Your bfd changes broke sh sim

Your bfd changes broke sh sim

[Hans-Peter Nilsson]

Check out the cvs module "sim" (like "binutils")
configure with --target=sh-elf
make all-sim

Observe:
gcc -DHAVE_CONFIG_H     -DPROFILE=1 -DWITH_PROFILE=-1          -DDEFAULT_INLINE=0              -I. -I/tmp/hpautotest-sim/src/sim/sh -I../common -I/tmp/hpautotest-sim/src/sim/sh/../common -I../../include -I/tmp/hpautotest-sim/src/sim/sh/../../include -I../../bfd -I/tmp/hpautotest-sim/src/sim/sh/../../bfd -I../../opcodes -I/tmp/hpautotest-sim/src/sim/sh/../../opcodes  -g -O2 -o run \
	  run.o libsim.a ../../bfd/libbfd.a ../../opcodes/libopcodes.a  ../../libiberty/libiberty.a -lz -lnsl  -lm
../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1bc8): undefined reference to `pe_print_ce_compressed_pdata'
../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1d28): undefined reference to `pe_print_ce_compressed_pdata'
collect2: ld returned 1 exit status
make[3]: Leaving directory `/tmp/hpautotest-sim/sh-elf/sim/sh'
make[3]: *** [run] Error 1

Please fix.

brgds, H-P

Re: Your bfd changes broke sh sim

[Danny Backx]

I've tested (and tested again) with
	./configure --enable-targets=all

Is this different ? Are some problems not caught this way ?

	Danny


On Fri, 2008-07-18 at 18:22 +0200, Hans-Peter Nilsson wrote:
> Check out the cvs module "sim" (like "binutils")
> configure with --target=sh-elf
> make all-sim
> 
> Observe:
> gcc -DHAVE_CONFIG_H     -DPROFILE=1 -DWITH_PROFILE=-1          -DDEFAULT_INLINE=0              -I. -I/tmp/hpautotest-sim/src/sim/sh -I../common -I/tmp/hpautotest-sim/src/sim/sh/../common -I../../include -I/tmp/hpautotest-sim/src/sim/sh/../../include -I../../bfd -I/tmp/hpautotest-sim/src/sim/sh/../../bfd -I../../opcodes -I/tmp/hpautotest-sim/src/sim/sh/../../opcodes  -g -O2 -o run \
> 	  run.o libsim.a ../../bfd/libbfd.a ../../opcodes/libopcodes.a  ../../libiberty/libiberty.a -lz -lnsl  -lm
> ../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1bc8): undefined reference to `pe_print_ce_compressed_pdata'
> ../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1d28): undefined reference to `pe_print_ce_compressed_pdata'
> collect2: ld returned 1 exit status
> make[3]: Leaving directory `/tmp/hpautotest-sim/sh-elf/sim/sh'
> make[3]: *** [run] Error 1
> 
> Please fix.
> 
> brgds, H-P
> 
-- 
Danny Backx ; danny.backx - at - scarlet.be ; http://danny.backx.info

Re: Your bfd changes broke sh sim

[Hans-Peter Nilsson]

> From: Danny Backx <danny.backx@scarlet.be>
> Date: Fri, 18 Jul 2008 20:12:39 +0200

> I've tested (and tested again) with
> 	./configure --enable-targets=all
> 
> Is this different ? Are some problems not caught this way ?

Read again.  This is *sim*, not *binutils*, though both are
users of the BFD library.

> On Fri, 2008-07-18 at 18:22 +0200, Hans-Peter Nilsson wrote:
> > Check out the cvs module "sim" (like "binutils")
> > configure with --target=sh-elf
> > make all-sim
> > 
> > Observe:
> > gcc -DHAVE_CONFIG_H     -DPROFILE=1 -DWITH_PROFILE=-1          -DDEFAULT_INLINE=0              -I. -I/tmp/hpautotest-sim/src/sim/sh -I../common -I/tmp/hpautotest-sim/src/sim/sh/../common -I../../include -I/tmp/hpautotest-sim/src/sim/sh/../../include -I../../bfd -I/tmp/hpautotest-sim/src/sim/sh/../../bfd -I../../opcodes -I/tmp/hpautotest-sim/src/sim/sh/../../opcodes  -g -O2 -o run \
> > 	  run.o libsim.a ../../bfd/libbfd.a ../../opcodes/libopcodes.a  ../../libiberty/libiberty.a -lz -lnsl  -lm
> > ../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1bc8): undefined reference to `pe_print_ce_compressed_pdata'
> > ../../bfd/libbfd.a(coff-sh.o):(.rodata+0x1d28): undefined reference to `pe_print_ce_compressed_pdata'
> > collect2: ld returned 1 exit status
> > make[3]: Leaving directory `/tmp/hpautotest-sim/sh-elf/sim/sh'
> > make[3]: *** [run] Error 1

brgds, H-P

Re: Your bfd changes broke sh sim

[Hans-Peter Nilsson]

> From: Danny Backx <danny.backx@scarlet.be>
> Date: Fri, 18 Jul 2008 20:12:39 +0200

> I've tested (and tested again) with
> 	./configure --enable-targets=all

Looks like you didn't even try and build binutils *without*
--enable-targets=all, because you broke e.g. sh-elf (and yes,
sh-pe too).  For --target=sh-elf, make all:

/bin/sh ./libtool --tag=CC --mode=link gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2   -o size  size.o bucomm.o version.o filemode.o ../bfd/libbfd.la ../libiberty/libiberty.a  -lz 
libtool: link: gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2 -o size size.o bucomm.o version.o filemode.o  ../bfd/.libs/libbfd.a ../libiberty/libiberty.a -lz
../bfd/.libs/libbfd.a(coff-sh.o)(.rodata+0x488): undefined reference to `pe_print_ce_compressed_pdata'
../bfd/.libs/libbfd.a(coff-sh.o)(.rodata+0xc68): undefined reference to `pe_print_ce_compressed_pdata'
collect2: ld returned 1 exit status
make[4]: *** [size] Error 1
make[4]: Leaving directory `/home/hp/combined/binuo/binutils'

For --target=sh-pe, make all:

/bin/sh ./libtool --tag=CC --mode=link gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2   -o size  size.o bucomm.o version.o filemode.o ../bfd/libbfd.la ../libiberty/libiberty.a  -lz 
libtool: link: gcc -W -Wall -Wstrict-prototypes -Wmissing-prototypes -Werror -g -O2 -o size size.o bucomm.o version.o filemode.o  ../bfd/.libs/libbfd.a ../libiberty/libiberty.a -lz
../bfd/.libs/libbfd.a(pe-sh.o)(.rodata+0x488): undefined reference to `pe_print_ce_compressed_pdata'
../bfd/.libs/libbfd.a(pei-sh.o)(.rodata+0x488): undefined reference to `pe_print_ce_compressed_pdata'
../bfd/.libs/libbfd.a(coff-sh.o)(.rodata+0x488): undefined reference to `pe_print_ce_compressed_pdata'
../bfd/.libs/libbfd.a(coff-sh.o)(.rodata+0xc68): undefined reference to `pe_print_ce_compressed_pdata'
collect2: ld returned 1 exit status
make[4]: *** [size] Error 1
make[4]: Leaving directory `/home/hp/combined/binuo/binutils'

I suggest you revert your change and move
pe_print_ce_compressed_pdata out of pe-arm-wince.c (clearly the
wrong place) and into peXXigen.c or its own file (don't forget
to adjust the build machinery), making it the default of a
target-dependent hook, overridden for pe-mips or something.

brgds, H-P

Re: Your bfd changes broke sh sim

[Danny Backx]

On Sat, 2008-07-19 at 05:52 +0200, Hans-Peter Nilsson wrote:
> > From: Danny Backx <danny.backx@scarlet.be>
> > Date: Fri, 18 Jul 2008 20:12:39 +0200
> 
> > I've tested (and tested again) with
> > 	./configure --enable-targets=all
> 
> Looks like you didn't even try and build binutils *without*
> --enable-targets=all, because you broke e.g. sh-elf (and yes,
> sh-pe too).  For --target=sh-elf, make all:

I was told to check --enable-targets=all in addition to what I was doing
(several arm targets). I did that.

Here's what I just found (unless I overlooked something) :
- I argued with Nick about where this code needed to be, the initial
  place was peXXigen.c but I argued that this appeared to be the wrong
  place because it could get generated more than once.
- Then I was given a choice where to put it, evidently I made the wrong
  choice (pe-arm-wince.c).
- I just moved it back and tested with three builds :
  configure --enable-targets=all
  configure --target=arm-wince
  configure --target=sh-elf

  and found that only the last one failed to build, even with the code
  in peXXigen.c.
- Inspection reveals that peXXigen.c isn't compiled for sh-elf, so this
  was the wrong choice too.

I see three options :
- remove the line that causes this in coff-sh.c and change it as in all
  the other architectures. Then all this only works for ARM, not for ARM
  and SH as it is supposed to.
- remove all the smart configuration stuff that Nick talked me into and
  revert all this to a dumb ARM-only patch
- find a better place to store the code

Please advise.

	Danny

> I suggest you revert your change and move
> pe_print_ce_compressed_pdata out of pe-arm-wince.c (clearly the
> wrong place) and into peXXigen.c or its own file (don't forget
> to adjust the build machinery), making it the default of a
> target-dependent hook, overridden for pe-mips or something.
> 
> brgds, H-P
> 
-- 
Danny Backx ; danny.backx - at - scarlet.be ; http://danny.backx.info

Re: Your bfd changes broke sh sim

[Hans-Peter Nilsson]

> From: Danny Backx <danny.backx@scarlet.be>
> Date: Sat, 19 Jul 2008 07:52:27 +0200

> I see three options :
> - find a better place to store the code

Yeah, like I suggested below, the "its own file" alternative.
If you don't understand the target-dependent-hook part, skip it.

> > I suggest you revert your change and move
> > pe_print_ce_compressed_pdata out of pe-arm-wince.c (clearly the
> > wrong place) and into peXXigen.c or its own file (don't forget
> > to adjust the build machinery), making it the default of a
> > target-dependent hook, overridden for pe-mips or something.

brgds, H-P