#271 Assert 250 upon semaphone wait timeout

[Anonymous]

In qxk_sema.cpp, line 153,

// did the blocking time-out? (signal of zero means that it did)
if (curr->m_timeEvt.sig != static_cast<qsignal>(0)) {</qsignal>

The code does NOT agree with the comment. Checking how the m_timeEvt.sig is used in QP, it looks like the comment is correct. For example, there is a similar code in qxk_mutex.cpp and it looks like it's correct there.

This issue was found when running 6.5.1. The source code of the latest 6.8.1 indicates that it is still an issue.

[Quantum Leaps]

Thank you for reporting. The bug is the reversed condition tested in qxk_sema.cpp:153. Here is the patch to fix this:

@@ -153,5 +153,5 @@ bool QXSemaphore::wait(std::uint_fast16_t const nTicks) noexcept {

         // did the blocking time-out? (signal of zero means that it did)

-        if (curr->m_timeEvt.sig != 0U) {
+        if (curr->m_timeEvt.sig == 0U) {
             if (m_waitSet.hasElement(p)) { // still waiting?
                 m_waitSet.rmove(p); // remove the unblocked thread

The bug has been fixed in the QP/C++ git repositories (on GitHub and SourceForge). The QP-bundle will be fixed in the next release.

NOTE: this bug is NOT present in QP/C, where the corresponding code is correct.

--MMS

[Quantum Leaps]

Fixed in QP/C++ 6.8.2.
--MMS