[OSSTEST PATCH 00/13] Build and test rump kernel xenstore-ls

[OSSTEST PATCH 00/13] Build and test rump kernel xenstore-ls

[Ian Jackson]

Preparatory:

 01/13  selectguest: log a message
 02/13  BuildSupport: make selectbuildhost modify @ARGV
 03/13  sg-run-job: Support toggling `adding' by repeating `+'
 04/13  building: Delete when cloning, not when preparing $builddir
 05/13  make-flight: rumpuserxen tests: use same buildjob for all guests

Building:

 06/13  ts-xen-build: Allow specification of make targets
 07/13  ts-xen-build: Honour cmdprefix_configure, cmdprefix_make
 08/13  rump kernel tests: Do a xen.git tools build

Testing:

 09/13  rump kernel tests: Use guest-specific fs images location
 10/13  rump kernel tests: Specify kernel via runvar
 11/13  rump kernel tests: Specify cmdline via runvar
 12/13  ts-guest-destroy-hard: New script
 13/13  rump kernel tests: Run xenstore-ls demo                             

I have not yet run a full flight with this - I have only bench-tested
it.  I'm about to set a flight off to see how it does.

[OSSTEST PATCH 01/13] selectguest: log a message

[Ian Jackson]

Print the guest's name and the host it's on; this can help with
interpreting the logs.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 Osstest/TestSupport.pm |    1 +
 1 file changed, 1 insertion(+)

diff --git a/Osstest/TestSupport.pm b/Osstest/TestSupport.pm
index 23ad106..22dcbff 100644
--- a/Osstest/TestSupport.pm
+++ b/Osstest/TestSupport.pm
@@ -1227,6 +1227,7 @@ sub selectguest ($$) {
     foreach my $opt (guest_var_commalist($gho,'options')) {
         $gho->{Options}{$opt}++;
     }
+    logm("guest: using $gn on $gho->{Host}{Name}");
     guest_find_lv($gho);
     guest_find_ether($gho);
     guest_find_tcpcheckport($gho);
-- 
1.7.10.4

[OSSTEST PATCH 02/13] BuildSupport: make selectbuildhost modify @ARGV

[Ian Jackson]

We pass it \@ARGV everywhere now, and it eats the argument (if it
finds one).  This allows build scripts to take extra arguments.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 Osstest/BuildSupport.pm |    5 +++--
 ts-kernel-build         |    2 +-
 ts-libvirt-build        |    2 +-
 ts-rumpuserxen-build    |    2 +-
 ts-xen-build            |    2 +-
 5 files changed, 7 insertions(+), 6 deletions(-)

diff --git a/Osstest/BuildSupport.pm b/Osstest/BuildSupport.pm
index 6d43cbc..0938810 100644
--- a/Osstest/BuildSupport.pm
+++ b/Osstest/BuildSupport.pm
@@ -56,8 +56,9 @@ our ($builddir,$makeflags);
 our ($xendist);
 
 sub selectbuildhost {
-    # pass @ARGV
-    ($whhost) = @_;
+    # pass \@ARGV
+    my ($av) = @_;
+    $whhost = shift @$av;
     $whhost ||= 'host';
     $ho= selecthost($whhost);
 }
diff --git a/ts-kernel-build b/ts-kernel-build
index 984a935..3b48920 100755
--- a/ts-kernel-build
+++ b/ts-kernel-build
@@ -22,7 +22,7 @@ use Osstest::TestSupport;
 use Osstest::BuildSupport;
 
 tsreadconfig();
-selectbuildhost(@ARGV);
+selectbuildhost(\@ARGV);
 builddirsprops();
 
 my $archparms = {
diff --git a/ts-libvirt-build b/ts-libvirt-build
index 5f493e4..940c034 100755
--- a/ts-libvirt-build
+++ b/ts-libvirt-build
@@ -22,7 +22,7 @@ use Osstest::TestSupport;
 use Osstest::BuildSupport;
 
 tsreadconfig();
-selectbuildhost(@ARGV);
+selectbuildhost(\@ARGV);
 builddirsprops();
 
 sub libvirtd_init ();
diff --git a/ts-rumpuserxen-build b/ts-rumpuserxen-build
index ee2db97..a1f31cd 100755
--- a/ts-rumpuserxen-build
+++ b/ts-rumpuserxen-build
@@ -22,7 +22,7 @@ use Osstest::TestSupport;
 use Osstest::BuildSupport;
 
 tsreadconfig();
-selectbuildhost(@ARGV);
+selectbuildhost(\@ARGV);
 builddirsprops();
 
 our %submodmap = qw(nblibs nblibs
diff --git a/ts-xen-build b/ts-xen-build
index 828c2b4..3913273 100755
--- a/ts-xen-build
+++ b/ts-xen-build
@@ -24,7 +24,7 @@ use Osstest::TestSupport;
 use Osstest::BuildSupport;
 
 tsreadconfig();
-selectbuildhost(@ARGV);
+selectbuildhost(\@ARGV);
 builddirsprops();
     
 sub checkout () {
-- 
1.7.10.4

[OSSTEST PATCH 03/13] sg-run-job: Support toggling `adding' by repeating `+'

[Ian Jackson]

Specifying `+' as an argument to spawn-ts suppresses the automatic
addition of subsequent arguments to the testid.  Make it possible to
toggle this by repeating `+'.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 sg-run-job |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/sg-run-job b/sg-run-job
index e903869..b02a755 100755
--- a/sg-run-job
+++ b/sg-run-job
@@ -122,7 +122,7 @@ proc spawn-ts {iffail testid ts args} {
     set host_testid_suffix {}
     foreach arg $args {
         if {![string compare + $arg]} {
-            set adding 0
+            set adding [expr {!$adding}]
             continue
         }
         lappend real_args $arg
-- 
1.7.10.4

[OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Jackson]

Make prepbuilddirs no longer delete $builddir; instead:
 - have prepbuilddirs delete *-stamp files and dist/;
 - have build_clone delete the specific source directories;
 - any individual ts-*-build scripts that use other things
   would have to delete them (but searching for $builddir
   reveals none).

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 Osstest/BuildSupport.pm |    2 +-
 Osstest/TestSupport.pm  |    4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/Osstest/BuildSupport.pm b/Osstest/BuildSupport.pm
index 0938810..874e27e 100644
--- a/Osstest/BuildSupport.pm
+++ b/Osstest/BuildSupport.pm
@@ -74,7 +74,7 @@ sub builddirsprops {
 
 sub prepbuilddirs {
     my (@xbuilddirs) = @_;
-    my $cmd = "rm -rf $builddir && mkdir $builddir";
+    my $cmd = "mkdir -p $builddir && rm -rf $builddir/*-stamp $builddir/dist";
     $cmd .= " && mkdir $builddir/$_" foreach @xbuilddirs;
     target_cmd($ho,$cmd,600);
 }
diff --git a/Osstest/TestSupport.pm b/Osstest/TestSupport.pm
index 22dcbff..3930595 100644
--- a/Osstest/TestSupport.pm
+++ b/Osstest/TestSupport.pm
@@ -1061,9 +1061,12 @@ sub build_clone ($$$$) {
 	}
     }
 
+    my $rm = "rm -rf $subdir";
+
     if ($vcs eq 'hg') {
         
         target_cmd_build($ho, $timeout, $builddir, <<END.
+	    $rm
 	    hg clone '$tree' $subdir
 	    cd $subdir
 END
@@ -1075,6 +1078,7 @@ END
 	my $eff_tree = git_massage_url($tree);
 
         target_cmd_build($ho, $timeout, $builddir, <<END.
+            $rm
             git clone '$eff_tree' $subdir
             cd $subdir
 END
-- 
1.7.10.4

[OSSTEST PATCH 05/13] make-flight: rumpuserxen tests: use same buildjob for all guests

[Ian Jackson]

Change the name of the runvar for the build job to be
guests_rumpuserxenbuildjob, which is the default (fallback) for
guest_var.

(We are going to introduce another step which uses another guest but
wants the same build job.)

No resulting change in the system's overall behaviour.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>

BUILDJOB GUESTS_
---
 make-flight |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/make-flight b/make-flight
index fbda440..8aa5396 100755
--- a/make-flight
+++ b/make-flight
@@ -122,7 +122,7 @@ do_rumpkernel_tests () {
   job_create_test test-$xenarch$kern-$dom0arch-rumpuserxen-$rumparch \
                       test-rumpuserxen xl \
             $xenarch $dom0arch                                       \
-            rump_rumpuserxenbuildjob=build-$rumparch-rumpuserxen     \
+            guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
             all_hostflags=$most_hostflags
 }
 
-- 
1.7.10.4

[OSSTEST PATCH 06/13] ts-xen-build: Allow specification of make targets

[Ian Jackson]

This allows sg-run-job (or a manual caller of ts-xen-build) to specify
which target(s) to pass to `make'.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 ts-xen-build |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ts-xen-build b/ts-xen-build
index 3913273..7c9e804 100755
--- a/ts-xen-build
+++ b/ts-xen-build
@@ -25,6 +25,7 @@ use Osstest::BuildSupport;
 
 tsreadconfig();
 selectbuildhost(\@ARGV);
+# remaining arguments are passed as targets to "make"
 builddirsprops();
     
 sub checkout () {
@@ -108,7 +109,7 @@ END
 END
 #/;
     buildcmd_stamped_logged(9000, 'build', '',<<END,'');
-            make $makeflags
+            make $makeflags @ARGV
 END
 }
 
-- 
1.7.10.4

[OSSTEST PATCH 07/13] ts-xen-build: Honour cmdprefix_configure, cmdprefix_make

[Ian Jackson]

These runvars can be set to specify a string (containing shell code)
to be prefixed to the executions of ./configure and make.

If not set, the behaviour is as as before.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 ts-xen-build |    7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/ts-xen-build b/ts-xen-build
index 7c9e804..661f186 100755
--- a/ts-xen-build
+++ b/ts-xen-build
@@ -94,6 +94,9 @@ sub build () {
     my $xend_opt= $r{enable_xend} =~ m/true/ ? "--enable-xend" : "--disable-xend";
     my $ovmf_opt= $r{enable_ovmf} =~ m/true/ ? "--enable-ovmf" : "--disable-ovmf";
 
+    my $configure_prefix = $r{cmdprefix_configure} // '';
+    my $make_prefix =      $r{cmdprefix_make}      // '';
+
     buildcmd_stamped_logged(600, 'configure', <<END,<<END,<<END);
             if test -f configure; then
                 if grep -q -- $xend_opt tools/configure ; then
@@ -103,13 +106,13 @@ sub build () {
                     ovmf=$ovmf_opt
                 fi
 END
-                ./configure --sysconfdir=/etc \$xend \$ovmf
+               $configure_prefix ./configure --sysconfdir=/etc \$xend \$ovmf
 END
             fi
 END
 #/;
     buildcmd_stamped_logged(9000, 'build', '',<<END,'');
-            make $makeflags @ARGV
+            $make_prefix make $makeflags @ARGV
 END
 }
 
-- 
1.7.10.4

[OSSTEST PATCH 08/13] rump kernel tests: Do a xen.git tools build

[Ian Jackson]

Having laid the groundwork, we can now:

* In make-flight, for the rumpuserxen builds, set the runvars that
  specify which xen.git bits etc. to build.  (We pass qemuu now,
  because we intend that eventually we will want to build that too.)

* In ts-rumpuserxen-build, record the locations of the app-tools
  configure and make wrappers in the runvars which will be honoured by
  ts-xen-build.  Thus ts-xen-build in rumpuserxen build jobs will use
  these app-tools and do a rump cross build.

* Add a ts-xen-build step to the build-rumpuserxen recipe, passing
  `tools' as the target to build.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 mfi-common           |    4 ++++
 sg-run-job           |    1 +
 ts-rumpuserxen-build |    7 +++++++
 3 files changed, 12 insertions(+)

diff --git a/mfi-common b/mfi-common
index 2055bab..aba0b15 100644
--- a/mfi-common
+++ b/mfi-common
@@ -204,6 +204,10 @@ create_build_jobs () {
                 tree_rumpuserxen=$TREE_RUMPUSERXEN			\
           ${TREEVCS_RUMPUSERXEN:+treevcs_rumpuserxen=}${TREEVCS_RUMPUSERXEN} \
                 revision_rumpuserxen=$REVISION_RUMPUSERXEN		\
+                tree_xen=$TREE_XEN                                           \
+                tree_qemuu=$TREE_QEMU_UPSTREAM                               \
+                revision_xen=$REVISION_XEN                                   \
+                revision_qemuu=$REVISION_QEMU_UPSTREAM                       \
                 tree_rumpuserxen_rumpsrc=$TREE_RUMPUSERXEN_RUMPSRC	\
                 ${TREEVCS_RUMPUSERXEN_RUMPSRC:+treevcs_rumpuserxen_rumpsrc=}${TREEVCS_RUMPUSERXEN_RUMPSRC}
                 # rumpsrc-related runvars needed only for old rumpuser-xen
diff --git a/sg-run-job b/sg-run-job
index b02a755..ab9f333 100755
--- a/sg-run-job
+++ b/sg-run-job
@@ -338,6 +338,7 @@ proc run-job/build-libvirt {} {
 
 proc run-job/build-rumpuserxen {} {
     run-ts . = ts-rumpuserxen-build
+    run-ts . = ts-xen-build + host tools
 }
 
 proc prepare-build-host {} {
diff --git a/ts-rumpuserxen-build b/ts-rumpuserxen-build
index a1f31cd..e0b18a7 100755
--- a/ts-rumpuserxen-build
+++ b/ts-rumpuserxen-build
@@ -97,6 +97,12 @@ sub build() {
 END
 }
 
+sub recordtools() {
+    my $apptool = "$rux/app-tools/rumpxen-app";
+    store_runvar('cmdprefix_configure', "$apptool-configure");
+    store_runvar('cmdprefix_make',      "$apptool-make");
+}
+
 sub install() {
     my $destdir="dist/usr/local/lib/xen/rump-kernel";
     target_cmd_build($ho, 300, $builddir, <<END);
@@ -110,5 +116,6 @@ END
 checkout();
 massage();
 build();
+recordtools();
 install();
 built_stash($ho, $builddir, 'dist', 'rumpuserxendist');
-- 
1.7.10.4

[OSSTEST PATCH 09/13] rump kernel tests: Use guest-specific fs images location

[Ian Jackson]

We are going to want to have multiple different guests in the same
job.  Don't use the same locations for all of their fs images.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 ts-rumpuserxen-demo-setup |   11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/ts-rumpuserxen-demo-setup b/ts-rumpuserxen-demo-setup
index 4ad77f8..68ec7d4 100755
--- a/ts-rumpuserxen-demo-setup
+++ b/ts-rumpuserxen-demo-setup
@@ -42,7 +42,7 @@ sub prep () {
 
     my @images;
 
-    my $jobdir = target_jobdir($ho);
+    my $fsimagesdir = target_jobdir($ho)."/$gn-images";
 
     my $cfgfile = "$gn.cfg";
     my $cfgpath = "/etc/xen/$cfgfile";
@@ -62,7 +62,7 @@ sub prep () {
 	    if ($in eq 'disk') {
 		s{\b img/(\w+.ffs) \b}{
                     push @images, $1;
-                    "$jobdir/$1";
+                    "$fsimagesdir/$1";
                 }xeg;
 	    }
 
@@ -72,8 +72,11 @@ sub prep () {
 
     store_runvar("$gho->{Guest}_cfgpath", $cfgpath);
 
-    target_cmd($ho, "cp $rkdist/img/$_ $jobdir/$_", 200)
-	foreach @images;
+    target_cmd($ho, <<END.(join "\n", map { <<END; } @images), 200);
+                   mkdir -p $fsimagesdir
+END
+                   cp $cfg_rkdist/img/$_ $fsimagesdir/$_
+END
 }
 
 prep();
-- 
1.7.10.4

[OSSTEST PATCH 10/13] rump kernel tests: Specify kernel via runvar

[Ian Jackson]

The runvar <gn>_builtimage is set to <distpart>:<distpath>.
Previously we hardcoded `/usr/local/lib/xen/rump-kernel' in
`rumpuserxen'.

The ffs filesystem images, and the demo config template, still come
from a single particular build, which we find from the
`guests_rumpuserxenbuildjob' runvar.  This runvar is then doing double
duty as the build job for the ffs filesystem images and example
configs, and the `default' build job for guest kernels.

The subpath /usr/local/lib/xen/rump-kernel is duplicated, because it
appears once (in make-flight, for setting a runvar) to find the demo
guest's kernel, and once (hardcoded in ts-rumpuserxen-demo-setup) for
the filesystem images and config file.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 make-flight               |    1 +
 ts-rumpuserxen-demo-setup |   17 +++++++++++++----
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/make-flight b/make-flight
index 8aa5396..cede756 100755
--- a/make-flight
+++ b/make-flight
@@ -123,6 +123,7 @@ do_rumpkernel_tests () {
                       test-rumpuserxen xl \
             $xenarch $dom0arch                                       \
             guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
+            rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
             all_hostflags=$most_hostflags
 }
 
diff --git a/ts-rumpuserxen-demo-setup b/ts-rumpuserxen-demo-setup
index 68ec7d4..34da7cb 100755
--- a/ts-rumpuserxen-demo-setup
+++ b/ts-rumpuserxen-demo-setup
@@ -35,10 +35,19 @@ our $rkdist;
 sub prep () {
     $gho = prepareguest($ho, $gn, $gn, 4096, undef, 30);
 
+    my $cfg_rkdist = target_extract_jobdistpath_subdir
+	($ho, 'rumpuserxen--base', 'rumpuserxen',
+	 $r{guests_rumpuserxenbuildjob});
+    $cfg_rkdist .= '/usr/local/lib/xen/rump-kernel';
+
     my $buildjob = guest_var($gho, 'rumpuserxenbuildjob', undef);
+    my $builtimage = guest_var($gho, 'builtimage', undef);
+    $builtimage =~ m/\:/ or die "$builtimage ?";
+    my $builtimage_part = $`;
+    my $builtimage_subpath = $'; #';
     $rkdist = target_extract_jobdistpath_subdir
-	($ho, 'rumpuserxen', 'rumpuserxen', $buildjob);
-    $rkdist .= '/usr/local/lib/xen/rump-kernel';
+	($ho, "rumpuserxen-$gn", $builtimage_part, $buildjob);
+    my $imagefile = $rkdist.$builtimage_subpath;
 
     my @images;
 
@@ -46,7 +55,7 @@ sub prep () {
 
     my $cfgfile = "$gn.cfg";
     my $cfgpath = "/etc/xen/$cfgfile";
-    target_editfile_root($ho, "$rkdist/domain_config", $cfgfile, $cfgpath,
+    target_editfile_root($ho, "$cfg_rkdist/domain_config", $cfgfile, $cfgpath,
 			 sub {
 	while (<EI>) {
 	    next if m/^\s*\#/;
@@ -55,7 +64,7 @@ sub prep () {
 	    my $in = $& if m/^\w\S*/;
 
 	    s# = .*              # " = '$gn'"          #xe  if $in eq 'name';
-	    s# =\s* ['"](.*)['"] # " = '$rkdist/$1'"   #xe  if $in eq 'kernel';
+	    s# = .*              # " = '$imagefile'"   #xe  if $in eq 'kernel';
 	    s#\b mac=[0-9a-f:]+  # "mac=$gho->{Ether}" #xe  if $in eq 'vif';
 	    s#\b 4             \b# 3                   #xe  if $in eq 'extra';
 
-- 
1.7.10.4

[OSSTEST PATCH 11/13] rump kernel tests: Specify cmdline via runvar

[Ian Jackson]

The runvar <gn>_cmdline is used for the rumpuserxen guest command
line.  Previously "3" (a binary flags word meaningful to the test/demo
program included in rumpuser-xen.git) was hardcoded.

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 make-flight               |    1 +
 ts-rumpuserxen-demo-setup |    3 ++-
 2 files changed, 3 insertions(+), 1 deletion(-)

diff --git a/make-flight b/make-flight
index cede756..ccfa3e3 100755
--- a/make-flight
+++ b/make-flight
@@ -124,6 +124,7 @@ do_rumpkernel_tests () {
             $xenarch $dom0arch                                       \
             guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
             rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
+            rump_cmdline=3                                           \
             all_hostflags=$most_hostflags
 }
 
diff --git a/ts-rumpuserxen-demo-setup b/ts-rumpuserxen-demo-setup
index 34da7cb..04ed0b9 100755
--- a/ts-rumpuserxen-demo-setup
+++ b/ts-rumpuserxen-demo-setup
@@ -48,6 +48,7 @@ sub prep () {
     $rkdist = target_extract_jobdistpath_subdir
 	($ho, "rumpuserxen-$gn", $builtimage_part, $buildjob);
     my $imagefile = $rkdist.$builtimage_subpath;
+    my $cmdline = guest_var($gho, 'cmdline', undef);
 
     my @images;
 
@@ -66,7 +67,7 @@ sub prep () {
 	    s# = .*              # " = '$gn'"          #xe  if $in eq 'name';
 	    s# = .*              # " = '$imagefile'"   #xe  if $in eq 'kernel';
 	    s#\b mac=[0-9a-f:]+  # "mac=$gho->{Ether}" #xe  if $in eq 'vif';
-	    s#\b 4             \b# 3                   #xe  if $in eq 'extra';
+	    s# = .*              # " = '$cmdline'"     #xe  if $in eq 'extra';
 
 	    if ($in eq 'disk') {
 		s{\b img/(\w+.ffs) \b}{
-- 
1.7.10.4

[OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Jackson]

This just calls xl destroy; it doesn't check that the guest is
running, or on the network, or anything.  (It also doesn't check that
the guest is down afterwards.)

We are going to want this to be able to destroy the rumpuserxen
xenstorels domain, which is short-lived (and will normally have
terminated by the time we want to destroy it).

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 ts-guest-destroy-hard |   27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)
 create mode 100755 ts-guest-destroy-hard

diff --git a/ts-guest-destroy-hard b/ts-guest-destroy-hard
new file mode 100755
index 0000000..5c2430c
--- /dev/null
+++ b/ts-guest-destroy-hard
@@ -0,0 +1,27 @@
+#!/usr/bin/perl -w
+# This is part of "osstest", an automated testing framework for Xen.
+# Copyright (C) 2009-2013 Citrix Inc.
+# 
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+# 
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Affero General Public License for more details.
+# 
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+use strict qw(vars);
+use DBI;
+use Osstest;
+use Osstest::TestSupport;
+
+tsreadconfig();
+
+our ($ho,$gho) = ts_get_host_guest(@ARGV);
+
+guest_destroy($ho, $gho);
-- 
1.7.10.4

[OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo

[Ian Jackson]

We can use the newly-generalised ts-rumpuserxen-demo-setup to set up
the config file etc. to run xenstore-ls from the xen.git build.

We have a bespoke script to actually execute it: we extract the actual
program output from the guest console log, and compare the results to
those we expect (based on running xenstore-ls in dom0).

Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
---
 make-flight                    |    2 +
 sg-run-job                     |    4 ++
 ts-rumpuserxen-demo-xenstorels |  111 ++++++++++++++++++++++++++++++++++++++++
 3 files changed, 117 insertions(+)
 create mode 100755 ts-rumpuserxen-demo-xenstorels

diff --git a/make-flight b/make-flight
index ccfa3e3..b02720f 100755
--- a/make-flight
+++ b/make-flight
@@ -125,6 +125,8 @@ do_rumpkernel_tests () {
             guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
             rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
             rump_cmdline=3                                           \
+            xenstorels_builtimage=:/usr/local/bin/xenstore           \
+            xenstorels_cmdline='ls -fp device'                       \
             all_hostflags=$most_hostflags
 }
 
diff --git a/sg-run-job b/sg-run-job
index ab9f333..2cf810a 100755
--- a/sg-run-job
+++ b/sg-run-job
@@ -315,6 +315,10 @@ proc run-job/test-rumpuserxen {} {
     run-ts . =   ts-rumpuserxen-demo-setup + host $g
     run-ts . =   ts-guest-start            + host $g
     run-ts . =   ts-guest-destroy          + host $g
+    set g xenstorels
+    run-ts . =   ts-rumpuserxen-demo-setup      + host + $g
+    run-ts . =   ts-rumpuserxen-demo-xenstorels + host + $g
+    run-ts . =   ts-guest-destroy-hard          + host + $g
 }
 
 #---------- builds ----------
diff --git a/ts-rumpuserxen-demo-xenstorels b/ts-rumpuserxen-demo-xenstorels
new file mode 100755
index 0000000..07388a4
--- /dev/null
+++ b/ts-rumpuserxen-demo-xenstorels
@@ -0,0 +1,111 @@
+#!/usr/bin/perl -w
+# This is part of "osstest", an automated testing framework for Xen.
+# Copyright (C) 2009-2013 Citrix Inc.
+# 
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU Affero General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+# 
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU Affero General Public License for more details.
+# 
+# You should have received a copy of the GNU Affero General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+use strict qw(vars);
+use DBI;
+use Osstest;
+use Osstest::TestSupport;
+use Data::Dumper;
+
+tsreadconfig();
+
+our ($ho,$gho) = ts_get_host_guest(@ARGV);
+
+our $domid;
+
+our $gn = $gho->{Guest};
+
+sub start () {
+    my $cmd= toolstack()->{Command}." create ".
+        $r{ $gho->{Guest}.'_'. toolstack()->{CfgPathVar} };
+    target_cmd_root($ho, $cmd, 30);
+
+    $domid = target_cmd_output_root($ho,"xl domid $gho->{Guest}");
+}
+
+sub await_end () {
+    poll_loop(30,3, "await ending of $gho->{Guest}", sub {
+        my $st= guest_get_state($ho,$gho);
+	return undef if $st =~ m/s/;
+        return "guest state is $st";
+    });
+}
+
+our (%file);
+our (%output);
+
+sub stash ($$) {
+    my ($data, $what) = @_;
+    my $leaf = "xenstore-ls-device--$gn--$what";
+    my $stashh = open_unique_stashfile(\$leaf);
+    print $stashh $data or die $!;
+    close $stashh or die $!;
+    $file{$what} = "$stash/$leaf";
+}
+
+sub some_xenstorels ($$$) {
+    my ($what, $massage, $cmd) = @_;
+    $output{$what} = target_cmd_output_root($ho,$cmd);
+    stash($output{$what}, "$what-raw");
+    $massage->();
+    $output{$what} = join "\n", sort split /\n/, $output{$what};
+    $output{$what} .= "\n";
+    stash($output{$what}, "$what-massaged");
+}
+
+sub our_xenstorels () {
+    some_xenstorels('ours', sub {
+	$output{ours} =~ s{^/local/domain/$domid/}{}gm;
+    }, <<END);
+        xenstore-ls -fp /local/domain/$domid/device
+END
+}
+
+sub their_xenstorels () {
+    some_xenstorels('theirs', sub {
+	$output{theirs} =~ s{\r\n}{\n}g;
+	while ($output{theirs} =~ m{\n=== calling main\(\) ===\n\n}) {
+	    $output{theirs} = $'; #';
+	}
+	$output{theirs} =~ m{\n=== main\(\) returned (\d+) ===\n} or die;
+	$output{theirs} = $`;
+	die "$1 ?" if $1 ne '0';
+	$output{theirs} =~ s{^STUB \`\`\w+'' called\n}{}mg;
+    }, <<END);
+        cat /var/log/xen/console/guest-$gn.log
+END
+}
+
+sub check_output () {
+    eval {
+	our_xenstorels();
+	their_xenstorels();
+	1;
+    };
+    if ($@) {
+	die Dumper(\%output)."$@";
+    }
+    if ($output{ours} ne $output{theirs}) {
+	system qw(diff -u), map { $file{"$_-massaged"} } qw(ours theirs);
+	die "$? COMPARISON FAILED";
+    }
+}
+
+start();
+await_end();
+check_output();
+logm("all good.");
-- 
1.7.10.4

Re: [OSSTEST PATCH 03/13] sg-run-job: Support toggling `adding' by repeating `+'

[Ian Campbell]

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> Specifying `+' as an argument to spawn-ts suppresses the automatic
> addition of subsequent arguments to the testid.  Make it possible to
> toggle this by repeating `+'.

So that's what that syntax was doing. I had been boggling over that!

> 
> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> ---
>  sg-run-job |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/sg-run-job b/sg-run-job
> index e903869..b02a755 100755
> --- a/sg-run-job
> +++ b/sg-run-job
> @@ -122,7 +122,7 @@ proc spawn-ts {iffail testid ts args} {
>      set host_testid_suffix {}
>      foreach arg $args {
>          if {![string compare + $arg]} {
> -            set adding 0
> +            set adding [expr {!$adding}]
>              continue
>          }
>          lappend real_args $arg

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Campbell]

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> Make prepbuilddirs no longer delete $builddir; instead:
>  - have prepbuilddirs delete *-stamp files and dist/;
>  - have build_clone delete the specific source directories;
>  - any individual ts-*-build scripts that use other things
>    would have to delete them (but searching for $builddir
>    reveals none).

Is it also implied that ts-*-build ought to do some sort of "make clean"
like thing? Deleting dist is only really relevant for ts-xen-build I
think, e.g. I don't think ts-kernel-build uses it.

Re: [OSSTEST PATCH 05/13] make-flight: rumpuserxen tests: use same buildjob for all guests

[Ian Campbell]

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> Change the name of the runvar for the build job to be
> guests_rumpuserxenbuildjob, which is the default (fallback) for
> guest_var.
> 
> (We are going to introduce another step which uses another guest but
> wants the same build job.)
> 
> No resulting change in the system's overall behaviour.
> 
> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> 
> BUILDJOB GUESTS_

Stray?

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Campbell]

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> This just calls xl destroy; it doesn't check that the guest is
> running, or on the network, or anything.  (It also doesn't check that
> the guest is down afterwards.)
> 
> We are going to want this to be able to destroy the rumpuserxen
> xenstorels domain, which is short-lived (and will normally have
> terminated by the time we want to destroy it).

You could use on_poweroff=preserve to avoid racing with that?

> 
> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> ---
>  ts-guest-destroy-hard |   27 +++++++++++++++++++++++++++
>  1 file changed, 27 insertions(+)
>  create mode 100755 ts-guest-destroy-hard
> 
> diff --git a/ts-guest-destroy-hard b/ts-guest-destroy-hard
> new file mode 100755
> index 0000000..5c2430c
> --- /dev/null
> +++ b/ts-guest-destroy-hard
> @@ -0,0 +1,27 @@
> +#!/usr/bin/perl -w
> +# This is part of "osstest", an automated testing framework for Xen.
> +# Copyright (C) 2009-2013 Citrix Inc.

Out of date

Re: [OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo

[Ian Campbell]

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> We can use the newly-generalised ts-rumpuserxen-demo-setup to set up
> the config file etc. to run xenstore-ls from the xen.git build.
> 
> We have a bespoke script to actually execute it: we extract the actual
> program output from the guest console log, and compare the results to
> those we expect (based on running xenstore-ls in dom0).
> 
> Signed-off-by: Ian Jackson <Ian.Jackson@eu.citrix.com>
> ---
>  make-flight                    |    2 +
>  sg-run-job                     |    4 ++
>  ts-rumpuserxen-demo-xenstorels |  111 ++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 117 insertions(+)
>  create mode 100755 ts-rumpuserxen-demo-xenstorels
> 
> diff --git a/make-flight b/make-flight
> index ccfa3e3..b02720f 100755
> --- a/make-flight
> +++ b/make-flight
> @@ -125,6 +125,8 @@ do_rumpkernel_tests () {
>              guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
>              rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
>              rump_cmdline=3                                           \
> +            xenstorels_builtimage=:/usr/local/bin/xenstore           \

Really no distpart here? Doesn't that look in the main build job which
built the regular Xen not the rump one?

Assuming this is a path within the rump build target -- might it be wise
to override --prefix when building that to avoid surprising clashes?

Ian.

Re: [OSSTEST PATCH 00/13] Build and test rump kernel xenstore-ls

[Ian Campbell]

I've commented on a bunch of these. Most of them look good. 

You can add my acked-by to the following (some of them I commented on,
but only in passing):

On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:

>  01/13  selectguest: log a message
>  02/13  BuildSupport: make selectbuildhost modify @ARGV
>  03/13  sg-run-job: Support toggling `adding' by repeating `+'

>  05/13  make-flight: rumpuserxen tests: use same buildjob for all guests

>  06/13  ts-xen-build: Allow specification of make targets
>  07/13  ts-xen-build: Honour cmdprefix_configure, cmdprefix_make
>  08/13  rump kernel tests: Do a xen.git tools build

>  09/13  rump kernel tests: Use guest-specific fs images location
>  10/13  rump kernel tests: Specify kernel via runvar
>  11/13  rump kernel tests: Specify cmdline via runvar
>  12/13  ts-guest-destroy-hard: New script

Ian.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > Make prepbuilddirs no longer delete $builddir; instead:
> >  - have prepbuilddirs delete *-stamp files and dist/;
> >  - have build_clone delete the specific source directories;
> >  - any individual ts-*-build scripts that use other things
> >    would have to delete them (but searching for $builddir
> >    reveals none).
> 
> Is it also implied that ts-*-build ought to do some sort of "make clean"
> like thing? Deleting dist is only really relevant for ts-xen-build I
> think, e.g. I don't think ts-kernel-build uses it.

No, because the builds all happen in trees which have been recloned.

Ian.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Campbell]

On Wed, 2014-07-16 at 10:58 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> > On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > > Make prepbuilddirs no longer delete $builddir; instead:
> > >  - have prepbuilddirs delete *-stamp files and dist/;
> > >  - have build_clone delete the specific source directories;
> > >  - any individual ts-*-build scripts that use other things
> > >    would have to delete them (but searching for $builddir
> > >    reveals none).
> > 
> > Is it also implied that ts-*-build ought to do some sort of "make clean"
> > like thing? Deleting dist is only really relevant for ts-xen-build I
> > think, e.g. I don't think ts-kernel-build uses it.
> 
> No, because the builds all happen in trees which have been recloned.

You mean that currently happens to be the case for ts-foo-build for
foo != xen, for Xen you are in the process or making this not hold, I
think?

Perhaps prepbuilddirs should take a list of paths to clean?

Ian.

Re: [OSSTEST PATCH 05/13] make-flight: rumpuserxen tests: use same buildjob for all guests

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 05/13] make-flight: rumpuserxen tests: use same buildjob for all guests"):
> On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > BUILDJOB GUESTS_
> 
> Stray?

Oops, yes.

Ian.

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script"):
> On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > We are going to want this to be able to destroy the rumpuserxen
> > xenstorels domain, which is short-lived (and will normally have
> > terminated by the time we want to destroy it).
> 
> You could use on_poweroff=preserve to avoid racing with that?

No, because it still insists that the guest is up and responding to
pings etc.

> > diff --git a/ts-guest-destroy-hard b/ts-guest-destroy-hard
> > new file mode 100755
> > index 0000000..5c2430c
> > --- /dev/null
> > +++ b/ts-guest-destroy-hard
> > @@ -0,0 +1,27 @@
> > +#!/usr/bin/perl -w
> > +# This is part of "osstest", an automated testing framework for Xen.
> > +# Copyright (C) 2009-2013 Citrix Inc.
> 
> Out of date

Will fix.

Thanks,
Ian.

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Campbell]

On Wed, 2014-07-16 at 11:10 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script"):
> > On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > > We are going to want this to be able to destroy the rumpuserxen
> > > xenstorels domain, which is short-lived (and will normally have
> > > terminated by the time we want to destroy it).
> > 
> > You could use on_poweroff=preserve to avoid racing with that?
> 
> No, because it still insists that the guest is up and responding to
> pings etc.

Ah, I was thinking of the other thing which jut parses xl list to see if
it exists and in what state.

> 
> > > diff --git a/ts-guest-destroy-hard b/ts-guest-destroy-hard
> > > new file mode 100755
> > > index 0000000..5c2430c
> > > --- /dev/null
> > > +++ b/ts-guest-destroy-hard
> > > @@ -0,0 +1,27 @@
> > > +#!/usr/bin/perl -w
> > > +# This is part of "osstest", an automated testing framework for Xen.
> > > +# Copyright (C) 2009-2013 Citrix Inc.
> > 
> > Out of date
> 
> Will fix.
> 
> Thanks,
> Ian.

Re: [OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo"):
> On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > We can use the newly-generalised ts-rumpuserxen-demo-setup to set up
> > the config file etc. to run xenstore-ls from the xen.git build.
...
> >              guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
> >              rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
> >              rump_cmdline=3                                           \> > +            xenstorels_builtimage=:/usr/local/bin/xenstore           \
> 
> Really no distpart here? Doesn't that look in the main build job which
> built the regular Xen not the rump one?

No, the job is that specified in guests_rumpuserxenbuildjob.

The distpart is <foo> in the runvar path_<foo>dist.  In this case we
want the main output rather than the hypervisor (path_xendist, not
built because we say `make tools') or kernel (path_kerndist, not built
nowadays).  So the distpart is `'.

> Assuming this is a path within the rump build target -- might it be wise
> to override --prefix when building that to avoid surprising clashes?

I think it is easier to deal with this on the output end.  If you
install the rump xen.git build's dist/install/ in your system's / you
have only yourself to blame.

Ian.

Re: [OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo

[Ian Campbell]

On Wed, 2014-07-16 at 11:15 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 13/13] rump kernel tests: Run xenstore-ls demo"):
> > On Tue, 2014-07-15 at 16:22 +0100, Ian Jackson wrote:
> > > We can use the newly-generalised ts-rumpuserxen-demo-setup to set up
> > > the config file etc. to run xenstore-ls from the xen.git build.
> ...
> > >              guests_rumpuserxenbuildjob=build-$rumparch-rumpuserxen   \
> > >              rump_builtimage=rumpuserxen:/usr/local/lib/xen/rump-kernel \
> > >              rump_cmdline=3                                           \> > +            xenstorels_builtimage=:/usr/local/bin/xenstore           \
> > 
> > Really no distpart here? Doesn't that look in the main build job which
> > built the regular Xen not the rump one?
> 
> No, the job is that specified in guests_rumpuserxenbuildjob.
> 
> The distpart is <foo> in the runvar path_<foo>dist.  In this case we
> want the main output rather than the hypervisor (path_xendist, not
> built because we say `make tools') or kernel (path_kerndist, not built
> nowadays).  So the distpart is `'.

Ah yes, I was confusing this with ${thing}buildjob.

> > Assuming this is a path within the rump build target -- might it be wise
> > to override --prefix when building that to avoid surprising clashes?
> 
> I think it is easier to deal with this on the output end.  If you
> install the rump xen.git build's dist/install/ in your system's / you
> have only yourself to blame.

OK.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> On Wed, 2014-07-16 at 10:58 +0100, Ian Jackson wrote:
> > Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> > > Is it also implied that ts-*-build ought to do some sort of "make clean"
> > > like thing? Deleting dist is only really relevant for ts-xen-build I
> > > think, e.g. I don't think ts-kernel-build uses it.
> > 
> > No, because the builds all happen in trees which have been recloned.
> 
> You mean that currently happens to be the case for ts-foo-build for
> foo != xen, for Xen you are in the process or making this not hold, I
> think?

No.  build_clone always gives you a fresh tree.  ts-xen-build always
builds in a directory from build_clone.

The rumpuserxen xen.git build is done (also, must be done) in a tree
which has not had any ordinary build done in it.

> Perhaps prepbuilddirs should take a list of paths to clean?

That would be fine if it was in fact necessary to use that feature
:-).

Ian.

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script"):
> On Wed, 2014-07-16 at 11:10 +0100, Ian Jackson wrote:
> > No, because it still insists that the guest is up and responding to
> > pings etc.
> 
> Ah, I was thinking of the other thing which jut parses xl list to see if
> it exists and in what state.

That wouldn't work either because the state would be `crashed'.

The rump environment currently uses minios's exit arrangements, and
the authors of minios apparently felt that if you return from main
you wanted a stack trace...

Ian.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Campbell]

On Wed, 2014-07-16 at 11:18 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> > On Wed, 2014-07-16 at 10:58 +0100, Ian Jackson wrote:
> > > Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> > > > Is it also implied that ts-*-build ought to do some sort of "make clean"
> > > > like thing? Deleting dist is only really relevant for ts-xen-build I
> > > > think, e.g. I don't think ts-kernel-build uses it.
> > > 
> > > No, because the builds all happen in trees which have been recloned.
> > 
> > You mean that currently happens to be the case for ts-foo-build for
> > foo != xen, for Xen you are in the process or making this not hold, I
> > think?
> 
> No.  build_clone always gives you a fresh tree.  ts-xen-build always
> builds in a directory from build_clone.
> 
> The rumpuserxen xen.git build is done (also, must be done) in a tree
> which has not had any ordinary build done in it.

OK, then I'm just confused about the purpose of this change ;-)

I thought you were reusing the build tree.

Ian.

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Campbell]

On Wed, 2014-07-16 at 11:20 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script"):
> > On Wed, 2014-07-16 at 11:10 +0100, Ian Jackson wrote:
> > > No, because it still insists that the guest is up and responding to
> > > pings etc.
> > 
> > Ah, I was thinking of the other thing which jut parses xl list to see if
> > it exists and in what state.
> 
> That wouldn't work either because the state would be `crashed'.
> 
> The rump environment currently uses minios's exit arrangements, and
> the authors of minios apparently felt that if you return from main
> you wanted a stack trace...

As in SCHEDOP_crash? How exciting... I stupidly assumed it would do the
sane thing and SCHEDOP_shutdown...

Ian.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> On Wed, 2014-07-16 at 11:18 +0100, Ian Jackson wrote:
> > The rumpuserxen xen.git build is done (also, must be done) in a tree
> > which has not had any ordinary build done in it.
> 
> OK, then I'm just confused about the purpose of this change ;-)
> 
> I thought you were reusing the build tree.

This change is needed because previously prepbuilddirs would delete
the whole per-job directory.

I will add some text to the commit messages to clarify this:

  building: Delete when cloning, not when preparing $builddir
  ...

  This makes it possible for later steps within a build job to look at
  the build trees from previous steps.  This is helpful if for any
  reason it isn't easy or desirable to package up a formal set of build
  outputs for reuse.

And

  rump kernel tests: Do a xen.git tools build
  ...

  ts-rumpuserxen-build does not generate a formal set of build outputs.
  Instead, the ts-xen-build step references the files generated by
  ts-rumpuserxen-build directly (and is told where to find them via the
  new command prefix runvars).

Ian.

Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script

[Ian Jackson]

Ian Campbell writes ("Re: [OSSTEST PATCH 12/13] ts-guest-destroy-hard: New script"):
> On Wed, 2014-07-16 at 11:20 +0100, Ian Jackson wrote:
> > The rump environment currently uses minios's exit arrangements, and
> > the authors of minios apparently felt that if you return from main
> > you wanted a stack trace...
> 
> As in SCHEDOP_crash? How exciting... I stupidly assumed it would do the
> sane thing and SCHEDOP_shutdown...

Yes, indeed.  Quite.

Ian.

Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir

[Ian Campbell]

On Wed, 2014-07-16 at 11:29 +0100, Ian Jackson wrote:
> Ian Campbell writes ("Re: [OSSTEST PATCH 04/13] building: Delete when cloning, not when preparing $builddir"):
> > On Wed, 2014-07-16 at 11:18 +0100, Ian Jackson wrote:
> > > The rumpuserxen xen.git build is done (also, must be done) in a tree
> > > which has not had any ordinary build done in it.
> > 
> > OK, then I'm just confused about the purpose of this change ;-)
> > 
> > I thought you were reusing the build tree.
> 
> This change is needed because previously prepbuilddirs would delete
> the whole per-job directory.

Ah, I see.

> I will add some text to the commit messages to clarify this:
> 
>   building: Delete when cloning, not when preparing $builddir
>   ...
> 
>   This makes it possible for later steps within a build job to look at
>   the build trees from previous steps.  This is helpful if for any
>   reason it isn't easy or desirable to package up a formal set of build
>   outputs for reuse.

Thanks, makes sense.

> 
> And
> 
>   rump kernel tests: Do a xen.git tools build
>   ...
> 
>   ts-rumpuserxen-build does not generate a formal set of build outputs.
>   Instead, the ts-xen-build step references the files generated by
>   ts-rumpuserxen-build directly (and is told where to find them via the
>   new command prefix runvars).
> 
> Ian.