existing system, now raises: ppcx64 can't be executed, error message: -11 lazarus / 62-General ---------------------------------------------------------------------- Ozz_Nixon · 2017-05-22 17:46 # /usr/bin/ppcx64 Free Pascal Compiler version 2.6.4 [2014/03/07] for x86_64 * not aware of any yum updates or anything... but, now I get: ./buildproduction Free Pascal Compiler version 2.6.4 [2014/03/07] for x86_64 Copyright (c) 1993-2014 by Florian Klaempfl and others Target OS: Linux for x86-64 Compiling modernpascal.pp Compiling _parser.pas Compiling _types.pas Compiling _exceptions.pas Compiling _compiler.pas Compiling _vartypes.pas Compiling _emitter.pas Compiling _interpreter.pas Compiling _compiler.pas Compiling _vartypes.pas Compiling _emitter.pas Compiling _vartypes2.pas Compiling _eval.pas Compiling _vartypes3.pas Compiling _astree.pas Compiling _vartypes4.pas Compiling _vartypes3.pas Compiling /source/bpdx/dxutil_environment.pas Compiling /source/bpdx/dxutil_common.pas Compiling /source/bpdx/dxutil_numeric.pas Compiling /source/bpdx/dxutil_string.pas Error: /usr/bin/ppcx64 can't be executed, error message: Failed to execute "/usr/bin/ppcx64", error code: -11  dxutil* source has not been changed in over a decade. I manually dropped all *.ppu and *.o files and forced with a -B, still same error same file. Skimmed dxutil_string.pas - nothing appears corrupt. ---------------------------------------------------------------------- Thaddy · 2017-05-22 18:42 chmod +x /usr/bin/ppcx64 maybe? It can also be that that is a symlink and ppcx64 is simply missing in /usr/lib/fpc/ or /usr/local/lib/fpc/ (version is 3.0.0, 3.0.2 or something) Because on most nixes that is the real place where the ppcx64 executable is installed and /usr/bin or /usr/local/bin contains a symlink. If the package was built properly or you installed correctly. ---------------------------------------------------------------------- Ozz_Nixon · 2017-05-22 19:51 Thanks @Thaddy, however as said - this is an existing system (e.g. 2.6.4 has been on it since last year) - this is a production build server, I copy source changes to modernpascal (my free pascal interpreter), and run build to deploy to my web servers. I made a change to my process locking code (3 lines) and now cannot deploy the update to my users. * Code change was rolled back, still same problem. # whereis ppcx64 /usr/bin/ppcx64 # ls -alrt /usr/bin/ppcx64 lrwxrwxrwx 1 root root 27 Feb 22  2016 /usr/bin/ppcx64 -> /usr/lib64/fpc/2.6.4/ppcx64 # ls -alrt /usr/lib64/fpc/2.6.4/ppcx64 -rwxr-xr-x 1 root root 2869176 Mar  7  2014 /usr/lib64/fpc/2.6.4/ppcx64 *** anyone know what the -11 error code means, or where to find the error numbers for ppc itself? THANKS! ---------------------------------------------------------------------- Jonas Maebe · 2017-05-22 19:59 I guess it's signal 11, which is a segmentation violation. It means the compiler binary has crashed. ---------------------------------------------------------------------- Ozz_Nixon · 2017-05-22 20:53 found the code which is causing it. rolled back 1 more version of changes, and viola everything works. so tracking through the code, it is in an include file, the method looks something like: procedure doGlobalLock; inline; {$ifdef unix} label tryagain; {$endif} begin {$ifdef unix} tryagain:    if {I can lock then globallocksemaphore}    else begin       {sleep 10ms}       goto tryagain;    end; {$endif}    doCrossThreadLocking; end; -- the goto tryagain raises the -11 error. ---------------------------------------------------------------------- Jonas Maebe · 2017-05-22 21:01 The problem is the "goto" in an inline routine. IIRC that was fixed in 3.0. ---------------------------------------------------------------------- Ozz_Nixon · 2017-05-22 21:02 Quote from: ozznixon on May 22, 2017, 10:53:57 pmso tracking through the code, it is in an include file, the method looks something like: procedure doGlobalLock; inline; {$ifdef unix} label tryagain; {$endif} So, I am not sure if it is the goto inside an inline? if I comment the goto call, all works, so I recoded to be a while true loop, just break out on success instead of goto tryagain on fail. Thanks @jonas (saw your post after I posted this)...