Bug 25733: Avoid assert failure if all circuits time out.

Prior to #23100, we were not counting HS circuit build times in our
calculation of the timeout. This could lead to a condition where our timeout
was set too low, based on non HS circuit build times, and then we would
abandon all HS circuits, storing no valid timeouts in the histogram.

This commit avoids the assert.

changes/bug25733

@@ -0,0 +1,4 @@
+  o Minor bugfixes (Assert crash):
+    - Avoid an assert in the circuit build timeout code if we fail to
+      allow any circuits to actually complete. Fixes bug 25733;
+      bugfix on 0.2.2.2-alpha.

src/or/circuitstats.c

@@ -767,11 +767,23 @@ circuit_build_times_get_xm(circuit_build_times_t *cbt)
              histogram[nth_max_bin[n]]);
   }
 
-  /* The following assert is safe, because we don't get called when we
-   * haven't observed at least CBT_MIN_MIN_CIRCUITS_TO_OBSERVE circuits. */
+  /* bin_counts can become zero if all of our last CBT_NCIRCUITS_TO_OBSERVE
+   * circuits were abandoned before they completed. This shouldn't happen,
+   * though. We should have reset/re-learned a lower timeout first. */
+  if (bin_counts == 0) {
+    ret = 0;
+    log_warn(LD_CIRC,
+               "No valid circuit build time data out of %d times, %u modes, "
+               "have_timeout=%d, %lfms", cbt->total_build_times, num_modes,
+               cbt->have_computed_timeout, cbt->timeout_ms);
+    goto done;
+  }
+
   tor_assert(bin_counts > 0);
 
   ret /= bin_counts;
+
+ done:
   tor_free(histogram);
   tor_free(nth_max_bin);
 
@@ -1057,6 +1069,10 @@ circuit_build_times_update_alpha(circuit_build_times_t *cbt)
    * and less frechet-like. */
   cbt->Xm = circuit_build_times_get_xm(cbt);
 
+  /* If Xm came back 0, then too many circuits were abandoned. */
+  if (cbt->Xm == 0)
+    return 0;
+
   tor_assert(cbt->Xm > 0);
 
   for (i=0; i< CBT_NCIRCUITS_TO_OBSERVE; i++) {