Re: [PATCH v4 1/3] test: Extend test scripts to allow running exeter tests.

On Wed, Aug 20, 2025 at 11:52:18AM +0200, Stefano Brivio wrote:

On Wed, 20 Aug 2025 12:55:58 +1000 David Gibson wrote:

...

On Tue, Aug 19, 2025 at 04:27:38PM +0200, Stefano Brivio wrote:

...

On Thu, 7 Aug 2025 21:32:35 +1000 David Gibson wrote:

...

Introduce some trivial testcases based on the exeter library. These run passt and pasta with --help and --version options. Extend our test scripts to run these tests.

Signed-off-by: David Gibson --- test/.gitignore | 1 + test/Makefile | 5 ++++- test/lib/exeter | 46 +++++++++++++++++++++++++++++++++++++++++++++ test/run | 9 ++++++++- test/smoke/smoke.sh | 27 ++++++++++++++++++++++++++ 5 files changed, 86 insertions(+), 2 deletions(-) create mode 100644 test/lib/exeter create mode 100755 test/smoke/smoke.sh

diff --git a/test/.gitignore b/test/.gitignore index 3573444f..cf48b885 100644 --- a/test/.gitignore +++ b/test/.gitignore @@ -11,3 +11,4 @@ nstool rampstream guest-key guest-key.pub +/exeter/ diff --git a/test/Makefile b/test/Makefile index bf63db87..332f3f3e 100644 --- a/test/Makefile +++ b/test/Makefile @@ -50,7 +50,7 @@ UBUNTU_NEW_IMGS = xenial-server-cloudimg-powerpc-disk1.img \ jammy-server-cloudimg-s390x.img UBUNTU_IMGS = $(UBUNTU_OLD_IMGS) $(UBUNTU_NEW_IMGS)

-DOWNLOAD_ASSETS = mbuto podman \ +DOWNLOAD_ASSETS = exeter mbuto podman \ $(DEBIAN_IMGS) $(FEDORA_IMGS) $(OPENSUSE_IMGS) $(UBUNTU_IMGS)

Pre-existing, but this broke a while after you switched the image preparation to Makefile targets, so you can't really expect people to issue 'make' at the moment.

Sorry, I'm not quite sure what the problem is you're getting at.

This: --

Hm, could you avoid using -- to mark quotes like this. mutt mistakes it for the sig and cuts it out; it's kind of a pain to paste it back in for repying.

$ mkdir psst $ cd psst $ git clone git://passt.top/passt [...] $ cd passt/test/ $ make [...] wget -c -O debian-11-generic-ppc64el.qcow2 https://cloud.debian.org/images/cloud/bullseye/latest/debian-11-generic-ppc6... --2025-08-20 11:27:05-- https://cloud.debian.org/images/cloud/bullseye/latest/debian-11-generic-ppc6... Resolving cloud.debian.org (cloud.debian.org)... 2001:6b0:19::165, 2001:6b0:19::173, 2001:6b0:19::163, ... Connecting to cloud.debian.org (cloud.debian.org)|2001:6b0:19::165|:443... connected. HTTP request sent, awaiting response... 404 Not Found 2025-08-20 11:27:05 ERROR 404: Not Found.

make: *** [Makefile:150: debian-11-generic-ppc64el.qcow2] Error 8

Ah, right. All the distro wgets are so slow from Australia that I've always avoided them by keeping the image files in place (or even copying the back from backups). So, I didn't hit this. Yeah.. we should do something about that.

The (very) old logic would make the "distro" tests fail, but it's been a while that one can't really run 'make' under test/ anymore (unless you have everything already prepared, that is).

On one hand, we don't have a replacement for those tests, and, while ugly and clumsy, they were actually useful as they would have told us in advance about stuff such as:

https://bugs.passt.top/show_bug.cgi?id=121

...and probably much more that we're ignoring now. I don't even test routinely on big-endian, we just have build tests from distributions. That's bad, and it will be ages before we can reasonably get a replacement for that.

Yeah. On the other hand, we could at least remove the links that are actually broken. The idea of testing across distros seems to me more important than the exact set of distros/versions to check - that's relatively easily expanded if we have the structure.

So it would be still valuable to get them working again, and it shouldn't be much effort.

On the other hand... you can't do 'make' there, because of tests we don't even run. Maybe those failures should be warnings instead. If somebody finds a "moment" it would be nice to fix those though...

...

...

I ran 'make exeter' to get further. It would be nice to update test/README.md (or fix the Makefile... or both).

Hrm.. "make check" already depends on "make assets" which should (now) include exeter. I'm not sure why that's not working for you.

...and not 'make assets', either.

...

...

...

TESTDATA_ASSETS = small.bin big.bin medium.bin \ rampstream @@ -70,6 +70,9 @@ assets: $(ASSETS) pull-%: % git -C $* pull

+exeter: + git clone https://gitlab.com/dgibson/exeter.git + mbuto: git clone git://mbuto.sh/mbuto

diff --git a/test/lib/exeter b/test/lib/exeter new file mode 100644 index 00000000..4f7bcff9 --- /dev/null +++ b/test/lib/exeter @@ -0,0 +1,46 @@ +#!/bin/sh +# +# SPDX-License-Identifier: GPL-2.0-or-later +# +# PASST - Plug A Simple Socket Transport +# for qemu/UNIX domain socket mode +# +# PASTA - Pack A Subtle Tap Abstraction +# for network namespace/tap device mode +# +# test/lib/exeter - Run exeter tests within the rest of passt's tests +# +# Copyright Red Hat +# Author: David Gibson + +exeter() {

It would be nice to have the usual arguments / function description here.

Good point, added.

...

...

+ STATESETUP="${STATEBASE}/$1" + mkdir -p "${STATESETUP}" + + context_setup_host host + layout_host

Indentation here was done with tabs, this adds spaces instead.

Oops. For some reason I'd added emacs configuration to explicitly *disable* tab indent mode for shell mode in the passt tree. Weird, and now reversed.

...

...

+ + cd test + + __ntests=$("$@" --list | wc -l) + if [ $? != 0 ]; then + info "Failed to get exeter manifest for $@" + pause_continue \ + "Press any key to pause test session" \ + "Resuming in " \ + "Paused, press any key to continue" \ + 5 + return + fi + + status_file_start "Exeter tests: $*" ${__ntests} + + for __testid in $("$@" --list); do + status_test_start "${__testid}" + context_run host "$@" "${__testid}" && status_test_ok || status_test_fail + done + + cd .. + + teardown_context_watch ${PANE_HOST} host +} diff --git a/test/run b/test/run index f73c3119..9e183682 100755 --- a/test/run +++ b/test/run @@ -53,6 +53,7 @@ COMMIT="$(git log --oneline --no-decorate -1)" . lib/layout_ugly . lib/test . lib/video +. lib/exeter

# cleanup() - Remove temporary files cleanup() { @@ -67,6 +68,8 @@ run() { perf_init [ ${CI} -eq 1 ] && video_start ci

+ exeter smoke/smoke.sh + setup build test build/all test build/cppcheck @@ -223,6 +226,10 @@ run_selected() {

__setup= for __test; do + if "test/${__test}" --list; then + exeter "${__test}" + continue + fi # HACK: the migrate tests need the setup repeated for # each test if [ "${__test%%/*}" != "${__setup}" -o \ @@ -234,7 +241,7 @@ run_selected() {

test "${__test}" done - teardown "${__setup}" + [ -n "${__setup}" ] && teardown "${__setup}"

log "PASS: ${STATUS_PASS}, FAIL: ${STATUS_FAIL}, SKIPPED: ${STATUS_SKIPPED}"

diff --git a/test/smoke/smoke.sh b/test/smoke/smoke.sh new file mode 100755 index 00000000..26a98d31 --- /dev/null +++ b/test/smoke/smoke.sh @@ -0,0 +1,27 @@ +#! /bin/sh

...this is desirable, I think (and, for consistency, I'd use "#!/bin/sh") as Bash is not universally available, but:

So.. I believe "#! " (with the space) is actually the traditional / correct version. I did once encounter a situation where it mattered, although to be fair, that might have been with 25 year old SunOS or something equally obscure. Nonetheless I've habitually included the space ever since (see find-arm64-firmware.sh, prepare-distro-img.sh and rapstream-check.sh).

Oh, I didn't know. I thought it was just something that happened to work somehow, but I didn't realise about this:

https://en.wikipedia.org/wiki/Shebang_(Unix)#Version_8_improved_shell_script...

and while this has some details too:

https://www.in-ulm.de/~mascheck/various/shebang/#blankrequired

...sure, makes sense, let's keep those.

-- Stefano

-- David Gibson (he or they) | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you, not the other way | around. http://www.ozlabs.org/~dgibson