h8300-coff vs h8300-rtems

h8300-coff vs h8300-rtems

[Joel Sherrill]

HI,

I have found the discrepancy between h8300-coff and h8300-rtems
with regards to Fortran.  The question is how best to resolve it.
In the top level configure.in, there is this stanza around line

*-*-rtems*)
    noconfigdirs="$noconfigdirs target-libgloss ${libgcj}"

and this one around line 766

 h8300*-*-* | \
  h8500-*-*)
    noconfigdirs="$noconfigdirs ${libstdcxx_version} target-librx
target-libgloss ${libgcj} target-libf2c"

so *-*-rtems is being tripped before the CPU one.  Thus it tries
to build target-libf2c which is not supported.  Should I add a
stanza above *-*-rtems for h8300*-*-rtems* to cover this? Or
add a nested case statement in *-*-rtems* to further eliminate
target-libf2c?

Either way, this should be a minor RTEMS specific fix which I 
would like to see in 3.1.

Thanks.


-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel@OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985

Re: h8300-coff vs h8300-rtems

[law]

In message <3CC8319F.374F3EFE@OARcorp.com>, Joel Sherrill writes:
 > 
 > HI,
 > 
 > I have found the discrepancy between h8300-coff and h8300-rtems
 > with regards to Fortran.  The question is how best to resolve it.
 > In the top level configure.in, there is this stanza around line
 > 
 > *-*-rtems*)
 >     noconfigdirs="$noconfigdirs target-libgloss ${libgcj}"
 > 
 > and this one around line 766
 > 
 >  h8300*-*-* | \
 >   h8500-*-*)
 >     noconfigdirs="$noconfigdirs ${libstdcxx_version} target-librx
 > target-libgloss ${libgcj} target-libf2c"
 > 
 > so *-*-rtems is being tripped before the CPU one.  Thus it tries
 > to build target-libf2c which is not supported.  Should I add a
 > stanza above *-*-rtems for h8300*-*-rtems* to cover this? Or
 > add a nested case statement in *-*-rtems* to further eliminate
 > target-libf2c?
 > 
 > Either way, this should be a minor RTEMS specific fix which I 
 > would like to see in 3.1.
I have no strong opinion either way.

Note that for the H8 you'll also want to prevent it from building the
C++ runtime libraries.

jeff