On Apr 2, 2007, at 2:02 AM, Johnny Billquist wrote:

> Matt Thomas wrote:
>> I just committed about 3 weeks of testing and development of  
>> getting GCC to produce PIC code for VAX and greatly improved 64bit  
>> math support.
>
> Hmm. Problems. Trying to build -current as of yesterday, cross- 
> compiling from i386, I got stuck at libedit.
>
> Two problems:
>
> 1) Lots of "/usr/src/lib/libedit/chared.c(68): warning: conversion  
> from 'long' to 'int' may lose accuracy [132]" which seems silly, as  
> I would think long and int are the same size.
>
> 2) Which is more serious:
> #   compile  libedit/editline.o
> /usr/src/obj/tooldir.NetBSD-3.99.8-i386/bin/vax--netbsdelf-gcc -O2 - 
> Wall -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Wno- 
> sign-compare -Wno-traditional -Wreturn-type -Wswitch -Wshadow - 
> Wcast-qual -Wwrite-strings -Wextra -Wno-unused-parameter  -Werror - 
> I. -I/usr/src/lib/libedit -I. -I/usr/src/lib/libedit     -nostdinc - 
> isystem /usr/src/obj/destdir.vax/usr/include -c    editline.c -o  
> editline.o
> /usr/src/lib/libedit/tty.c: In function 'tty_stty':
> /usr/src/lib/libedit/tty.c:1287: error: unable to generate reloads  
> for:
> (insn:QI 478 476 479 69 (parallel [
>             (set (reg/v:SI 30 [ c.5528 ])
>                 (ffs:SI (reg:SI 7 %r7 [orig:27 temp.5532 ] [27])))
>             (set (cc0)
>                 (reg/v:SI 30 [ c.5528 ]))
>         ]) 168 {ffssi2_internal} (nil)
>     (expr_list:REG_DEAD (reg:SI 7 %r7 [orig:27 temp.5532 ] [27])
>         (nil)))
> /usr/src/lib/libedit/tty.c:1287: internal compiler error: in  
> find_reloads, at reload.c:3738
> Please submit a full bug report,
> with preprocessed source if appropriate.
>
> Since you probably knows better what's going on, Matt, perhaps you  
> can look at it?

I committed a fix for this.