Broken mips-wince-pe, mips-pe ?

[Karl Trygve Kalleberg]

Hi.

Are the mips-wince-pe and mips-pe targets out of order in binutils-2.10
and current cvs ?

(I've noticed that there has been some debate in whether 'wince' should be
in, replace 'pe' or just be dropped altogether. I've tried mips-wince, but
that choked the configure. mips-pe and mips-wince-pe make the tools
compile, but with the same problem, as described below).

I've tried

../binutils-2.10/configure --target=mips-wince-pe --prefix=/usr/local/mips
and
../binutils-2.10/configure --target=mips-pe --prefix=/usr/local/mips

(after having added peigen.lo to bfd_backends and BFD_BACKENDS in
bfd/Makefile).

The resulting mips-{wince-}pe-ld has a few problems. 

It reports 

/usr/local/mips/bin/mips-pe-ld:built in linker script:4:
zsh:segmentation fault /usr/local/mips/bin/mips-pe-ld


When I compile from yesterday's CVS, the segmentation fault disappears,
but the linker script error persists:

/usr/local/mips/bin/mips-wince-pe-ld:built in linker script:4: parse error

strace shows that the only place where the binary tries to open the linker
scripts is in a stat:
stat("/usr/local/mips/mips-wince-pe/lib/ldscripts",
{st_mode=S_IFDIR|S_ISGID|0755, st_size=1024, ...}) = 0


Compiling the sh-wince-pe target results in working binutils, and I've
managed (with no major problems) to compile a sh-wince-pe version of gcc
2.95.2 afterwards.

The host is a GNU/Linux Debian x86 SMP machine. The host compiler is
gcc 2.95.2.


Should I file a bug report on this with proper stack traces, or am I
simply doing something wrong ?



Regards,

Karl T