HEAD b0rked re coff-tic80?

HEAD b0rked re coff-tic80?

[Dave Korn]

  Just did a cvsup not half an hour ago.  After configuring with
--enable-targets=all, (which I don't usually bother with), my build failed
with....


/bin/sh ./libtool --mode=compile gcc -DHAVE_CONFIG_H -I.
-I/usr/build/binutils-s
rc/src/bfd -I. -D_GNU_SOURCE     -I. -I/usr/build/binutils-src/src/bfd
-I/usr/bu
ild/binutils-src/src/bfd/../include
-I/usr/build/binutils-src/src/bfd/../intl -
I../intl   -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2
-c -
o coff-tic80.lo /usr/build/binutils-src/src/bfd/coff-tic80.c
gcc -DHAVE_CONFIG_H -I. -I/usr/build/binutils-src/src/bfd -I. -D_GNU_SOURCE
-I.
-I/usr/build/binutils-src/src/bfd
-I/usr/build/binutils-src/src/bfd/../include -
I/usr/build/binutils-src/src/bfd/../intl -I../intl -W -Wall
-Wstrict-prototypes
-Wmissing-prototypes -Werror -g -O2 -c
/usr/build/binutils-src/src/bfd/coff-tic8
0.c -o coff-tic80.o
In file included from /usr/build/binutils-src/src/bfd/coff-tic80.c:31:
/usr/build/binutils-src/src/include/coff/tic80.h:71:1: "_CONST" redefined
In file included from /usr/include/stdio.h:29,
                 from /usr/build/binutils-src/src/bfd/sysdep.h:33,
                 from /usr/build/binutils-src/src/bfd/coff-tic80.c:29:
/usr/include/_ansi.h:30:1: this is the location of the previous definition
make[3]: *** [coff-tic80.lo] Error 1
make[3]: Leaving directory `/usr/build/binutils-obj/bfd'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/usr/build/binutils-obj/bfd'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/usr/build/binutils-obj/bfd'
make: *** [all-bfd] Error 2
dk@mace /usr/build/binutils-obj> start .
dk@mace /usr/build/binutils-obj>


  This is for a i686-pc-cygwin native build.  Sorry, I'm not going to look
into this myself right this very moment.  If nobody jumps up and says they
know what's going on here, I'll file a bugzilla for it shortly.  Neither
bfd/coff-tic80.c nor include/coff/tic80.h have been changed recently.



    cheers, 
      DaveK
-- 
Can't think of a witty .sigline today....

Re: HEAD b0rked re coff-tic80?

[Nick Clifton]

Hi Dave,

>   Just did a cvsup not half an hour ago.  After configuring with
> --enable-targets=all, (which I don't usually bother with), my build failed
> with....

> In file included from /usr/build/binutils-src/src/bfd/coff-tic80.c:31:
> /usr/build/binutils-src/src/include/coff/tic80.h:71:1: "_CONST" redefined
> In file included from /usr/include/stdio.h:29,
>                  from /usr/build/binutils-src/src/bfd/sysdep.h:33,
>                  from /usr/build/binutils-src/src/bfd/coff-tic80.c:29:
> /usr/include/_ansi.h:30:1: this is the location of the previous definition

This does not happen for me, but I can certainly see how the problem 
might arise.  Since _CONST is not actually used in the tic80 specific 
sources the simplest thing seems to be to remove it.  So... please could 
you try out the patch and let me know if you have any problems with it.

Cheers
   Nick

Index: include/coff/tic80.h
===================================================================
RCS file: /cvs/src/src/include/coff/tic80.h,v
retrieving revision 1.2
diff -c -3 -p -r1.2 tic80.h
*** include/coff/tic80.h	14 Mar 2001 02:27:44 -0000	1.2
--- include/coff/tic80.h	26 Apr 2005 16:56:58 -0000
*************** struct external_scnhdr
*** 67,77 ****
--- 67,81 ----
   #define _TEXT	".text"
   #define _DATA	".data"
   #define _BSS	".bss"
+ #if 0 /* The following are not used in the binutils sources and the 
definition
+ 	 of _CONST can clash with a definition provided by the system header
+ 	 file _ansi.h  */
   #define _CINIT	".cinit"
   #define _CONST	".const"
   #define _SWITCH	".switch"
   #define _STACK	".stack"
   #define _SYSMEM	".sysmem"
+ #endif

   #define	SCNHDR	struct external_scnhdr
   #define	SCNHSZ	40

RE: HEAD b0rked re coff-tic80?

[Dave Korn]

----Original Message----
>From: Nick Clifton
>Sent: 26 April 2005 17:57

> Hi Dave,
> 
>>   Just did a cvsup not half an hour ago.  After configuring with
>> --enable-targets=all, (which I don't usually bother with), my build
>> failed with....
> 
>> In file included from /usr/build/binutils-src/src/bfd/coff-tic80.c:31:
>> /usr/build/binutils-src/src/include/coff/tic80.h:71:1: "_CONST" redefined

> This does not happen for me, but I can certainly see how the problem
> might arise.  Since _CONST is not actually used in the tic80 specific
> sources the simplest thing seems to be to remove it.  So... please could
> you try out the patch and let me know if you have any problems with it.
> 
> Cheers
>    Nick
> 
> Index: include/coff/tic80.h
> ===================================================================


  That helped, but now it falls apart a bit further along, and this time
with a different target:

gcc -DHAVE_CONFIG_H -I. -I/usr/build/binutils-src/src/gas -I. -D_GNU_SOURCE
-I.
-I/usr/build/binutils-src/src/gas -I../bfd
-I/usr/build/binutils-src/src/gas/con
fig -I/usr/build/binutils-src/src/gas/../include
-I/usr/build/binutils-src/src/g
as/.. -I/usr/build/binutils-src/src/gas/../bfd
-I/usr/build/binutils-src/src/gas
/../intl -I../intl -DLOCALEDIR="\"/usr/build/install/share/locale\""   -W
-Wall
-Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2 -c
/usr/build/binutils-s
rc/src/gas/config/obj-coff.c
In file included from /usr/build/binutils-src/src/gas/config/obj-multi.h:26,
                 from obj-format.h:1,
                 from /usr/build/binutils-src/src/gas/config/te-pe.h:7,
                 from targ-env.h:1,
                 from /usr/build/binutils-src/src/gas/as.h:631,
                 from /usr/build/binutils-src/src/gas/config/obj-coff.c:25:
/usr/build/binutils-src/src/gas/config/obj-coff.h:73:1: "TARGET_FORMAT"
redefine
d
In file included from targ-cpu.h:1,
                 from /usr/build/binutils-src/src/gas/config/obj-coff.h:38,
                 from /usr/build/binutils-src/src/gas/config/obj-multi.h:26,
                 from obj-format.h:1,
                 from /usr/build/binutils-src/src/gas/config/te-pe.h:7,
                 from targ-env.h:1,
                 from /usr/build/binutils-src/src/gas/as.h:631,
                 from /usr/build/binutils-src/src/gas/config/obj-coff.c:25:
/usr/build/binutils-src/src/gas/config/tc-i386.h:72:1: this is the location
of t
he previous definition
make[3]: *** [obj-coff.o] Error 1
make[3]: Leaving directory `/usr/build/binutils-obj-new/gas'
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory `/usr/build/binutils-obj-new/gas'
make[1]: *** [all] Error 2
make[1]: Leaving directory `/usr/build/binutils-obj-new/gas'
make: *** [all-gas] Error 2
dk@mace /usr/build/binutils-obj-new>


  There's something more fundamental here, isn't there?  Presumably
obj-coff.h and tc-i386.h have not previously been being included in the same
compilation unit.  Previously it was coff-tic80.h and the system's
/usr/include/_ansi.h, so something's now pulling lots of coff-related
headers into places they didn't used to be, perhaps?


    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....

Re: HEAD b0rked re coff-tic80?

[Nick Clifton]

Hi Dave,

>   There's something more fundamental here, isn't there?  Presumably
> obj-coff.h and tc-i386.h have not previously been being included in the same
> compilation unit.  Previously it was coff-tic80.h and the system's
> /usr/include/_ansi.h, so something's now pulling lots of coff-related
> headers into places they didn't used to be, perhaps?

Indeed.  How exactly did you configure this toolchain ?  And what type 
of host are you building it on ?

Cheers
   Nick

RE: HEAD b0rked re coff-tic80?

[Dave Korn]

----Original Message----
>From: Nick Clifton
>Sent: 28 April 2005 11:36

> Hi Dave,
> 
>>   There's something more fundamental here, isn't there?  Presumably
>> obj-coff.h and tc-i386.h have not previously been being included in the
>> same compilation unit.  Previously it was coff-tic80.h and the system's
>> /usr/include/_ansi.h, so something's now pulling lots of coff-related
>> headers into places they didn't used to be, perhaps?
> 
> Indeed.  How exactly did you configure this toolchain ?  And what type
> of host are you building it on ?



dk@mace /usr/build/binutils-obj> head -7 config.status  | tail -1
# /usr/build/binutils-src/src/configure  --prefix=/usr/build/install -v
dk@mace /usr/build/binutils-obj> /usr/build/binutils-src/src/config.guess
i686-pc-cygwin
dk@mace /usr/build/binutils-obj>


    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....

RE: HEAD b0rked re coff-tic80?

[Dave Korn]

----Original Message----
>From: Dave Korn
>Sent: 28 April 2005 11:44

> ----Original Message----
>> From: Nick Clifton
>> Sent: 28 April 2005 11:36
> 
>> Hi Dave,
>> 
>>>   There's something more fundamental here, isn't there?  Presumably
>>> obj-coff.h and tc-i386.h have not previously been being included in the
>>> same compilation unit.  Previously it was coff-tic80.h and the system's
>>> /usr/include/_ansi.h, so something's now pulling lots of coff-related
>>> headers into places they didn't used to be, perhaps?
>> 
>> Indeed.  How exactly did you configure this toolchain ?  And what type
>> of host are you building it on ?
> 
> 
> 
> dk@mace /usr/build/binutils-obj> head -7 config.status  | tail -1
> # /usr/build/binutils-src/src/configure  --prefix=/usr/build/install -v
> dk@mace /usr/build/binutils-obj> /usr/build/binutils-src/src/config.guess
> i686-pc-cygwin
> dk@mace /usr/build/binutils-obj>

  Oops.  Wrong build dir.  Host still applies, but configure command did of
course have all targets enabled:

dk@mace /usr/build/binutils-obj-new> head -7 config.status  | tail -1
# /usr/build/binutils-src/src/configure  --prefix=/usr/build/install
--enable-ta
rgets=all -v

..pardon the noise.  I'm just trying it again with a freshly-checkedout cvs
tree; if anything changes I'll report.

    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....

Re: HEAD b0rked re coff-tic80?

[Nick Clifton]

Hi Dave,

>>Indeed.  How exactly did you configure this toolchain ?  And what type
>>of host are you building it on ?

> dk@mace /usr/build/binutils-obj> head -7 config.status  | tail -1
> # /usr/build/binutils-src/src/configure  --prefix=/usr/build/install -v
> dk@mace /usr/build/binutils-obj> /usr/build/binutils-src/src/config.guess
> i686-pc-cygwin

I have no idea why a cygwin native build should be behaving like this. 
Of course this would happen when my own cygwin machine is busted.  All I 
can suggest is that you investigate some more and try to find out where 
these tic80 files are being included.  There should be no need for a 
cygwin native build to access them at all.

Cheers
   Nick

RE: HEAD b0rked re coff-tic80?

[Dave Korn]

----Original Message----
>From: Nick Clifton
>Sent: 28 April 2005 13:07

> Hi Dave,
> 
>>> Indeed.  How exactly did you configure this toolchain ?  And what type
>>> of host are you building it on ?
> 
>> dk@mace /usr/build/binutils-obj> head -7 config.status  | tail -1
>> # /usr/build/binutils-src/src/configure  --prefix=/usr/build/install -v
>> dk@mace /usr/build/binutils-obj> /usr/build/binutils-src/src/config.guess
>> i686-pc-cygwin
> 
> I have no idea why a cygwin native build should be behaving like this.
> Of course this would happen when my own cygwin machine is busted.  All I
> can suggest is that you investigate some more and try to find out where
> these tic80 files are being included.  There should be no need for a
> cygwin native build to access them at all.
> 

  I refer the honourable gentleman to my earlier reply!  It was of course an
all-targets build.

  I notice that the last-modified date on my /usr/include/_ansi.h is 1
April.  I'll see if maybe the w32api package changed recently.


    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....

Re: HEAD b0rked re coff-tic80?

[Nick Clifton]

Hi Dave,

>   I refer the honourable gentleman to my earlier reply!  It was of course an
> all-targets build.

Our emails crossed in the ether...

>   I notice that the last-modified date on my /usr/include/_ansi.h is 1
> April.  I'll see if maybe the w32api package changed recently.

Hmmm - that would still not explain the tc-i386.h vs obj-coff.h conflict 
though.  Since I cannot reproduce this problem locally however, I am 
going to have to leave it with you to come up with a solution.  Perhaps 
something as simple as protecting the #define TARGET_FORMAT 
declarations, although it would be nice to know why it is breaking for 
you and not for me.

Cheers
   Nick

RE: HEAD b0rked re coff-tic80?

[Dave Korn]

----Original Message----
>From: Nick Clifton
>Sent: 28 April 2005 17:57

> Hi Dave,
> 
>>   I refer the honourable gentleman to my earlier reply!  It was of
>> course an all-targets build.
> 
> Our emails crossed in the ether...
> 
>>   I notice that the last-modified date on my /usr/include/_ansi.h is 1
>> April.  I'll see if maybe the w32api package changed recently.
> 
> Hmmm - that would still not explain the tc-i386.h vs obj-coff.h conflict
> though.  Since I cannot reproduce this problem locally however, I am
> going to have to leave it with you to come up with a solution.  Perhaps
> something as simple as protecting the #define TARGET_FORMAT
> declarations, although it would be nice to know why it is breaking for
> you and not for me.


  OK, I'll try and take a look at how the #include hierarchy is interacting
with the generated source files.  You said your cygwin machine was dead, but
do you have any other x86 host hanging around?  I  might want to ask you to
send me the output from "make all" at some point, so I can compare the
command lines and options.  I'll look at it on my own first, though, and
tell you if I can track it down.


    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....