Subject: StrongARM K bug
To: None <port-arm32@netbsd.org>
From: Nicholas Clark <nick@flirble.org>
List: port-arm32
Date: 03/28/1999 22:02:51
I confess to lurking on this list.
You may be interested to know that following a discussion on the arm
linux list about working around the StrongARM revision K's LDM into
the pc bug, I was persuaded to have a go a writing a patch for gcc.

This I've done (for egcs 1.1.1ish, but the only changes are in the
arm backend), and posted it to the arm linux list.
I've tested it:

1: I'm now running on a linux kernel built with -mstrongarmk-fix
   which still hasn't crashed :-)
2: I rebuilt perl and the two regression tests that appeared to be
   failing with a SEGV due to the K bug now work perfectly.
   (suspect K bug because the problem only manifested itself in perl
   tests that fork()ed and then wrote to a pipe, and changing the order
   of symbols in the compiler assembler output, or adding/removing one
   instruction from the assembler made a perl that worked)

Is this of direct use to BSD?

Nicholas Clark