Linux Pcbnew Seg Faults When Creating a Via Over Smaller Pad

[Joseph Chen]

Linux Pcbnew seg faults almost every time when it is creating a Via over a pad of smaller size.

Here is the version info:

Application: kicad
Version: (2015-10-06 BZR 6248, Git 20a3e23)-product release build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1002,GCC 4.9.2,wx containers,compatible with 2.8)
Platform: Linux 3.19.0-30-generic x86_64, 64 bit, Little endian, wxGTK
Boost version: 1.59.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=ON
         KICAD_SCRIPTING_MODULES=ON
         KICAD_SCRIPTING_WXPYTHON=ON
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

Here is the backtrace of the core dump:

jchen@thermaltake:~ (kicad-local)$ kicad
Segmentation fault (core dumped)
jchen@thermaltake:~ (kicad-local)$ gdb kicad
GNU gdb (Ubuntu 7.9-1ubuntu1) 7.9
Copyright (C) 2015 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-linux-gnu".
Type "show configuration" for configuration details.
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Find the GDB manual and other documentation resources online at:
<http://www.gnu.org/software/gdb/documentation/>.
For help, type "help".
Type "apropos word" to search for commands related to "word"...
Reading symbols from kicad...(no debugging symbols found)...done.
(gdb) core core
warning: exec file is newer than core file.
[New LWP 20210]
[New LWP 20241]
[New LWP 20239]
[New LWP 20211]
[New LWP 20237]
[New LWP 20244]
[New LWP 20213]
[New LWP 20238]
[New LWP 20245]
[New LWP 20212]
[New LWP 20242]
[New LWP 20240]
[New LWP 20243]
[New LWP 20246]
[New LWP 20247]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
bCore was generated by `kicad'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 0x00007f1f0ef12ef3 in PNS_LINE_PLACER::makeVia(VECTOR2<int> const&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
(gdb) bt
#0 0x00007f1f0ef12ef3 in PNS_LINE_PLACER::makeVia(VECTOR2<int> const&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#1 0x00007f1f0ef19dcb in PNS_LINE_PLACER::rhShoveOnly(VECTOR2<int> const&, PNS_LINE&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#2 0x00007f1f0ef1ac06 in PNS_LINE_PLACER::routeStep(VECTOR2<int> const&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#3 0x00007f1f0ef1ad01 in PNS_LINE_PLACER::route(VECTOR2<int> const&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#4 0x00007f1f0ef1af93 in PNS_LINE_PLACER::Move(VECTOR2<int> const&, PNS_ITEM*) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#5 0x00007f1f0ef1275b in PNS_LINE_PLACER::ToggleVia(bool) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#6 0x00007f1f0eeb5d05 in ROUTER_TOOL::onViaCommand(VIATYPE_T) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#7 0x00007f1f0eebaa4d in ROUTER_TOOL::performRouting() () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#8 0x00007f1f0eebb180 in ROUTER_TOOL::mainLoop(PNS_ROUTER_MODE) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#9 0x00007f1f0eebbc38 in ROUTER_TOOL::RouteSingleTrace(TOOL_EVENT const&) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#10 0x00007f1f0f00baf1 in COROUTINE<int, TOOL_EVENT const&>::callerStub(long) () from /home/jchen/kicad-local-installed/usr/local/bin/_pcbnew.kiface
#11 0x00007f1f0fbe9211 in make_fcontext () from /usr/local/lib/libboost_context.so.1.59.0
#12 0x0000000000000000 in ?? ()
(gdb)

[Joseph Chen (joseph-chen59)]

BTW, Pcbnew is in OpenGL view when it seg faults.

[Tomasz Wlostowski (twlostow)]

Hi Joseph,

I'm unable to reproduce the issue here. Could you send us the board (a stripped-down version) which shows the problem?

Cheers,
Tom

[Joseph Chen (joseph-chen59)]

On 10/07/2015 02:24 AM, Tomasz Wlostowski wrote:
> Hi Joseph,
>
> I'm unable to reproduce the issue here. Could you send us the board (a
> stripped-down version) which shows the problem?
>
> Cheers,
> Tom
>

Hi Tom,

Attached you can find the tarball of very simple pcb file that I am able
to reproduce the seg fault 100%.

Here the steps I did:

1. Open kicad-via.pro
2. Click "pcbnew" within "kicad" so you get in the pcbnew.
3. Within pcbnew, Press "F11" key to switch to OpenGL canvas
4. Press "x" key to start routing tracks
5. Click on the pad "GND" of Pin 2 of footprint C1.
6. Press "v" key for creating a via.

Now Pcbnew crashes, everything disappears, because of seg fault.

FYI, the pcb file's design rules include a bigger via for GND. This may
be a factor.

Hope this helps,

--Joe

[firewalker (firew4lker)]

I can not reproduce it with 6254.

Application: kicad
Version: (2015-10-08 BZR 6254)-product release build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1009,GCC 5.2.0,wx containers,compatible with 2.8)
Platform: Linux 4.1.6-1-ARCH i686, 32 bit, Little endian, wxGTK
Boost version: 1.54.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=ON
         KICAD_SCRIPTING_MODULES=ON
         KICAD_SCRIPTING_WXPYTHON=ON
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

[Joseph Chen (joseph-chen59)]

Hi Firewalker, have you used the test board file that I attended ? Using that file enabled me to reproduce the segment fault 100%time.

------------------------------Sent from a mobile device, please excuse my brevity.

-------- Original message --------
From: firewalker <email address hidden>
Date: 10/08/2015 7:39 AM (GMT-07:00)
To: <email address hidden>
Subject: [Bug 1503571] Re: Linux Pcbnew Seg Faults When Creating a Via Over
  Smaller Pad

I can not reproduce it with 6254.

Application: kicad
Version: (2015-10-08 BZR 6254)-product release build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1009,GCC 5.2.0,wx containers,compatible with 2.8)
Platform: Linux 4.1.6-1-ARCH i686, 32 bit, Little endian, wxGTK
Boost version: 1.54.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=ON
         KICAD_SCRIPTING_MODULES=ON
         KICAD_SCRIPTING_WXPYTHON=ON
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

--
You received this bug notification because you are subscribed to the bug
report.
https://bugs.launchpad.net/bugs/1503571

Title:
  Linux Pcbnew Seg Faults When Creating a Via Over Smaller Pad

Status in KiCad:
  New

Bug description:
  Linux Pcbnew seg faults almost every time when it is creating a Via
  over a pad of smaller size.

  Here is the version info:

  Application: kicad
  Version: (2015-10-06 BZR 6248, Git 20a3e23)-product release build
  wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1002,GCC 4.9.2,wx containers,compatible with 2.8)
  Platform: Linux 3.19.0-30-generic x86_64, 64 bit, Little endian, wxGTK
  Boost version: 1.59.0
           USE_WX_GRAPHICS_CONTEXT=OFF
           USE_WX_OVERLAY=OFF
           KICAD_SCRIPTING=ON
           KICAD_SCRIPTING_MODULES=ON
           KICAD_SCRIPTING_WXPYTHON=ON
           USE_FP_LIB_TABLE=HARD_CODED_ON
           BUILD_GITHUB_PLUGIN=ON

  Here is the backtrace of the core dump:

  jchen@thermaltake:~ (kicad-local)$ kicad
  Segmentation fault (core dumped)
  jchen@thermaltake:~ (kicad-local)$ gdb kicad
  GNU gdb (Ubuntu 7.9-1ubuntu1) 7.9
  Copyright (C) 2015 Free Software Foundation, Inc.
  License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
  This is free software: you are free to change and redistribute it.
  There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
  and "show warranty" for details.
  This GDB was configured as "x86_64-linux-gnu".
  Type "show configuration" for configuration details.
  For bug reporting instructions, please see:
  <http://www.gnu.org/software/gdb/bugs/>.
  Find the GDB manual and other documentation resources online at:
  <http://www.gnu.org/software/gdb/documentation/>.
  For help, type "help".
  Type "apropos word" to search for commands related to "word"...
  Reading symbols from kicad...(no debugging symbols found)...done.
  (gdb) core core
  warning: exec file is newer than core file.
  [New LWP 20210]
  [New LWP 20241]
  [New LWP 20239]
  [New LWP 20211]
  [New LWP 20237]
  [New LWP 20244]
  [New LWP 20213]
  [New LWP 20238]
  [New LWP 20245]
  [New LWP 20212]
  [New LWP 20242]
  [New LWP 20...

[Joseph Chen (joseph-chen59)]

I am providing more source level info from a recompiled kicad that is not stripped. As you can see, the newer version of kicad still seg faults. Hope this new backtrace helps.

Application: kicad
Version: (2015-10-08 BZR 6256, Git 7d5cb75)-product debug build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1002,GCC 4.9.2,wx containers,compatible with 2.8)
Platform: Linux 3.19.0-30-generic x86_64, 64 bit, Little endian, wxGTK
Boost version: 1.59.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=ON
         KICAD_SCRIPTING_MODULES=ON
         KICAD_SCRIPTING_WXPYTHON=ON
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

(gdb)
(gdb) core core
[New LWP 15664]
[New LWP 15695]
[New LWP 15665]
[New LWP 15696]
[New LWP 15698]
[New LWP 15693]
[New LWP 15701]
[New LWP 15700]
[New LWP 15694]
[New LWP 15683]
[New LWP 15666]
[New LWP 15667]
[New LWP 15703]
[New LWP 15702]
[New LWP 15697]
[New LWP 15699]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
bt
Core was generated by `kicad'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0 0x00007fb88afc904a in VECTOR2<int>::operator= (this=0x498c890, aVector=...) at /home/jchen/git-kicad/kicad-source-mirror/include/math/vector2d.h:316
316 x = aVector.x;
(gdb) bt
#0 0x00007fb88afc904a in VECTOR2<int>::operator= (this=0x498c890, aVector=...) at /home/jchen/git-kicad/kicad-source-mirror/include/math/vector2d.h:316
#1 0x00007fb88b524591 in PNS_VIA::PNS_VIA (this=0x498c850, aPos=..., aLayers=..., aDiameter=762000, aDrill=508000, aNet=-1, aViaType=VIA_THROUGH) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_via.h:50
#2 0x00007fb88b559e21 in PNS_LINE_PLACER::makeVia (this=0x4e822b0, aP=...) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:79
#3 0x00007fb88b55b5ea in PNS_LINE_PLACER::rhShoveOnly (this=0x4e822b0, aP=..., aNewHead=...) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:459
#4 0x00007fb88b55bac1 in PNS_LINE_PLACER::routeHead (this=0x4e822b0, aP=..., aNewHead=...) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:521
#5 0x00007fb88b55c05e in PNS_LINE_PLACER::routeStep (this=0x4e822b0, aP=...) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:613
#6 0x00007fb88b55c1f9 in PNS_LINE_PLACER::route (this=0x4e822b0, aP=...) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:649
#7 0x00007fb88b55cc9e in PNS_LINE_PLACER::Move (this=0x4e822b0, aP=..., aEndItem=0x0) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:829
#8 0x00007fb88b559ee1 in PNS_LINE_PLACER::ToggleVia (this=0x4e822b0, aEnabled=true) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_line_placer.cpp:91
#9 0x00007fb88b523eb6 in PNS_ROUTER::ToggleViaPlacement (this=0x5531090) at /home/jchen/git-kicad/kicad-source-mirror/pcbnew/router/pns_router.cpp:1020
#10 0x00007fb88b4f71ed in ROUTER_TOOL::onViaCommand (this=0x5e8cf20, aType=VIA_THROUGH) at /home/jchen/git-kicad/kicad-sour...

[firewalker (firew4lker)]

Yes, I tried with the attached board.

[jean-pierre charras (jp-charras)]

I also tried with the attached board, and I have no crash (both Windows 7 and Linux Kubuntu 14.04)

[Joseph Chen (joseph-chen59)]

On 10/09/2015 11:51 AM, jean-pierre charras wrote:
> I also tried with the attached board, and I have no crash (both Windows
> 7 and Linux Kubuntu 14.04)
>
Those who is unable to reproduce the crash with using my attached PCB
file, could you copy and paste you Kicad version info? Thus we can
compare if there is any difference.

Also, Has anyone looked at my latest backtrace of the dumped core? By
looking at it, the crash occurs at

Inside of template VECTOR2's "operator=(const VECTOR<t>& aVector), at
line 316 of file vector2d.h, which is part of Kicad source. invoked by
the constructor of PNS_VIA().

I suspect that the constructor of PNS_VIA(), on the crash condition, is
supplied with a invalid VECTOR2 instance. Please look at the code there
to see if you can figure it out.

Plus, there is a interesting finding: when the via dia size is "0,03"
inch, pcbnew crashes when pressing key "v" on pad GND of C1, while using
a smaller via dia size "0.02" inch it does not crash. Does this ring a
bell for you?

FYI, my running kicad is on Ubuntu 15.04.

--JC

[Joseph Chen (joseph-chen59)]

Just try to be a bit helpful with more debug info, since I can reproduce the crash ALL the times.....

I've narrowed down a bit to the source code that seems to be the root cause to the crash:

My debugging printouts have shown that the root cause to the crash is a BAD vector of type "VECTOR2I" is returned by the following expression:

l.CPoint( -1 )

which is invoked at line 459 of file "pcbnew/router/pns_line_placer.cpp", shown here:

  if( m_placingVia && viaOk )
    {
        PNS_VIA v1( makeVia( l.CPoint( -1 ) ) ); //---------------------- Line 459
        PNS_VIA v2( makeVia( l2.CPoint( -1 ) ) );

        l.AppendVia( v1 );
        l2.AppendVia( v2 );
    }

My printouts show that when creating a 0.02 inch via, the referenced address is a good one:

&l.CPoint(-1) = 0x34118c8

When creating a 0.03 inch via, the address is a BAD one:

&l.CPoint(-1) = 0xfffffffffffffff8

Again, the key step to reproduce the crash is to create a Via of Size 0.03 inch on the top of the pad of SMD 0402 footprint.

Hope this is helpful to you guys.

--JC

[Joseph Chen (joseph-chen59)]

To correct my comment above: The footprint should be smd 0201, not smd 0402.

So ...

The key step to reproduce the crash is to create a via of size 0.03 inch on the top of a pad of smd 0201, in OpenGL view.

[Wayne Stambaugh (stambaughw)]

Can anyone reproduce this issue or has it been fixed? Please set the status as invalid or fix released if it is no longer valid.

[Jon Neal (reportingsjr)]

I just tried recreating it with the board Joseph provided and I can't get it to segfault. Joseph, can you provide a video that shows exactly what you are doing? (although there is a lot of debug info here, maybe Tom can figure it out anyways)

My debug info:
Application: pcbnew
Version: (2015-10-08 BZR 6256, Git 7d5cb75)-product debug build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1008,GCC 5.1.1,wx containers,compatible with 2.8)
Platform: Linux 4.0.4-303.fc22.x86_64 x86_64, 64 bit, Little endian, wxGTK
Boost version: 1.57.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=OFF
         KICAD_SCRIPTING_MODULES=OFF
         KICAD_SCRIPTING_WXPYTHON=OFF
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

[Joseph Chen (joseph-chen59)]

Hi Jon Neal, I see only difference between your kicad and mime is the Boost version, yours is 1.57.0. My kicad crashes with either of Bost 1.59.0 from Boost direct source, or 1.47.0 or with and mine is 1.54.0 that comes with the Launchpad.

[Joseph Chen (joseph-chen59)]

Now it seems to be narrowed down to condition that pcbnew crashes: a wider track width.

Attached you can find my second file set for reproducing the crash.
 This time, the file simpler: all two footprints are smd 0201, and the "default" design rule's track width is changed to 0.02 inch. This way, any key "v" pressing on a top of any pad will cause Pcbnew to crash.

Here is the values for the "default" under the "Net Classes Editor":

Clearance: 0.008
Track Width: 0.02
Via Dia: 0.02
Via Drill: 0.015
uVia Dia: 0.01
uVia Drill: 0.004.

Steps for reproducing the crash:

1. Open the PCB file with Pcbnew
2. Zoom in so the footprints are large enough so you can click on a pad right at the center
3. Press key "F11" to switch to OpenGL view.
4. Press key "X" to switch to "Routing" mode
5. Right-Click to get the "Interactive Router Settings"
6. Select "Shove" Mode (this is the key option for reproducing the crash)
7. Put check marks for these options:
"Remove redundant tracks"
8. Put chec

6. Click "OK"

[Joseph Chen (joseph-chen59)]

I accidentally push the button (wish there is "Edit" option to change or correct the posted comments )

Here is the full comment entry ...

Now it seems to be narrowed down to condition that pcbnew crashes: a wider track width.

Attached you can find my second file set for reproducing the crash.
 This time, the file simpler: all two footprints are smd 0201, and the "default" design rule's track width is changed to 0.02 inch. This way, any key "v" pressing on a top of any pad will cause Pcbnew to crash.

Here is the values for the "default" under the "Net Classes Editor":

Clearance: 0.008
Track Width: 0.02
Via Dia: 0.02
Via Drill: 0.015
uVia Dia: 0.01
uVia Drill: 0.004.

Steps for reproducing the crash:

1. Open the PCB file with Pcbnew
2. Zoom in so the footprints are large enough so you can click on a pad right at the center
3. Press key "F11" to switch to OpenGL view.
4. Press key "X" to switch to "Routing" mode
5. Right-Click to get the "Interactive Router Settings"
6. Select "Shove" Mode (this is the key option for reproducing the crash)
7. Put check marks for these options:
         "Remove redundant tracks"
        "Automatic neckdonw"
      "Smooth dragged segments"
8. Slide "Optimizer effort" to all the way to "high"
9. Click "OK" to close the pop-up window
10. Click on the center of any of the pads of the two footprints.
11. Without moving the mouse, press key "V".

Now you can see pcbnew crashes.

[firewalker (firew4lker)]

Confirmed. It crashed.

Application: kicad
Version: (2015-10-08 BZR 6254)-product release build
wxWidgets: Version 3.0.2 (debug,wchar_t,compiler with C++ ABI 1009,GCC 5.2.0,wx containers,compatible with 2.8)
Platform: Linux 4.1.6-1-ARCH i686, 32 bit, Little endian, wxGTK
Boost version: 1.54.0
         USE_WX_GRAPHICS_CONTEXT=OFF
         USE_WX_OVERLAY=OFF
         KICAD_SCRIPTING=ON
         KICAD_SCRIPTING_MODULES=ON
         KICAD_SCRIPTING_WXPYTHON=ON
         USE_FP_LIB_TABLE=HARD_CODED_ON
         BUILD_GITHUB_PLUGIN=ON

[Tomasz Wlostowski (twlostow)]

Hi,

Many thanks to all of you for the investigation and sample files!

Crash occurs when the shove algorithm fails to find a non-colliding track that leads to the via being placed. Here's a patch. We'll commit it tomorrow.

Tom

[Joseph Chen (joseph-chen59)]

On 10/11/2015 05:44 PM, Tomasz Wlostowski wrote:
> Hi,
>
> Many thanks to all of you for the investigation and sample files!
>
> Crash occurs when the shove algorithm fails to find a non-colliding
> track that leads to the via being placed. Here's a patch. We'll commit
> it tomorrow.
>
> Tom
>
> ** Patch added: "0001-router-fix-segfault-when-placing-a-via-with-failed-s.patch"
> https://bugs.launchpad.net/kicad/+bug/1503571/+attachment/4491974/+files/0001-router-fix-segfault-when-placing-a-via-with-failed-s.patch
>
Thank YOU! Now I can continue doing the Via-In-Pad layout. I love the
great "Push-n-Shove".

Appreciate it very much,

--JC