RC1 20.0.0.7 MAC spindumps

[rajkej]

I have had issues with SE 19 getting the spinning wheel for 2-3 months now. We never were able to track it down even though I reported issues. Now I'm getting spindumps from V20 RC1. I'm not sure what might cause this but wanted to report it in case there's something in the spin file that stands out for you.

from system.log:

Oct 26 12:10:02 iMac.local WindowServer[169]: _CGXRemoveWindowFromWindowMovementGroup: Window not in group
Oct 26 12:10:03 iMac.local vs[806]: modalSession has been exited prematurely - check for a reentrant call to endModalSession:
Oct 26 12:10:20 iMac.local spindump[457]: Saved spin report for vs version 20.0.0 (20.0.0.7) to /Library/Logs/DiagnosticReports/vs_2015-10-26-121020_iMac.spin
Oct 26 12:10:40 iMac.local spindump[457]: Saved spin report for vs version 20.0.0 (20.0.0.7) to /Library/Logs/DiagnosticReports/vs_2015-10-26-121040_iMac.spin

from first spin file
Date/Time:       2015-10-26 12:10:10 -0600
OS Version:      10.10.5 (Build 14F1021)
Architecture:    x86_64
Report Version:  21

Command:         SlickEdit Pro 2015
Path:            /Applications/SlickEditPro2015.app/Contents/MacOS/vs
Version:         20.0.0 (20.0.0.7)
PID:             806

Event:           spin-lite
Duration:        9.10s (process was unresponsive for 2 seconds before sampling)
Steps:           92 (100ms sampling interval)

Hardware model:  iMac11,3
Active cpus:     4

Fan speed:       997 rpm

Heaviest stack for the main thread of the target process:
  92  start + 52 (vs + 22708) [0x1000058b4]
  92  main + 9 (vs + 318649) [0x10004dcb9]
  92  xmain + 42 (vs + 16425130) [0x100faa0aa]
  92  vmain(int, char**) + 3336 (vs + 204984) [0x1000320b8]
  92  QCoreApplication::exec() + 188 (QtCore + 1327020) [0x1027defac]
  92  QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) + 324 (QtCore + 1316660) [0x1027dc734]
  92  QEventLoop::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) + 68 (QtCore + 1315716) [0x1027dc384]
  92  QDesktopWidget::resizeEvent(QResizeEvent*) + 14167 (QtGui + 276839) [0x102a2a967]
  92  -[NSApplication run] + 711 (AppKit + 551784) [0x7fff8ee3bb68]
  92  -[SlickEditApplication sendEvent:] + 181 (vs + 19276581) [0x101262325]
  92  -[NSApplication sendEvent:] + 3537 (AppKit + 1431067) [0x7fff8ef1261b]
  92  QMacInputContext::reset() + 24932 (QtGui + 209860) [0x102a1a3c4]
  92  QMacCocoaAutoReleasePool::~QMacCocoaAutoReleasePool() + 18795 (QtGui + 255307) [0x102a2554b]
  92  QApplicationPrivate::sendMouseEvent(QWidget*, QMouseEvent*, QWidget*, QWidget*, QWidget**, QPointer<QWidget>&, bool) + 333 (QtGui + 602381) [0x102a7a10d]
  92  QCoreApplication::notifyInternal(QObject*, QEvent*) + 124 (QtCore + 1320044) [0x1027dd46c]
  92  QApplication::notify(QObject*, QEvent*) + 8752 (QtGui + 597376) [0x102a78d80]
  92  QApplicationPrivate::notify_helper(QObject*, QEvent*) + 142 (QtGui + 571822) [0x102a729ae]
  92  QCoreApplicationPrivate::sendThroughObjectEventFilters(QObject*, QEvent*) + 151 (QtCore + 1318679) [0x1027dcf17]
  92  QAbstractScrollArea::qt_metacall(QMetaObject::Call, int, void**) + 336 (QtGui + 5323120) [0x102efa970]
  92  QAbstractScrollArea::viewportEvent(QEvent*) + 23 (QtGui + 5314023) [0x102ef85e7]
  92  QFrame::event(QEvent*) + 44 (QtGui + 4738396) [0x102e6bd5c]
  92  QWidget::event(QEvent*) + 2479 (QtGui + 948207) [0x102ace7ef]
  92  vsCtlEditor::mouseMoveEvent(QMouseEvent*) + 524 (vs + 1220012) [0x100129dac]
  92  vsQTSlickCExecuteMouseMoveEvent + 229 (vs + 1412341) [0x100158cf5]
  92  vsCtlSlickCExecuteEvent + 833 (vs + 1409857) [0x100158341]
  92  gui_edit_event_message + 182 (vs + 2039222) [0x1001f1db6]
  92  vs_call_key_new(filewindow_t*, int, int) + 805 (vs + 2765525) [0x1002a32d5]
  92  call_key_common(filewindow_t*, int, int, unsigned int, unsigned int, bool, unsigned char) + 481 (vs + 2768481) [0x1002a3e61]
  91  run_proc(int) + 481 (vs + 590161) [0x100090151]
  46  call_dllpc(int, int, namelist_t*) + 1551 (vs + 580911) [0x10008dd2f]
  45  filewindow_t::updateTextChangeCallbacks() + 115 (vs + 2543283) [0x10026ceb3]
  25  vs_clex_comments2(filewindow_t*) + 3404 (vs + 2541276) [0x10026c6dc]
  2   clex_continue_comment(char*, int, unsigned long long, clexembedded_data_t*) + 7824 (vs + 1565968) [0x10017e510] (running)

  Binary Images:
         0x100000000 -        0x101977fff  com.slickedit.SlickEdit 20.0.0 (20.0.0.7) <ED73EE39-CEBD-3BF6-A252-22A0CB95011E>  /Applications/SlickEditPro2015.app/Contents/MacOS/vs
         0x10269b000 -        0x102975fff  QtCore 4.8 (4.                          <1587CD48-DAD6-622D-3DCA-AF93CD2D02FC>  /Applications/SlickEditPro2015.app/Contents/Frameworks/QtCore.framework/Versions/4/QtCore
         0x1029e7000 -        0x103418fff  QtGui 4.8 (4.                           <CC6F8CFF-D8A3-0D1F-3A23-CDAF1BC60AA4>  /Applications/SlickEditPro2015.app/Contents/Frameworks/QtGui.framework/Versions/4/QtGui
      0x7fff8edb5000 -     0x7fff8f936fff  com.apple.AppKit 6.9 (1348.17)            <E485D56D-3E72-34B7-99BB-BFDEE2D07BF5>  /System/Library/Frameworks/AppKit.framework/Versions/C/AppKit

The second spin file has two more stack entries. Here's the last 4.
  24  vs_clex_comments2(filewindow_t*) + 3404 (vs + 2541276) [0x10026c6dc]
  4   clex_continue_comment(char*, int, unsigned long long, clexembedded_data_t*) + 3486 (vs + 1561630) [0x10017d41e]
  1   zstrieq(char const*, char const*) + 21 (vs + 16497989) [0x100fbbd45]
  1   strlen + 39 (libsystem_c.dylib + 4039) [0x7fff930f7fc7] (running)

[Dennis]

vsclex_continue_comment() is comment color coding related, but it could also be related to embedded code.  This will be nearly impossible to replicate without a test case (source file).

[rajkej]

Well... This latest set of spins is when I'm using .cu (cuda) source files that have a mixture of c++ and cuda specific constructs. Cuda here is the Nvidia GPU c/c++ extension for GPU parallel programming.

Extra routines are declared in 'cuda' format that looks a lot like C but with special declarations. Then the routines are called using a format as the following:

static __global__ void xxx( ... )
xxx<<<grids, threads>>>( ... )

The xxx<<<>>> is how you 'call' a cuda function on the GPU and is translated by a special compiler into the appropriate driver calls/etc.

I can definitely say that the issues I had in V19 weren't cuda as I wasn't working on cuda code at the time. There might have been cuda code within the tag file though.

[Dennis]

Could you obfuscate one of the actual source files that exhibit this problem and post it or send it into support?

[rajkej]

I can send it to support. Can you tell me what to put on subject line/who to send to?

[Dennis]

A reference to RC1, this thread, and ask them to forward the file to me.

[Dennis]

Thanks for the good test case.  I've reproduced the problem.  I'm glad you have a good workaround in the meantime.

[Dennis]

Fixed for next release candidate.

[rajkej]

Glad you found something. That'll make using SE a lot nicer!

[Dennis]

I had actually replied to the wrong thread originally, but I think I have also tracked down the problem you were running into as well and am putting a fix in for the next build.

You could help me confirm the problem by using Slick-C profiling (Macro > Slick-C Profiling > ...), then replicating the delay and then posting the results.

Also, do you have the Boost libraries tagged?  That was the source of the specific delays I was able to replicate.

[rajkej]

I do not have boost libraries tagged. I do have linux header files tagged.