Improve QEMU execution by henrybear327 · Pull Request #382 · sysprog21/elfuse

henrybear327 · 2026-09-14T16:36:28Z

Improve error reporting from qemu execution, in preparation for conformance test failed runs.

Summary by cubic

Improves qemu-runner.sh lifecycle handling and diagnostics so a VM can be started and stopped from different shells without risking cleanup of a recycled PID. It also centralizes the test VM's SSH options and adds no-VM regression coverage.

start --state-file PATH records the port, key, and pidfile; stop --state-file PATH validates process identity before signaling and preserves state if SIGKILL fails.
Failed boots and guest /tmp setup now report the serial console and QEMU output before cleanup.
Both QEMU lanes use shared SSH options, and test-qemu-runner.sh runs in the check lane with stand-ins.

Migration

Pass --state-file PATH to both start and stop for cross-shell use; without it, stop cannot find the VM.

^{Written for commit a886dd0. Summary will update on new commits.}

qemu-runner.sh and test-matrix.sh each spelled the ssh option list for the test VM, and the copies had drifted: the runner kept the peer alive for 10 s x 6 and the matrix for 15 s x 4. tests/lib/qemu-ssh.sh now holds the one list. qemu_ssh_opts fills QEMU_SSH_OPTS at call time so a caller can still wrap its ssh in timeout(1), which cannot wrap a shell function. Both lanes keep the 60 s dead-peer budget.

cubic-dev-ai

3 issues found across 5 files

Prompt for AI agents (unresolved issues)


Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="tests/test-qemu-runner.sh">

<violation number="1" location="tests/test-qemu-runner.sh:94">
P2: The `kill -0`-based `alive()` reports a killed-but-unreaped child as "alive": `$own` is a child of this test script, killed by the separate `bash $RUNNER stop` subprocess, so it stays a zombie until the script reaps it and `kill -0` keeps returning success. The assertion `... dead "$(alive "$own")"` therefore always fails (verified: the runner reports `(got alive, want dead) [ FAIL ]` and exits 1), which fails the `test-qemu-runner` lane wired into `make check`. Reap the child before asserting liveness.</violation>

<violation number="2" location="tests/test-qemu-runner.sh:122">
P3: The failed-start case does not verify run-directory cleanup: `qemu_write_state` is never reached, so the state-file check passes regardless of whether `qemu_fail_start` removes the generated directory. Make the stub create or record `-pidfile`, then assert that its run directory is gone.</violation>
</file>

<file name="tests/lib/qemu-ssh.sh">

<violation number="1" location="tests/lib/qemu-ssh.sh:21">
P2: On hosts whose `ssh-agent` contains several identities, `-i` does not restrict authentication to `QEMU_SSH_KEY`; Dropbear can exhaust its authentication attempts before trying the VM key, causing every QEMU SSH command to fail. Add `IdentitiesOnly=yes` to this shared option list.</violation>
</file>

_{Reply with feedback, questions, or to request a fix.

Re-trigger cubic}

cubic-dev-ai · 2026-09-14T16:43:47Z

+        -o ConnectTimeout=10
+        -o ServerAliveInterval=10
+        -o ServerAliveCountMax=6
+        -i "$QEMU_SSH_KEY"


P2: On hosts whose ssh-agent contains several identities, -i does not restrict authentication to QEMU_SSH_KEY; Dropbear can exhaust its authentication attempts before trying the VM key, causing every QEMU SSH command to fail. Add IdentitiesOnly=yes to this shared option list.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At tests/lib/qemu-ssh.sh, line 21: <comment>On hosts whose `ssh-agent` contains several identities, `-i` does not restrict authentication to `QEMU_SSH_KEY`; Dropbear can exhaust its authentication attempts before trying the VM key, causing every QEMU SSH command to fail. Add `IdentitiesOnly=yes` to this shared option list.</comment> <file context> @@ -0,0 +1,24 @@ + -o ConnectTimeout=10 + -o ServerAliveInterval=10 + -o ServerAliveCountMax=6 + -i "$QEMU_SSH_KEY" + -p "$QEMU_PORT" + ) </file context>

cubic-dev-ai · 2026-09-14T16:43:47Z

+rc=0
+bash "$RUNNER" stop --state-file "$state" > /dev/null 2>&1 || rc=$?
+check "stop returns 0 for the run's own process" 0 "$rc"
+check "the run's own process is terminated" dead "$(alive "$own")"


P2: The kill -0-based alive() reports a killed-but-unreaped child as "alive": $own is a child of this test script, killed by the separate bash $RUNNER stop subprocess, so it stays a zombie until the script reaps it and kill -0 keeps returning success. The assertion ... dead "$(alive "$own")" therefore always fails (verified: the runner reports (got alive, want dead) [ FAIL ] and exits 1), which fails the test-qemu-runner lane wired into make check. Reap the child before asserting liveness.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At tests/test-qemu-runner.sh, line 94: <comment>The `kill -0`-based `alive()` reports a killed-but-unreaped child as "alive": `$own` is a child of this test script, killed by the separate `bash $RUNNER stop` subprocess, so it stays a zombie until the script reaps it and `kill -0` keeps returning success. The assertion `... dead "$(alive "$own")"` therefore always fails (verified: the runner reports `(got alive, want dead) [ FAIL ]` and exits 1), which fails the `test-qemu-runner` lane wired into `make check`. Reap the child before asserting liveness.</comment> <file context> @@ -0,0 +1,125 @@ +rc=0 +bash "$RUNNER" stop --state-file "$state" > /dev/null 2>&1 || rc=$? +check "stop returns 0 for the run's own process" 0 "$rc" +check "the run's own process is terminated" dead "$(alive "$own")" +check "stop removes the state file after a kill" gone "$(present "$state")" + </file context>

Suggested change

check "the run's own process is terminated" dead "$(alive "$own")"

wait "$own" 2> /dev/null

check "the run's own process is terminated" dead "$(alive "$own")"

cubic-dev-ai · 2026-09-14T16:43:47Z

+check "the failure names the timeout" yes "$(has "$out" "did not boot")"
+check "the serial console reaches the caller" yes "$(has "$out" SERIAL-MARKER)"
+check "qemu's own output reaches the caller" yes "$(has "$out" STDERR-MARKER)"
+check "a failed start writes no state file" gone "$(present "$state")"


P3: The failed-start case does not verify run-directory cleanup: qemu_write_state is never reached, so the state-file check passes regardless of whether qemu_fail_start removes the generated directory. Make the stub create or record -pidfile, then assert that its run directory is gone.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At tests/test-qemu-runner.sh, line 122: <comment>The failed-start case does not verify run-directory cleanup: `qemu_write_state` is never reached, so the state-file check passes regardless of whether `qemu_fail_start` removes the generated directory. Make the stub create or record `-pidfile`, then assert that its run directory is gone.</comment> <file context> @@ -0,0 +1,125 @@ +check "the failure names the timeout" yes "$(has "$out" "did not boot")" +check "the serial console reaches the caller" yes "$(has "$out" SERIAL-MARKER)" +check "qemu's own output reaches the caller" yes "$(has "$out" STDERR-MARKER)" +check "a failed start writes no state file" gone "$(present "$state")" + +report_summary </file context>

qemu-runner.sh writes a state file so start and stop can run in separate shells, and stop signals the recorded pid only while its argv still names the run's pidfile. The command that starts a VM installs the cleanup trap: the old source-time trap 'qemu_stop' EXIT replaced the cleanup_fixtures; cleanup_qemu trap test-matrix.sh registers earlier, since a bare trap ... EXIT replaces the chain rather than adding to it. test-matrix.sh now marks the VM active before qemu_start, so its own trap covers the boot window the removed trap used to, and clears the mark if the start fails. A start whose guest /tmp cannot be prepared fails, and a failed start prints the serial console and qemu's own output before its run directory goes. tests/test-qemu-runner.sh pins this against stand-ins, so no VM boots: a recycled pid, a process whose argv names the pidfile, and a start that never opens the port.

henrybear327 requested a review from jserv September 14, 2026 16:36

henrybear327 self-assigned this Sep 14, 2026

cubic-dev-ai Bot reviewed Sep 14, 2026

View reviewed changes

jserv reviewed Sep 14, 2026

View reviewed changes

Comment thread tests/test-qemu-runner.sh Outdated

henrybear327 force-pushed the conf/qemu-runner branch from a6ae8ef to a886dd0 Compare September 15, 2026 17:03

Navigation Menu

Uh oh!

Improve QEMU execution - #382

Improve QEMU execution#382
henrybear327 wants to merge 2 commits into
sysprog21:mainfrom
henrybear327:conf/qemu-runner

henrybear327 commented Sep 14, 2026 •

edited by cubic-dev-ai Bot

Loading

Uh oh!

cubic-dev-ai Bot left a comment

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

	check "the run's own process is terminated" dead "$(alive "$own")"
	wait "$own" 2> /dev/null
	check "the run's own process is terminated" dead "$(alive "$own")"

Uh oh!

Conversation

henrybear327 commented Sep 14, 2026 • edited by cubic-dev-ai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary by cubic

Uh oh!

cubic-dev-ai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Choose a reason for hiding this comment

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Choose a reason for hiding this comment

Uh oh!

cubic-dev-ai Bot Sep 14, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

henrybear327 commented Sep 14, 2026 •

edited by cubic-dev-ai Bot

Loading