tests: cover the channel open, close and req callbacks

[ejohnstown]

The channel open confirm and fail hooks, the close hook, and the exec and
subsystem request hooks had no caller in tests/, examples/ or apps/, so
each shipped unexercised.

• Cover the confirm and fail callbacks: the peer's channel id, window and
  packet size are recorded, the channel is still findable in the fail
  arm, and a rejecting hook fails the receive.
• Cover the close callback running before the channel is retired, and a
  rejecting one changing nothing.
• Cover the exec and subsystem callbacks seeing the session type and the
  command string, and their answer drawing CHANNEL_SUCCESS or
  CHANNEL_FAILURE.

[Copilot]

🟡 Changes recommended

The new tests’ comments/intent imply validation of callback-observed state ordering and channel discoverability during callbacks, but the current implementation does not actually assert those properties.

Once you've addressed the issues Copilot identified, you can request another Copilot review.

Pull request overview

This PR expands tests/regress.c to exercise previously untested channel lifecycle and request callbacks in the wolfSSH server harness, aiming to ensure these hooks are invoked and their observable effects are validated.

Changes:

• Adds regression tests for SSH_MSG_CHANNEL_OPEN_CONFIRMATION / SSH_MSG_CHANNEL_OPEN_FAILURE open-response callbacks.
• Adds regression tests for the channel close callback behavior during peer close handling.
• Adds regression tests for session "exec" and "subsystem" request callbacks and their resulting CHANNEL_SUCCESS / CHANNEL_FAILURE replies.

File summaries

File	Description
tests/regress.c	Adds new regression harness helpers and test cases covering channel open-response, close, and request callbacks.

Review details

Suppressed comments (3)

tests/regress.c:2979

• The open-confirm test comment/PR description implies the peer parameters are already populated when the confirm callback runs, but the callback currently doesn't record them. Likewise, the open-fail path is supposed to invoke the callback while the channel is still on the session's list, but that is not captured. Record peer params and channelListSz inside the callbacks so the tests can assert what the callback could actually observe.

static int RecordingChannelOpenConfCb(WOLFSSH_CHANNEL* channel, void* ctx)
{
    AssertNotNull(channel);
    openConfCbCalls++;
    openRespCbChannel = channel->channel;

tests/regress.c:3059

• To validate that the peer parameters are populated before the open-confirm callback runs (as the comment indicates), assert against the values captured inside the callback rather than only checking the final channel fields after DoReceive().

    /* The peer's numbers are in place before the callback can use them. */
    AssertIntEQ(channel->peerChannel, 7);
    AssertIntEQ(channel->peerWindowSz, 0x4000);
    AssertIntEQ(channel->peerMaxPacketSz, 0x8000);

tests/regress.c:3094

• The open-failure test is meant to ensure the failure callback runs while the channel is still on the session list (so it remains discoverable during the callback). Add an assertion for the list size captured inside the callback.

    AssertIntEQ(openFailCbCalls, 1);
    AssertIntEQ(openConfCbCalls, 0);
    AssertIntEQ(openRespCbChannel, selfChannelId);
    AssertTrue(openRespCbCtx == &cbCtx);
    AssertIntEQ(harness.ssh->channelListSz, 0);

• Files reviewed: 1/1 changed files
• Comments generated: 2
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[wolfSSL-Fenrir-bot]

Fenrir Automated Review — PR #1227

Scan targets checked: wolfssh-bugs, wolfssh-src

Fenrir result: Approved ✅

No new issues found in the changed files.

Advisory only — this automated result does not count as a GitHub approval.