Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!sun-barr!olivea!uunet!fernwood!uupsi!sunic!sics.se!ifi.uio.no!runene From: runene@ifi.uio.no (Rune Nerg}rd) Newsgroups: comp.sys.nsc.32k Subject: gas-1.38 for NS32K (and gcc-1.40) Message-ID: Date: 21 Jun 91 14:11:06 GMT Sender: runene@ifi.uio.no (Rune Nerg}rd) Organization: Dept. of Informatics, University of Oslo, Norway Lines: 42 Nntp-Posting-Host: lyberga.ifi.uio.no Originator: runene@lyberga I'm sitting trying to kludge together a cross-compiling system for a embedded system using a NS32GX32 processor. Right now i'm looking at the gnu assembler, and wandering about some (for me) strange things. If anybody has seen these problems/anomalities before and found any explanations/solutions I would be very glad if you could enlighten a lost soul. It is gas-1.38 if it makes any difference. 1: Why does gas incorrectly assemble "br .+10" to 0x1a0f and not to 0x1a0a ?? 2: Why is adressing mode 20 default for labels, when gcc (v. 1.40) outputs movd _integer1,tos jsr _part2 .comm _integer1,4 for int integer1; part2(integer1) and that is assembled into what whould be the same as "addr @_integer1,tos" which is not what gcc ment. Should gcc output movd @_integer1,tos or should gas be changed to accept gcc output?? (or is that a matter of taste ??) And here is one about gcc (-1.40): 1: Why does gcc output "addr @255,_integer1" instead of "movd 255,_integer1" ?? Probably for speed, I could not find it out from the tables for ns32032 and ns32332, but for ns32532 and ns32GX32 the latter is faster, I think. Has anybody used the gnu tools for a embedded system using a NS32K processor? If you have (and even if you haven't) and have any advice, I would be glad to hear from you :) Rune N. "We are sorry, but the number you have dialed is imaginary." "Please rotate your phone 90 degrees and try again." ------ Rune Nergaard runene@ifi.uio.no (Internet) Institutt For Informatikk ..!uunet!ifi!runene (UUCP) University of Oslo, Norway