#1847 Z80: Caught signal 22: SIGABRT

open
nobody
z80 port (188)
Z80
1
2014-01-02
2011-09-16
No

The code from bug 3408654 crashes inside Boost when compiled with a recent snapshot #6854.

Discussion

  • Maarten Brock

    Maarten Brock - 2011-09-16
     
  • Philipp Klaus Krause

    I just did a clean checkout of sdcc revision #6854, configured and compiled.

    The attached file compiles fine for me using
    sdcc -mz80 bug3410415.c. Did you use any special options? On which platform did you observe the bug?

    Philipp

    P.S.: Could your problem have been caused by #3401455?

     
  • Maarten Brock

    Maarten Brock - 2011-09-19

    I also did a fresh checkout of #6863 and rebuilt it with MSVC 2010 Express. First I tried a Release build and it passed. Then I used a Debug build and it failed as reported. I have Boost 1.44 installed. I'll try to investigate a bit deeper.

     
  • Maarten Brock

    Maarten Brock - 2011-09-19

    Installing Boost 1.47 did not make any difference.

     
  • Lee Morrison

    Lee Morrison - 2011-10-03

    I checked out snapshot 6854 under cygwin, did (a clean) build, but did not get a crash when running:
    sdcc -mz80 -c bug3410415.c

    It was built using "gcc (GCC) 4.3.4 20090804 (release) 1"

     
  • Borut Ražem

    Borut Ražem - 2011-10-30

    Maarten, does it crash also if you use the snapshot build WIN32 binaries?

     
  • Maarten Brock

    Maarten Brock - 2011-10-30

    No, it doesn't crash with the MINGW32 built binary snapshot.

    It only crashes with a Microsoft Visual C Debug-build. A Release-build works fine. I also guess that all GCC builds are fine based on the replies here. I have a strong feeling the problem lies in the boost code or in Visual C++ although I cannot rule out improper use of boost by the sdcc code.

    I lowered the priority as this only impounds those debugging SDCC itself with Visual C++.

     
  • Maarten Brock

    Maarten Brock - 2011-10-30
    • priority: 5 --> 1
     
  • Philipp Klaus Krause

    Can you get a backtrace of this issue? Did sdcc emit any warning or error messages prior to crashing?

    Philipp

     
  • Philipp Klaus Krause

    Is this issue still there in current sdcc?

    Philipp

     
  • Maarten Brock

    Maarten Brock - 2012-08-19

    call stack trace

     
  • Maarten Brock

    Maarten Brock - 2012-08-19

    Unfortunately yes. I've installed Boost 1.50 and this is what I get now:

    Debug Assertion Failed!
    [...]
    Expression: Invalid operator <

    I've attached a call stack trace.

    It seems to be triggered by calling sort on the list alist in tree_dec_ralloc_forget.

     
  • Philipp Klaus Krause

    • status: open --> pending-works-for-me
    • Category: --> Z80
     
  • Maarten Brock

    Maarten Brock - 2014-01-02
    • status: pending-works-for-me --> open
     
  • Maarten Brock

    Maarten Brock - 2014-01-02

    I'm pretty sure this was not tested on MSVC again when it was set to "works-for-me". It still fails as explained with boost 1.55.0 on MSVC 2012 (Express) for a Debug build only.

     
    Last edit: Maarten Brock 2014-01-02

Log in to post a comment.

Get latest updates about Open Source Projects, Conferences and News.

Sign up for the SourceForge newsletter:





No, thanks