tests, task: fix load-dependent test failures

configs/sim/gmoccapy/simulated_home.hal

@@ -1,4 +1,3 @@
-loadrt or2 names=or2_0
 loadrt comp names=comp_x,comp_y,comp_z
 
 net Xhomeswpos => comp_x.in0
@@ -17,16 +16,10 @@ setp comp_x.hyst .02
 setp comp_y.hyst .02
 setp comp_z.hyst .02
 
-net Xhomesw <= comp_x.out
+net Xhomesw <= comp_x.out => joint.0.home-sw-in
 net Yhomesw <= comp_y.out => joint.1.home-sw-in
-net Zhomesw <= comp_z.out
-
-net Xhomesw => or2_0.in0
-net Zhomesw => or2_0.in1
-net XZhomesw or2_0.out => joint.0.home-sw-in joint.2.home-sw-in
+net Zhomesw <= comp_z.out => joint.2.home-sw-in
 
 addf comp_x servo-thread
 addf comp_y servo-thread
 addf comp_z servo-thread
-
-addf or2_0 servo-thread

configs/sim/touchy/touchy.ini

@@ -50,7 +50,7 @@ SERVO_PERIOD =               1000000
 
 HALFILE =                    core_sim.hal
 # add a test panel so people can actually test the screen
-HALCMD = loadusr gladevcp --always_above -c touchy_test touchy_test_panel.glade
+HALCMD = loadusr -Wn touchy_test gladevcp --always_above -c touchy_test touchy_test_panel.glade
 POSTGUI_HALFILE = touchy_postgui.hal
 
 [TRAJ]

src/emc/task/taskintf.cc

@@ -1704,7 +1704,9 @@ int emcTrajUpdate(EMC_TRAJ_STAT * stat)
 
     if (emcmotStatus.motionFlag & EMCMOT_MOTION_ERROR_BIT) {
 	stat->status = RCS_STATUS::ERROR;
-    } else if (stat->inpos && (stat->queue == 0)) {
+    } else if (stat->inpos && (stat->queue == 0) && !emcmotStatus.probing) {
+	// a probe move that ends without contact is reported a servo
+	// cycle after the planner goes idle, so wait for the verdict
 	stat->status = RCS_STATUS::DONE;
     } else {
 	stat->status = RCS_STATUS::EXEC;

tests/abort/stop-button-crazy-move/test-ui.py

@@ -10,7 +10,7 @@
 
 # Time increment and timeout, seconds
 TIME_INCR = 0.1
-TIMEOUT = 10.0
+TIMEOUT = 60.0
 
 #
 # connect to LinuxCNC

tests/hal-show/checkresult

@@ -0,0 +1,12 @@
+#!/bin/bash
+set -e
+
+TESTDIR=$(dirname "$1")
+cd "${TESTDIR}"
+
+# The owner column is the component ID, which depends on the order in which
+# halcmd and rtapi_app take IDs during each loadrt. That order changes with
+# CPU load, so mask it.
+sed --in-place --regexp-extended --expression='s/^ *[0-9]+( +[a-z0-9]+ +(IN|OUT|I\/O) .*)$/(ignored comp id)\1/' result
+
+diff -u expected result

tests/hal-show/expected

@@ -1,11 +1,11 @@
 Component Pins:
 Owner   Type  Dir                 Value  Name
-    10  bool  IN                  FALSE  conv-bool-uint.0.in <== net-conv-bool-uint.0.in
-    16  real  IN                      0  conv-real-sint.0.in <== net-conv-real-sint.0.in
-    19  sint  IN                      0  conv-sint-real.0.in <== net-conv-sint-real.0.in
-    22  sint  IN                      0  conv-sint-uint.0.in <== net-conv-sint-uint.0.in
-    13  uint  IN     0x0000000000000000  conv-uint-bool.0.in <== net-conv-uint-bool.0.in
-    25  uint  IN     0x0000000000000000  conv-uint-sint.0.in <== net-conv-uint-sint.0.in
+(ignored comp id)  bool  IN                  FALSE  conv-bool-uint.0.in <== net-conv-bool-uint.0.in
+(ignored comp id)  real  IN                      0  conv-real-sint.0.in <== net-conv-real-sint.0.in
+(ignored comp id)  sint  IN                      0  conv-sint-real.0.in <== net-conv-sint-real.0.in
+(ignored comp id)  sint  IN                      0  conv-sint-uint.0.in <== net-conv-sint-uint.0.in
+(ignored comp id)  uint  IN     0x0000000000000000  conv-uint-bool.0.in <== net-conv-uint-bool.0.in
+(ignored comp id)  uint  IN     0x0000000000000000  conv-uint-sint.0.in <== net-conv-uint-sint.0.in
 
 Signals:
 Type                  Value  Name     (linked to)
@@ -36,12 +36,12 @@ sint                      0  net-conv-uint-sint.0.out
 
 Component Pins:
 Owner   Type  Dir                 Value  Name
-    10  bool  IN                   TRUE  conv-bool-uint.0.in <== net-conv-bool-uint.0.in
-    16  real  IN           2.147484e+09  conv-real-sint.0.in <== net-conv-real-sint.0.in
-    19  sint  IN            -2147483648  conv-sint-real.0.in <== net-conv-sint-real.0.in
-    22  sint  IN   -9223372036854775808  conv-sint-uint.0.in <== net-conv-sint-uint.0.in
-    13  uint  IN     0x00000000FFFFFFFF  conv-uint-bool.0.in <== net-conv-uint-bool.0.in
-    25  uint  IN     0xFFFFFFFFFFFFFFFF  conv-uint-sint.0.in <== net-conv-uint-sint.0.in
+(ignored comp id)  bool  IN                   TRUE  conv-bool-uint.0.in <== net-conv-bool-uint.0.in
+(ignored comp id)  real  IN           2.147484e+09  conv-real-sint.0.in <== net-conv-real-sint.0.in
+(ignored comp id)  sint  IN            -2147483648  conv-sint-real.0.in <== net-conv-sint-real.0.in
+(ignored comp id)  sint  IN   -9223372036854775808  conv-sint-uint.0.in <== net-conv-sint-uint.0.in
+(ignored comp id)  uint  IN     0x00000000FFFFFFFF  conv-uint-bool.0.in <== net-conv-uint-bool.0.in
+(ignored comp id)  uint  IN     0xFFFFFFFFFFFFFFFF  conv-uint-sint.0.in <== net-conv-uint-sint.0.in
 
 Signals:
 Type                  Value  Name     (linked to)

tests/halmodule/stream/stream_reader.py

@@ -10,8 +10,9 @@
     assert reader.sampleno == i+1
 assert reader.read() is None
 assert reader.num_underruns == 1
+# report before ready(): halrun ends the run once it sees ready
+print("pass", flush=True)
 c.ready()
-print("pass")
 
 try:
     while 1: time.sleep(1)

tests/interp/m98m99/12-M99-endless-main-program/expected.motion-logger

@@ -176,6 +176,7 @@ SPINDLE_OFF
 DISABLE
 JOINT_UNHOME joint=-2
 FREE
+SET_TERM_COND termCond=2, tolerance=0.001
 JOG_ABORT joint=0
 JOG_ABORT joint=1
 JOG_ABORT joint=2

tests/interp/m98m99/12-M99-endless-main-program/test-ui.py

@@ -16,6 +16,22 @@
 e = linuxcnc.error_channel()
 
 
+#
+# Task queues the canon commands of interpreter init before it reads its
+# first command. Let it issue them, or an early ESTOP_RESET wipes them
+# and the log depends on how fast this script starts.
+#
+
+deadline = time.monotonic() + 60
+while True:
+    s.poll()
+    if s.state == linuxcnc.RCS_DONE: break
+    if time.monotonic() > deadline:
+        print("task did not finish startup")
+        sys.exit(1)
+    time.sleep(0.01)
+
+
 #
 # Come out of E-stop, turn the machine on, home, and switch to Auto mode.
 #

tests/motion-logger/basic/expected.builtin-startup.in

@@ -176,6 +176,7 @@ SPINDLE_OFF
 DISABLE
 JOINT_UNHOME joint=-2
 FREE
+SET_TERM_COND termCond=2, tolerance=0.001
 JOG_ABORT joint=0
 JOG_ABORT joint=1
 JOG_ABORT joint=2

tests/motion-logger/basic/test-ui.py

@@ -46,6 +46,22 @@ def end_log(logfile_name):
 e = linuxcnc.error_channel()
 
 
+#
+# Task queues the canon commands of interpreter init before it reads its
+# first command. Let it issue them, or an early ESTOP_RESET wipes them
+# and the log depends on how fast this script starts.
+#
+
+deadline = time.monotonic() + 60
+while True:
+    s.poll()
+    if s.state == linuxcnc.RCS_DONE: break
+    if time.monotonic() > deadline:
+        print("task did not finish startup")
+        sys.exit(1)
+    time.sleep(0.01)
+
+
 #
 # Come out of E-stop, turn the machine on, and switch to Auto mode.
 #

tests/motion-logger/mountaindew/expected.motion-logger

@@ -80,6 +80,7 @@ SPINDLE_OFF
 DISABLE
 JOINT_UNHOME joint=-2
 FREE
+SET_TERM_COND termCond=2, tolerance=0.001
 JOG_ABORT joint=0
 JOG_ABORT joint=1
 JOG_ABORT joint=2

tests/motion-logger/mountaindew/test-ui.py

@@ -16,6 +16,22 @@
 e = linuxcnc.error_channel()
 
 
+#
+# Task queues the canon commands of interpreter init before it reads its
+# first command. Let it issue them, or an early ESTOP_RESET wipes them
+# and the log depends on how fast this script starts.
+#
+
+deadline = time.monotonic() + 60
+while True:
+    s.poll()
+    if s.state == linuxcnc.RCS_DONE: break
+    if time.monotonic() > deadline:
+        print("task did not finish startup")
+        sys.exit(1)
+    time.sleep(0.01)
+
+
 #
 # Come out of E-stop, turn the machine on, home, and switch to Auto mode.
 #

tests/motion-logger/startup-gcode-abort/expected.motion-logger.in

@@ -114,12 +114,15 @@ JOINT_UNHOME joint=-2
 FREE
 
 #
-# Task is now got the 'motion.in-position' pin visible, and is
-# accepting abort calls from the UI.
+# Task now issues what interp_list holds: the interpreter's init and the
+# first move of the startup code, then waits in its dwell.
 #
 
+SET_TERM_COND termCond=2, tolerance=0.001
+SET_LINE x=-1, y=-2, z=-3, a=0, b=0, c=0, u=0, v=0, w=0, id=2, motion_type=1, vel=4.98888, ini_maxvel=4.98888, acc=1247.22, turn=-1
+
 #
-# The test UI immediately sent an Abort.
+# The test UI sent an Abort during the dwell.
 #
 
 JOG_ABORT joint=0

tests/motion-logger/startup-gcode-abort/test-ui.py

@@ -31,6 +31,15 @@
     time.sleep(0.1)
     waitlimit -= 1
 
+# Abort in the dwell of the startup code, after its first move went out.
+waitlimit = 600
+while 0 < waitlimit:
+    s.poll()
+    if s.exec_state == linuxcnc.EXEC_WAITING_FOR_DELAY:
+        break
+    time.sleep(0.1)
+    waitlimit -= 1
+
 print("UI abort")
 sys.stdout.flush()
 

tests/multiclick/test.hal

@@ -2,8 +2,8 @@
 loadrt threads name1=millisecond period1=1000000
 
 loadrt multiclick count=2
-loadrt sampler depth=100 cfg=bbbbbbbbbsbbbbbbbbbs
-loadrt streamer depth=100 cfg=bb
+loadrt sampler depth=128 cfg=bbbbbbbbbsbbbbbbbbbs
+loadrt streamer depth=128 cfg=bb
 
 
 net in0 streamer.0.pin.0 => multiclick.0.in sampler.0.pin.0
@@ -50,9 +50,9 @@ setp multiclick.1.output-hold-ns 1500000
 setp multiclick.1.invert-input   True
 
 
-# run halstreamer and let it pre-load the fifo
-loadusr halstreamer input-signals
-loadusr -w sleep 0.5
+# pre-load the whole input before the thread starts; halstreamer exits
+# once the last line is in the fifo
+loadusr -w halstreamer input-signals
 
 start
 loadusr -w halsampler -n 105

tests/mux/test.hal

@@ -1,8 +1,8 @@
 setexact_for_test_suite_only
 loadrt threads name1=slow period1=1000000
 loadrt mux_generic config="b2b,b2f,b2s,b2u,f2b,f2f,f2s,f2u,s2b,s2f,s2s,s2u,u2b,u2f,u2s,u2u,b2b,f5f"
-loadrt streamer depth=100 cfg=bu
-loadrt sampler depth=100 cfg=bfsubfsubfsubfsubfbu
+loadrt streamer depth=128 cfg=bu
+loadrt sampler depth=128 cfg=bfsubfsubfsubfsubfbu
 
 addf streamer.0 slow
 addf mux-gen.00 slow
@@ -80,8 +80,9 @@ net o15 mux-gen.15.out-uint sampler.0.pin.15
 net o16 mux-gen.16.out-bool sampler.0.pin.16
 net o17 mux-gen.17.out-real sampler.0.pin.17
 
-loadusr halstreamer input-signals
-loadusr -w sleep 0.5
+# pre-load the whole input before the thread starts; halstreamer exits
+# once the last line is in the fifo
+loadusr -w halstreamer input-signals
 
 start
 loadusr -w halsampler -n 105

tests/remap/remap-io/test-ui.py

@@ -76,14 +76,17 @@ def do_in(inp, d_in=True, wait_mode=None, timeout=None):
         timeout = (' Q%d' % timeout if timeout is not None else ''),
         )
 
+    # Read the pre-test values before setting the input pin; motion copies
+    # the pin into the status within a servo cycle
+    s.poll() # Update status channel
+
     # Set input pin
     if d_in:
         h['d_in'] = inp
     else:
         h['a_in'] = inp
 
     # Print pre-test values
-    s.poll() # Update status channel
     c.mdi('(print,cmd: "%s"; input: %.4f)' % (cmd, inp))
     c.mdi('(print,    M66 remapping pre:  5399=#5399; 100=#100; '
           'ain0=%.2f; din0=%d)' % (s.ain[0], s.din[0]))

tests/ui-smoke/_lib/drive.py

@@ -158,9 +158,15 @@ def home_all(cmd, stat, timeout):
     Ensures ESTOP_RESET + ON + MANUAL and retries the whole sequence
     (see HOME_RETRY_BUDGET)."""
     attempt_timeout = min(timeout, HOME_ATTEMPT_TIMEOUT_S)
+    deadline = time.monotonic() + timeout
     for attempt in range(1, HOME_RETRY_BUDGET + 1):
-        if not ensure_state(cmd, stat, linuxcnc.STATE_ESTOP_RESET,
-                            "STATE_ESTOP_RESET"):
+        # Task reports ESTOP_RESET only while the machine is off, and an
+        # ESTOP_RESET command does not turn it off, so requesting it on a
+        # retry after the machine came on would never be met.
+        stat.poll()
+        if (stat.task_state != linuxcnc.STATE_ON and
+                not ensure_state(cmd, stat, linuxcnc.STATE_ESTOP_RESET,
+                                 "STATE_ESTOP_RESET")):
             return False
         if not ensure_state(cmd, stat, linuxcnc.STATE_ON, "STATE_ON"):
             return False
@@ -171,10 +177,19 @@ def home_all(cmd, stat, timeout):
         stat.poll()
         njoints = stat.joints
         cmd.home(-1)
-        if wait_until_quiet(
+        all_homed = lambda s: all(s.homed[i] for i in range(njoints))
+        done = wait_until_quiet(stat, all_homed, attempt_timeout)
+        # On a slow machine homing can still be running here. Homing
+        # again would restart joints whose home switch another, already
+        # homed joint now holds active (the gmoccapy sim shares one switch
+        # between X and Z), so let a running sequence finish instead.
+        if not done and any(stat.joint[i]["homing"] for i in range(njoints)):
+            done = wait_until_quiet(
                 stat,
-                lambda s: all(s.homed[i] for i in range(njoints)),
-                attempt_timeout):
+                lambda s: all_homed(s) or not any(
+                    s.joint[i]["homing"] for i in range(njoints)),
+                max(0.0, deadline - time.monotonic())) and all_homed(stat)
+        if done:
             cmd.teleop_enable(1)
             cmd.wait_complete()
             return True

tests/ui-smoke/_lib/window-fit.sh

@@ -10,8 +10,9 @@
 #
 # window_fit_check <wm-class-pattern>
 #   <wm-class-pattern> is matched literally against the xwininfo tree
-#   line, e.g. '("touchy" "Touchy")'. The first matching window is checked
-#   against the root window (the screen). Needs xwininfo (x11-utils).
+#   line, e.g. '("touchy" "Touchy")'. The first viewable matching window
+#   is checked against the root window (the screen). Needs xwininfo
+#   (x11-utils).
 
 window_fit_check() {
     pattern="$1"
@@ -22,11 +23,30 @@ window_fit_check() {
     root=$(xwininfo -root 2>/dev/null)
     screen_w=$(echo "$root" | awk '/Width:/{print $2; exit}')
     screen_h=$(echo "$root" | awk '/Height:/{print $2; exit}')
-    geom=$(xwininfo -root -tree 2>/dev/null | grep -F "$pattern" \
-        | grep -oE '[0-9]+x[0-9]+\+[0-9]+\+[0-9]+' | head -1)
-    win_w=${geom%%x*}
-    rest=${geom#*x}
-    win_h=${rest%%+*}
+    # drive.py returns once task is up, which under load can be before
+    # the GUI maps its window. The tree also lists unmapped windows, at
+    # the size they were created with, so measure only a viewable one,
+    # and poll until its size reads the same twice in a row.
+    win_w=
+    win_h=
+    prev=
+    for _ in $(seq "${UI_SMOKE_FIT_TIMEOUT:-60}"); do
+        size=
+        for w in $(xwininfo -root -tree 2>/dev/null | grep -F "$pattern" \
+                | grep -oE '^ +0x[0-9a-f]+'); do
+            info=$(xwininfo -id "$w" 2>/dev/null)
+            echo "$info" | grep -q "Map State: IsViewable" || continue
+            size=$(echo "$info" | awk '/^ *Width:/{w=$2} /^ *Height:/{h=$2} END{print w "x" h}')
+            break
+        done
+        if [ -n "$size" ] && [ "$size" = "$prev" ]; then
+            win_w=${size%x*}
+            win_h=${size#*x}
+            break
+        fi
+        prev=$size
+        sleep 1
+    done
     if [ -z "$win_w" ] || [ -z "$win_h" ]; then
         echo "UI_SMOKE_FAIL: window-fit: no window matching $pattern found"
         return 1