Bug 299581

Summary: calc build fails on ppc with incompatible types error
Product: [Fedora] Fedora Reporter: Matthew Miller <mattdm>
Component: calcAssignee: Matthew Miller <mattdm>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: medium Docs Contact:
Priority: medium    
Version: rawhideCC: dwmw2
Target Milestone: ---   
Target Release: ---   
Hardware: powerpc   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2007-09-30 09:10:44 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---
Bug Depends On:    
Bug Blocks: 179260    

Description Matthew Miller 2007-09-21 01:48:39 UTC
I don't have a power pc system, but calc fails to build in the Fedora
buildsys on ppc32 (builds fine on 64 bit) with this error:

  gcc  -DCALC_SRC -DCUSTOM -Wall -W -Wno-comment   -fPIC -O2 -g -pipe -Wall
       -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector
       --param=ssp-buffer-size=4 -m32   -c -o file.o file.c
  file.c: In function 'filepos2z':
  file.c:1366: error: incompatible types in assignment
  make: *** [file.o] Error 1


The line in question is:

  SWAP_HALF_IN_FILEPOS(ret.v, &pos);

and from fposval.h,

  #define SWAP_HALF_IN_FILEPOS(dest, src) memcpy((void *)(dest), (void *)(src),
sizeof(FPOS_POS_BITS))

pos is a FILEPOS, which is typedef'd to fpos_t from <stdio.h>.

ret is a "ZVALUE", defined in zmath.h, so ret.v is a *HALF, which is
conditionally defined in zmath.h as either USB32 (unsigned long) or USB16
(unsigned short) depending on the value of HAVE_B64.

Something is clearly going wrong in all of that.

Comment 1 Denis Leroy 2007-09-21 11:22:56 UTC
The problem is fposval.h is automatically generated, and the code that generates
it (fposval.c) is completely broken. It tries to detect the size of fpos_t, but
fails.

The best solution is to avoid it to use fgetpos() and fsetpos() at all (and use
fseek() and ftell() instead, like most code nowadays), by adding this to the
compile line:

     HAVE_FPOS="-DHAVE_NO_FPOS" \


However, the rpmbuild process on the system that David kindly gave me access to,
fails at link time with some sort of architecture incompatiblity issue that I
don't quite understand, not being very familiar with ppc. At least, it should
help you move forward.


Comment 2 Matthew Miller 2007-09-21 12:50:20 UTC
Cool. Thanks everyone.

Comment 3 David Woodhouse 2007-09-30 09:10:44 UTC
This is fixed now, right?

Comment 4 Matthew Miller 2007-09-30 11:03:56 UTC
Yes; sorry -- I meant to close it with the comment above. Thanks again.