Alignment problem on OSF (egcs 2.91.57)

[Lars Hecking]

 Ghostscript 5.50, compiled with egcs 2.91.57 on alpha-dec-osf4.0b,
 produces the following output:

Unaligned access pid=8926 <gs> va=0x140128cba pc=0x1200c9c24 ra=0x1200c90b0 inst=0xa0290000
Unaligned access pid=8926 <gs> va=0x140128cba pc=0x1200c8c24 ra=0x1200c8be4 inst=0xa0290000
Unaligned access pid=8926 <gs> va=0x140128cbe pc=0x1200c85a8 ra=0x1200c8be4 inst=0xa0290000
Unaligned access pid=8926 <gs> va=0x140128cbe pc=0x1200c9b7c ra=0x1200c8be4 inst=0xa0290000
Unaligned access pid=8926 <gs> va=0x140128cbe pc=0x1200c9c24 ra=0x1200c8be4 inst=0xa0290000
Unaligned access pid=8926 <gs> va=0x140128cba pc=0x1200c85a8 ra=0x1200c90b0 inst=0xa0290000
[ad infinitum]

 It just keeps running, no core dummps, and I have no idea how to trace
 this with gdb (single-stepping a couple of 10k lines through the code 
 didn't produce much).

 Peter Deutsch, the ghostscript author, reckons that egcs-1.1 makes
 different assumptions about data alignment than 1.0.3, and is unable
 to help with no further information.

 [Btw, I'm not on the list.]

-- 
"I just need enough to tide me over until I need more."
                -- Bill Hoest