPTS += block-enbd block-nbd
-XEN_SCRIPTS += vtpm
+XEN_SCRIPTS += vtpm vtpm-delete
 XEN_SCRIPT_DATA = xen-script-common.sh
 XEN_SCRIPT_DATA += xen-hotplug-common.sh xen-network-common.sh vif-common.sh
 XEN_SCRIPT_DATA += block-common.sh vtpm-common.sh
xen-unstable.hg/tools/examples/vtpm-common.sh +41 -31

186 lines not on this page · raw message

xen-unstable.hg/tools/examples/vtpm-delete new file +18 -0
Index: xen/xen-unstable.hg/tools/examples/vtpm-delete
===================================================================
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+# This scripts must be called the following way:
+# vtpm-delete <domain name>
+
+if [ $1 != "remove" ]; then
+	exec sh -c "bash $0 remove $*"
+fi
+
+
+XENBUS_PATH=" "
+
+dir=$(dirname "$0")
+. "$dir/vtpm-common.sh"
+
+shift
+
+vtpm_delete_instance $1

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#1 lore raw
Quoted 5 lines · Stefan Berger

The attached patch adds 2 tests for the virtual TPM to the test suite (more to come over time). It skips the test if it cannot be run. It builds on top of the previous patch.

Stefan, I have a number of issues with this patch. Firstly, you have made significant changes to the scripts in tools/examples, plus a change to xm-test's core library, and yet your changelog message makes no mention of them.

xen-unstable.hg/tools/examples/vtpm-common.sh function find_instance () { · line 89

Quoted 9 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-common.sh
===================================================================

xen-unstable.hg/tools/examples/vtpm-common.sh +1 -1
@@ -82,16 +85,15 @@ function find_instance () {
 	if [ "$instance" != "" ]; then
 		ret=1
 	fi
-	return $ret
+	res=$ret
 }

Why have you made this change? Setting a global variable to return a result as a side effect is poor style. This occurs a number of times in your patch.

Quoted 13 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-delete
===================================================================

xen-unstable.hg/tools/examples/vtpm-delete new file +8 -0
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+# This scripts must be called the following way:
+# vtpm-delete <domain name>
+
+if [ $1 != "remove" ]; then
+	exec sh -c "bash $0 remove $*"
+fi

xen-unstable.hg/tools/examples/vtpm-delete line 18

Quoted 10 lines
xen-unstable.hg/tools/examples/vtpm-delete new file +10 -0
+
+
+XENBUS_PATH=" "
+
+dir=$(dirname "$0")
+. "$dir/vtpm-common.sh"
+
+shift
+
+vtpm_delete_instance $1

You are setting XENBUS_PATH just to hack your way into the hotplug scripts, and then relying on the fact that vtpm_delete_instance doesn't actually use that value. This is not a reasonable thing to be do. If there's a need for a tpm-manipulating library, separate from the hotplug infrastructure, then you should split vtpm-common.sh into vtpm-common.sh and vtpm-hotplug-common.sh (or similar). Note that vif-common.sh includes xen-hotplug-common.sh and xen-network-common.sh -- these are split apart for exactly that reason (xen-network-common.sh being used for the network-xyz scripts as well as for the vif-xyz hotplugging scripts).

Ewan.

#2 lore raw

[email protected] wrote on 02/22/2006 10:40:33 AM:

Quoted 9 lines · Stefan Berger
Quoted 6 lines

The attached patch adds 2 tests for the virtual TPM to the test suite (more to come over time). It skips the test if it cannot be run. It builds on top of the previous patch.

Stefan, I have a number of issues with this patch. Firstly, you have

made

Quoted 2 lines

significant changes to the scripts in tools/examples, plus a change to xm-test's core library, and yet your changelog message makes no mention

of

Quoted 13 lines

them.

xen-unstable.hg/tools/examples/vtpm-common.sh function find_instance () { · line 89

Quoted 9 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-common.sh
===================================================================

xen-unstable.hg/tools/examples/vtpm-common.sh +1 -1
@@ -82,16 +85,15 @@ function find_instance () {
    if [ "$instance" != "" ]; then
       ret=1
    fi
-   return $ret
+   res=$ret
 }

Why have you made this change? Setting a global variable to return a

result

Quoted 1 line

as a side effect is poor style. This occurs a number of times in your

patch.

To get rid of the 'set +e' command I cannot return a number in the return statement, otherwise the script just terminates.

Quoted 18 lines
Quoted 18 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-delete
===================================================================

xen-unstable.hg/tools/examples/vtpm-delete new file +12 -0
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+# This scripts must be called the following way:
+# vtpm-delete <domain name>
+
+if [ $1 != "remove" ]; then
+   exec sh -c "bash $0 remove $*"
+fi
+
+
+XENBUS_PATH=" "
+
Quoted 8 lines

xen-unstable.hg/tools/examples/vtpm-delete line 18

Quoted 6 lines
xen-unstable.hg/tools/examples/vtpm-delete new file +6 -0
+dir=$(dirname "$0")
+. "$dir/vtpm-common.sh"
+
+shift
+
+vtpm_delete_instance $1

You are setting XENBUS_PATH just to hack your way into the hotplug

scripts,

Quoted 1 line

and then relying on the fact that vtpm_delete_instance doesn't actually

use

Quoted 1 line

that value. This is not a reasonable thing to be do. If there's a need

for a

Quoted 1 line

tpm-manipulating library, separate from the hotplug infrastructure, then

you

Quoted 1 line

should split vtpm-common.sh into vtpm-common.sh and

vtpm-hotplug-common.sh (or

Quoted 3 lines

similar). Note that vif-common.sh includes xen-hotplug-common.sh and xen-network-common.sh -- these are split apart for exactly that reason (xen-network-common.sh being used for the network-xyz scripts as well as

for

Quoted 1 line

the vif-xyz hotplugging scripts).

It's a hack, indeed. I will change that.

-- Stefan

Quoted 7 lines

Ewan.

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#3 lore raw
Quoted 11 lines · Stefan Berger
Quoted 11 lines

xen-unstable.hg/tools/examples/vtpm-common.sh function find_instance () { · line 89

Quoted 9 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-common.sh
===================================================================

xen-unstable.hg/tools/examples/vtpm-common.sh +1 -1
@@ -82,16 +85,15 @@ function find_instance () {
    if [ "$instance" != "" ]; then
       ret=1
    fi
-   return $ret
+   res=$ret
 }

Why have you made this change? Setting a global variable to return a

Quoted 6 lines

result

Quoted 1 line

as a side effect is poor style. This occurs a number of times in your

patch.

To get rid of the 'set +e' command I cannot return a number in the return statement, otherwise the script just terminates.

Got you. I prefer

foo()
{
echo '1'
}

...

thing=$(foo "$param1")

That way you avoid the global variable, but can still return useful things from the function. With this, you can even have more descriptive return values than integer codes, too.

Cheers,

Ewan.

#4 lore raw
In reply to Ewan Mellor

[email protected] wrote on 02/22/2006 10:40:33 AM:

Quoted 2 lines · Stefan Berger

Ewan, if I cannot source the vtpm-hotplug-common.sh from the vtpm-delete anymore, I don't have access to the locking functions (claim_lock, release_lock) in vtpm-hotplug-common.sh which are needed by my script. May I separte those out to locking.sh so they can be 'sourced' by other scritps as well?

Stefan

Quoted 17 lines
Quoted 17 lines

Index: xen/xen-unstable.hg/tools/examples/vtpm-delete
===================================================================

xen-unstable.hg/tools/examples/vtpm-delete new file +12 -0
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+# This scripts must be called the following way:
+# vtpm-delete <domain name>
+
+if [ $1 != "remove" ]; then
+   exec sh -c "bash $0 remove $*"
+fi
+
+
+XENBUS_PATH=" "
+
Quoted 8 lines

xen-unstable.hg/tools/examples/vtpm-delete line 18

Quoted 6 lines
xen-unstable.hg/tools/examples/vtpm-delete new file +6 -0
+dir=$(dirname "$0")
+. "$dir/vtpm-common.sh"
+
+shift
+
+vtpm_delete_instance $1

You are setting XENBUS_PATH just to hack your way into the hotplug

scripts,

Quoted 1 line

and then relying on the fact that vtpm_delete_instance doesn't actually

use

Quoted 1 line

that value. This is not a reasonable thing to be do. If there's a need

for a

Quoted 1 line

tpm-manipulating library, separate from the hotplug infrastructure, then

you

Quoted 1 line

should split vtpm-common.sh into vtpm-common.sh and

vtpm-hotplug-common.sh (or

Quoted 3 lines

similar). Note that vif-common.sh includes xen-hotplug-common.sh and xen-network-common.sh -- these are split apart for exactly that reason (xen-network-common.sh being used for the network-xyz scripts as well as

for

Quoted 8 lines

the vif-xyz hotplugging scripts).

Ewan.

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#5 lore raw
Quoted 10 lines · Stefan Berger

[email protected] wrote on 02/22/2006 10:40:33 AM:

Quoted 2 lines · Stefan Berger

Ewan, if I cannot source the vtpm-hotplug-common.sh from the vtpm-delete anymore, I don't have access to the locking functions (claim_lock, release_lock) in vtpm-hotplug-common.sh which are needed by my script. May I separte those out to locking.sh so they can be 'sourced' by other scritps as well?

Sure, that makes sense.

Ewan.