hello Pedro, thanks for your answer!
On Tue, Jan 23, 2024 at 5:47 PM Pedro Tammela pctammela@mojatatu.com wrote:
On 23/01/2024 10:17, Davide Caratti wrote:
hi Pedro,
On Tue, Jan 23, 2024 at 1:28 PM Pedro Tammela pctammela@mojatatu.com wrote:
If 'jq' is not available the taprio tests that use this script will run forever. Check if it exists before entering the while loop.
[...]
nit: what about returning $KSFT_SKIP (that is 4) if jq is not there? so the test does not fail. thanks!
Since these scripts are run in the setup phase, it has a special treatment.
Take for example this run: ok 1 ba39 - Add taprio Qdisc to multi-queue device (8 queues) ok 2 9462 - Add taprio Qdisc with multiple sched-entry ok 3 8d92 - Add taprio Qdisc with txtime-delay ok 4 d092 - Delete taprio Qdisc with valid handle ok 5 8471 - Show taprio class ok 6 0a85 - Add taprio Qdisc to single-queue device ok 7 3e1e - Add taprio Qdisc with an invalid cycle-time ok 8 39b4 - Reject grafting taprio as child qdisc of software taprio # skipped - "-----> prepare stage" did not complete successfully
ok 9 e8a1 - Reject grafting taprio as child qdisc of offloaded taprio # skipped - skipped - previous setup failed 9 39b4
[...]
As of today it returns 0, success in ksft, even though it clearly wasn't. Looking at the code any failures in the setup/teardown phase will stop the run, skip all the remaining tests but still return success.
About returning skip from the script, aside from marking it as skip and continuing the suite, we would need to run a silent teardown, one that executes all commands in the specified teardown but ignores errors. In this case we are assuming all setup steps follow KSFT return codes. Not sure if it it's reasonable or not...
wouldn't this be fixed by adding this line:
"dependsOn" : "command -v jq >/dev/null"
to test scenarios 39b4 and e8a1 ? I'm asking this because jq is used also in verifyCmd after the script, to parse results. Background for this question: I see tdc skipping both setup and teardown stages for each test case in taprio.json where this line:
"dependsOn": "command -v ciao >/dev/null",
is present. Rather than doing a setup + silent teardown, just do nothing and go to the next test.
As your suggestion is not a blocker, I would rather address the above problems in a follow up series since they will require some refactoring. WDYT?
no objections, but I'm curious to see if "dependsOn" would fix this case :) thanks!