Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Mon, 14 Apr 2014 20:36:24 +0200

Am Montag, 14. April 2014, 17:34:27 schrieb Graber, Tim:
> Hi Ben-
> I added a few statements to the code in addition to the patch to print out
> more information.  The source codes are attached as well.

The are are no significant difference between your pvSync.i and mine, so I 
guess we can exclude CPP from the list of suspects.

>  When the build
> gets to the point where it is going to print the offending scope->type, the
> process hangs on the function:

> report_at_expr(scope, "scope->type == %d\n", scope->type);
> 	
> and I can't ctrl-c out of it.  I actually have to kill the Cygwin window and
> start over.  

That strongly suggests the 'scope' pointer is corrupt. I guess you will have 
to start it with a debugger, as Eric described. (Thanks, Eric, for the 
suggestion.)

You could also wait until I get a 64 Bit Windows VM so I can debug this 
myself. I ordered one today, but it could take a few days until I can get my 
hands on it.

Cheers
Ben

> Here is the part of the analysis.c code that I changed:
> 
> 	report_at_expr(scope, "scope->type == %d\n", scope->type);
> 	switch(scope->type)
> 	{
> 	case D_PROG:
> 		return scope->prog_defns;
> 	case D_SS:
> 		return scope->ss_defns;
> 	case D_STATE:
> 		return scope->state_defns;
> 	case D_WHEN:
> 		return scope->when_defns;
> 	case D_ENTRY:
> 		return scope->entry_defns;
> 	case D_EXIT:
> 		return scope->exit_defns;
> 	case S_CMPND:
> 		return scope->cmpnd_defns;
> 	default:
> 		report("In default, line = 209\n");
> 		report_at_expr(scope, "offending scope->type == %d\n", scope->type);
> 		assert(impossible); 
> 		return NULL;
> 
> Here is the output from the build:
> 
> ../../../bin/cygwin-x86_64/snc.exe   +m -i  pvSync.i -o pvSync.c.tmp
> ../pvSync.st:7: scope->type == 6
> ../pvSync.st:33: scope->type == 2
> ../pvSync.st:41: scope->type == 7
> ../pvSync.st:42: scope->type == 8
> ../pvSync.st:43: scope->type == 11
> ../pvSync.st:51: scope->type == 7
> ../pvSync.st:52: scope->type == 8
> ../pvSync.st:54: scope->type == 11
> ../pvSync.st:64: scope->type == 7
> ../pvSync.st:65: scope->type == 8
> ../pvSync.st:66: scope->type == 11
> ../pvSync.st:75: scope->type == 8
> ../pvSync.st:76: scope->type == 11
> ../pvSync.st:80: scope->type == 25
> 
> Tim
> 
> 
> -----Original Message-----
> From: Benjamin Franksen [mailto:[email protected]] 
> Sent: Monday, April 14, 2014 10:31 AM
> To: Mark Rivers
> Cc: Graber, Tim; [email protected]
> Subject: Re: Sequencer build problem cygwin64
> 
> Oops, no core file. That makes things more difficult.
> 
> Can you apply this patch:
> 
> diff -rN -u old-branch-2-1/src/snc/analysis.c new-
> branch-2-1/src/snc/analysis.c
 --- old-branch-2-1/src/snc/analysis.c  
> 2014-04-14 17:30:40.000000000 +0200
> +++ new-branch-2-1/src/snc/analysis.c   2014-04-14 17:30:40.000000000 
> +0200
> @@ -205,6 +205,7 @@
>         case S_CMPND:
>                 return scope->cmpnd_defns;
>         default:
> +               report_at_expr(scope, "offending scope->type == %d\n",
> scope->type);
>                 assert(impossible); return NULL;
>         }
>  }
> 
> and then run again and send the output?
> 
> Cheers
> Ben
> 
> On Monday, April 14, 2014 15:22:13 Mark Rivers wrote:
> 
> > The last time I checked on Cygwin it did not produce useful core 
> > files, so gdb could not be used with them.
> 
>  
> 
> > Maybe it has improved since I last tried it?
> > 
> > Mark
> > 
> > 
> > -----Original Message-----
> > From: [email protected]
> > [mailto:[email protected]] On Behalf Of Benjamin Franksen
> 
>  Sent: Monday, April 14, 2014 10:15 AM
> 
> > To: Graber, Tim; [email protected]
> > Subject: Re: Sequencer build problem cygwin64
> > 
> > Hi Tim
> > 
> > can you send me the original pvSync.i? Maybe there is a problem with 
> > the
> 
>  C preprocessor on your platform.
> 
> > 
> > Also, can you use gdb to analyse the core for the aborted snc run?
> > Start
> 
>  it like
> 
> > 
> > 
> >   gdb bin/cygwin-x86_64/snc.exe test/validate/O.cygwin-x86_64/core
> > 
> > 
> > and enter
> > 
> > 
> >   backtrace
> > 
> > 
> > then send me the output.
> > 
> > Thanks
> > Ben
> > 
> > On Monday, April 14, 2014 02:51:42 Graber, Tim wrote:
> > 
> > 
> > > Hi Ben-
> > > I did a bit more debugging and found the point in the pvSync.i file 
> > > that is causing the problem.  It's at line 84:
> > > 
> > > 
> > > 
> > >        if (i >= 1000 -1) {
> > > 	
> > > 	
> > > 	state done;
> > > 	
> > > 	
> > >        }
> > > 
> > > 
> > > 
> > > When I replaced the above by
> > > 
> > > %%       if (i >= 1000 -1) {
> > > 
> > > 
> > > 	state done;
> > > 
> > > 
> > > %%      }
> > > 
> > > and ran the command
> > > 
> > > ../../../bin/cygwin-x86_64/snc.exe   +m -i  pvSync.i -o
> > > pvSync.c.tmp
> > > 
> > > in the O.cygwin-x86_64 directory, it produced  pvSync.c.tmp without 
> > > error.
> > 
> > 
> > 
> > 
> > 
> > 
> > > 
> > > Tim
> > > -----Original Message-----
> > > From: [email protected]
> > > [mailto:[email protected]] On Behalf Of Johnson, Andrew 
> > > N. Sent: Sunday, April 13, 2014 9:38 PM
> > > To: Benjamin Franksen
> > > Cc: [email protected]
> > > Subject: Re: Sequencer build problem cygwin64
> > > 
> > > Hi Ben,
> > > 
> > > The OSCLASS for the cygwin-x86_64 architecture is 'cygwin32' so that 
> > > include path is actually correct. This is an accident of history, as 
> > > is the OSCLASS 'WIN32' of the windows-x64 architectures; the OS APIs 
> > > didn't change significantly when they went to 64-bit and we didn't 
> > > want to duplicate code unnecessarily, so we still use the old names.
> > > 
> > > Your "impossible" assertion condition might not be (whatever it is).
> > > 
> > > - Andrew
> > > 
> > > --
> > > Sent from my iPad
> > > 
> > > 
> > > 
> > > > On Apr 13, 2014, at 5:52 PM, "Benjamin Franksen"
> > > > <[email protected]> wrote:
> > > > 
> > > > Hi Tim
> > > > 
> > > > looking at the make output, I notice that your include path has - 
> > > > I../../../include/os/cygwin32, even though you compile for 
> > > > O.cygwin-x86_64. This looks inconsistent to me, and could perhaps 
> > > > explain the assertion failure. I may be wrong, though, in which 
> > > > case we'd have to look deeper into what's broken (I have never 
> > > > compiled on this platform, only on 32 bit Windows native).
> > > > 
> > > > Cheers
> > > > Ben
> > > > 
> > > > Am Sonntag, 13. April 2014, 17:17:11 schrieb Graber, Tim:
> > > > 
> > > > 
> > > >> I am trying to build the EPICS Sequencer (version seq-2.1.15) 
> > > >> using
> > > >> 64-bit Cygwin.   I get the following error when I build the test
> > > >> directory and a similar error when building the examples 
> > > >> directory.  If I comment out the test and examples directories in 
> > > >> the Makefile, the src directory builds fine.  Any help will be 
> > > >> appreciated. Thanks, Tim Graber
> > > >> 
> > > >> 
> > > >> gcc -c  -D_POSIX_THREADS -D_POSIX_TIMERS            -DUNIX 
> > > >> -DCYGWIN32 -O3   -Wall     -m64   -DEPICS_DLL_NO   -MMD -I.
> > > >> -I../O.Common -I. -I.. -I../../../include/os/cygwin32 
> > > >> -I../../../include
> > > >> -I/usr/local/EPICS/base/include/os/cygwin32
> > > >> -I/usr/local/EPICS/base/include>>
> > > >> 
> > > >> 
> > > >>      opttVar.c g++ -o opttVar.exe -static
> > > >> 
> > > >> 
> > > >> 
> > > >> -L/usr/local/epics/base-3.14.12.4/seq-2.1.15/lib/cygwin-x86_64
> > > >> -L/usr/local/EPICS/base-3.14.12.4/lib/cygwin-x86_64      -m64
> > > >> opttVar.o subThreadSleep.o
> > > >> seqSoftIoc_registerRecordDeviceDriver.o
> > > >> testSupport.o    -lseq -lpv -lrecIoc -lsoftDevIoc -lmiscIoc
> > > >> -lrsrvIoc
> > > >> -ldbtoolsIoc -lasIoc -ldbIoc -lregistryIoc -ldbStaticIoc -lca 
> > > >> -lCom
> > > >> -lpthread    -lm preprocessing ../pvSync.st
> > > >> gcc -x c -E  -D_POSIX_THREADS -D_POSIX_TIMERS            -DUNIX 
> > > >> -DCYGWIN32 -I. -I../O.Common -I. -I..
> > > >> -I../../../include/os/cygwin32
> > > >> -I../../../include -I/usr/local/EPICS/base/include/os/cygwin32
> > > >> -I/usr/local/EPICS/base/include       ../pvSync.st > pvSync.i
> > > >> converting pvSync.i
> > > >> ../../../bin/cygwin-x86_64/snc.exe   +m -i  pvSync.i -o
> > > >> pvSync.c.tmp
> > > >> assertion "impossible" failed: file "../analysis.c", line 209,
> > > >> function: defn_list_from_scope
> > > >> ../../../configure/RULES_BUILD:62:
> > > >> recipe for target 'pvSync.c' failed make[3]: *** [pvSync.c] 
> > > >> Aborted (core dumped)
> > > >> make[3]: Leaving directory
> > > >> '/usr/local/epics/base-3.14.12.4/seq-2.1.15/test/validate/O.cygwi
> > > >> n-
> > > >> x86_64' /usr/local/EPICS/base/configure/RULES_ARCHS:63: recipe 
> > > >> for target 'install.cygwin-x86_64' failed make[2]: *** 
> > > >> [install.cygwin-x86_64] Error 2
> > > >> make[2]: Leaving directory
> > > >> '/usr/local/epics/base-3.14.12.4/seq-2.1.15/test/validate'
> > > >> /usr/local/EPICS/base/configure/RULES_DIRS:70: recipe for target 
> > > >> 'validate.install' failed make[1]: *** [validate.install] Error
> > > >> 2
> > > >> make[1]: Leaving directory
> > > >> '/usr/local/epics/base-3.14.12.4/seq-2.1.15/test'
> > > >> /usr/local/EPICS/base/configure/RULES_DIRS:70: recipe for target 
> > > >> 'test.install' failed make: *** [test.install] Error 2
> > > >> 
> > > >> -------------------------------------------------
> > > >> Timothy Graber
> > > >> Research Professor
> > > >> Washington State University
> > > >> Manager, Dynamic Compression Sector c/o Argonne National 
> > > >> Laboratory Bld. 438F/F002
> > > >> 9700 S. Cass Ave
> > > >> Argonne, IL 60439
> > > >> Phone: (630)252-0462
> > > > 
> > > > 
> > > > 
> > > > --
> > > > Perl: There Is More Than One Way To Do It
> > > > Python: There Is One Way To Do It
> > > > Haskell: There Is One Way To Do It, Up To Isomorphisµ
> > 
> > 
> > --
> > "Make it so they have to reboot after every typo." ― Scott Adams
> 
> 
> --
> "Make it so they have to reboot after every typo." ― Scott Adams

-- 
Perl: There Is More Than One Way To Do It
Python: There Is One Way To Do It
Haskell: There Is One Way To Do It, Up To Isomorphisµ

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System