| Message ID | 20190722140544.29867-1-joe.lawrence@redhat.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | [v3] selftests/livepatch: add test skip handling | expand |
Hi Joe, On 7/22/19 8:05 AM, Joe Lawrence wrote: > Add a skip() message function that stops the test, logs an explanation, > and sets the "skip" return code (4). > > Before loading a livepatch self-test kernel module, first verify that > we've built and installed it by running a 'modprobe --dry-run'. This > should catch a few environment issues, including !CONFIG_LIVEPATCH and > !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new > skip() function. > > Reported-by: Jiri Benc <jbenc@redhat.com> > Suggested-by: Shuah Khan <shuah@kernel.org> > Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> > --- > > v3: tweak modprobe error message: check kernel config and run as root, > so output now looks like [shuah] : > Couple of small tweaks to user visible message below. > % make run_tests > TAP version 13 > 1..3 > # selftests: livepatch: test-livepatch.sh > # TEST: basic function patching ... SKIP: unable load module test_klp_livepatch, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root > not ok 1 selftests: livepatch: test-livepatch.sh # SKIP unable to > # selftests: livepatch: test-callbacks.sh > # TEST: target module before livepatch ... SKIP: unable load module test_klp_callbacks_mod, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root unable to > not ok 2 selftests: livepatch: test-callbacks.sh # SKIP > # selftests: livepatch: test-shadow-vars.sh > # TEST: basic shadow variable API ... SKIP: unable load module test_klp_shadow_vars, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root > unable to thanks for taking care of this. -- Shuah
On Mon 2019-07-22 10:05:44, Joe Lawrence wrote: > Add a skip() message function that stops the test, logs an explanation, > and sets the "skip" return code (4). > > Before loading a livepatch self-test kernel module, first verify that > we've built and installed it by running a 'modprobe --dry-run'. This > should catch a few environment issues, including !CONFIG_LIVEPATCH and > !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new > skip() function. > > Reported-by: Jiri Benc <jbenc@redhat.com> > Suggested-by: Shuah Khan <shuah@kernel.org> > Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> Reviewed-by: Petr Mladek <pmladek@suse.com> Best Regards, Petr PS: No need to send v4. The missing "to", suggested by Shuah, could get added when pushing the patch into the repo.
On 7/24/19 12:27 AM, Petr Mladek wrote: > On Mon 2019-07-22 10:05:44, Joe Lawrence wrote: >> Add a skip() message function that stops the test, logs an explanation, >> and sets the "skip" return code (4). >> >> Before loading a livepatch self-test kernel module, first verify that >> we've built and installed it by running a 'modprobe --dry-run'. This >> should catch a few environment issues, including !CONFIG_LIVEPATCH and >> !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new >> skip() function. >> >> Reported-by: Jiri Benc <jbenc@redhat.com> >> Suggested-by: Shuah Khan <shuah@kernel.org> >> Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> > > Reviewed-by: Petr Mladek <pmladek@suse.com> > > Best Regards, > Petr > > PS: No need to send v4. The missing "to", suggested by Shuah, could > get added when pushing the patch into the repo. > I can't make changes to the patch. I can adjust the commit log at times and correct merge conflicts. I would like to see v4 for this. thanks, -- Shuah
On 07/24/2019 12:45 PM, shuah wrote: > On 7/24/19 12:27 AM, Petr Mladek wrote: >> On Mon 2019-07-22 10:05:44, Joe Lawrence wrote: >>> Add a skip() message function that stops the test, logs an explanation, >>> and sets the "skip" return code (4). >>> >>> Before loading a livepatch self-test kernel module, first verify that >>> we've built and installed it by running a 'modprobe --dry-run'. This >>> should catch a few environment issues, including !CONFIG_LIVEPATCH and >>> !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new >>> skip() function. >>> >>> Reported-by: Jiri Benc <jbenc@redhat.com> >>> Suggested-by: Shuah Khan <shuah@kernel.org> >>> Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> >> >> Reviewed-by: Petr Mladek <pmladek@suse.com> >> >> Best Regards, >> Petr >> >> PS: No need to send v4. The missing "to", suggested by Shuah, could >> get added when pushing the patch into the repo. >> > > I can't make changes to the patch. I can adjust the commit log at times > and correct merge conflicts. > > I would like to see v4 for this. > We should clarify for which tree this and "[PATCH] selftests/livepatch: push and pop dynamic debug config" would be merged. I had assumed this would go to the livepatching tree, but Shuah, do you prefer that selftest-centric patches go through the selftests tree? I'm okay with either and can send up a v4 if needed. But also this would be good to know for future changesets like this, we can perhaps mark the destination branch in advance. Thanks, -- Joe
On 7/24/19 12:45 PM, Joe Lawrence wrote: > On 07/24/2019 12:45 PM, shuah wrote: >> On 7/24/19 12:27 AM, Petr Mladek wrote: >>> On Mon 2019-07-22 10:05:44, Joe Lawrence wrote: >>>> Add a skip() message function that stops the test, logs an explanation, >>>> and sets the "skip" return code (4). >>>> >>>> Before loading a livepatch self-test kernel module, first verify that >>>> we've built and installed it by running a 'modprobe --dry-run'. This >>>> should catch a few environment issues, including !CONFIG_LIVEPATCH and >>>> !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new >>>> skip() function. >>>> >>>> Reported-by: Jiri Benc <jbenc@redhat.com> >>>> Suggested-by: Shuah Khan <shuah@kernel.org> >>>> Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> >>> >>> Reviewed-by: Petr Mladek <pmladek@suse.com> >>> >>> Best Regards, >>> Petr >>> >>> PS: No need to send v4. The missing "to", suggested by Shuah, could >>> get added when pushing the patch into the repo. >>> >> >> I can't make changes to the patch. I can adjust the commit log at times >> and correct merge conflicts. >> >> I would like to see v4 for this. >> > > We should clarify for which tree this and "[PATCH] selftests/livepatch: > push and pop dynamic debug config" would be merged. I had assumed this > would go to the livepatching tree, but Shuah, do you prefer that > selftest-centric patches go through the selftests tree? > > I'm okay with either and can send up a v4 if needed. But also this > would be good to know for future changesets like this, we can perhaps > mark the destination branch in advance. > I prefer to take these through kselftest tree unless there is a dependency on other trees. It is the case for some feature dependent tests that might have dependency on the feature. This doesn't appear to have any dependency. thanks, -- Shuah
diff --git a/tools/testing/selftests/livepatch/functions.sh b/tools/testing/selftests/livepatch/functions.sh index 30195449c63c..8eb21fcc71de 100644 --- a/tools/testing/selftests/livepatch/functions.sh +++ b/tools/testing/selftests/livepatch/functions.sh @@ -13,6 +13,14 @@ function log() { echo "$1" > /dev/kmsg } +# skip(msg) - testing can't proceed +# msg - explanation +function skip() { + log "SKIP: $1" + echo "SKIP: $1" >&2 + exit 4 +} + # die(msg) - game over, man # msg - dying words function die() { @@ -43,6 +51,12 @@ function loop_until() { done } +function assert_mod() { + local mod="$1" + + modprobe --dry-run "$mod" &>/dev/null +} + function is_livepatch_mod() { local mod="$1" @@ -75,6 +89,9 @@ function __load_mod() { function load_mod() { local mod="$1"; shift + assert_mod "$mod" || + skip "unable load module ${mod}, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root" + is_livepatch_mod "$mod" && die "use load_lp() to load the livepatch module $mod" @@ -88,6 +105,9 @@ function load_mod() { function load_lp_nowait() { local mod="$1"; shift + assert_mod "$mod" || + skip "unable load module ${mod}, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root" + is_livepatch_mod "$mod" || die "module $mod is not a livepatch"
Add a skip() message function that stops the test, logs an explanation, and sets the "skip" return code (4). Before loading a livepatch self-test kernel module, first verify that we've built and installed it by running a 'modprobe --dry-run'. This should catch a few environment issues, including !CONFIG_LIVEPATCH and !CONFIG_TEST_LIVEPATCH. In these cases, exit gracefully with the new skip() function. Reported-by: Jiri Benc <jbenc@redhat.com> Suggested-by: Shuah Khan <shuah@kernel.org> Signed-off-by: Joe Lawrence <joe.lawrence@redhat.com> --- v3: tweak modprobe error message: check kernel config and run as root, so output now looks like [shuah] : % make run_tests TAP version 13 1..3 # selftests: livepatch: test-livepatch.sh # TEST: basic function patching ... SKIP: unable load module test_klp_livepatch, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root not ok 1 selftests: livepatch: test-livepatch.sh # SKIP # selftests: livepatch: test-callbacks.sh # TEST: target module before livepatch ... SKIP: unable load module test_klp_callbacks_mod, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root not ok 2 selftests: livepatch: test-callbacks.sh # SKIP # selftests: livepatch: test-shadow-vars.sh # TEST: basic shadow variable API ... SKIP: unable load module test_klp_shadow_vars, verify CONFIG_TEST_LIVEPATCH=m and run self-tests as root not ok 3 selftests: livepatch: test-shadow-vars.sh # SKIP v2: move assert_mod() call into load_mod() and load_lp_nowait(), before they check whether the module is a livepatch or not (a test-failing assertion). [mbenes, pmladek] .../testing/selftests/livepatch/functions.sh | 20 +++++++++++++++++++ 1 file changed, 20 insertions(+)