home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.aix
- Path: sparky!uunet!usc!cs.utexas.edu!geraldo.cc.utexas.edu!portal.austin.ibm.com!awdprime.austin.ibm.com!konopik.austin.ibm.com!konopik
- From: konopik@konopik.austin.ibm.com (Brad Konopik)
- Subject: Re: Micro Channel in RS/6000
- Sender: news@austin.ibm.com (News id)
- Message-ID: <BzDxK1.F12@austin.ibm.com>
- Date: Thu, 17 Dec 1992 03:55:12 GMT
- References: <1992Dec8.210721.15669@oakhill.sps.mot.com>
- Organization: IBM AIX Porting Center, Austin
- Keywords: Which Endian?
- Lines: 46
-
- In article <1992Dec8.210721.15669@oakhill.sps.mot.com> mack@oakhill.sps.mot.com (Gregg Mack ( Sun )) writes:
- >I am designing a memory-mapped I/O adapter to be put into a Micro
- >Channel slot in an IBM RS/6000 model 350 workstation. After reading
- >the IBM document SA23-2647-00 "Hardware Technical Reference Micro
- >Channel Architecture", I am still uncertain if Big Endian or Little
- >Endian addressing is used. Is this not part of the specification?
-
- Byte 0 is the LSB. Byte 31 is the MSB. Addressing is addressed in
- the Hardware Technical Reference General Information Manual
- (SA23-2643-00).
-
- >I know that the Micro Channel spec states that D(31) is the most
- >significant bit of the 32-bit data bus, and this is part of the
- >byte accessed when -BE(3) is asserted. Do the byte enable numbers
- >- -BE(0-3) correspond to the byte addresses?
-
- See the MC spec (document # SA23-2647-00), pg. 1-40, Note 4.
- BE0 is D(0-7), BE3 is D(24-31).
-
- >If so, that would
- >make -BE(0) = D(0-7) to reside at address 0 and -BE(3) = D(24-31)
- >reside at address 3. Now if a programmer were to read a 2-byte
- >datum at address 1, would the data be driven onto D(8-23) or D(0-15)?
- >Which of these two byte would be the most significant of the data?
-
- See MC spec pg. 1-40 and 1-41. Data would be driven on 8-23.
- -BE1 and -BE2 would be active (low). Also, D(8-15) would be
- replicated on D(0-7).
-
- >As a second question: The 8 POS registers are each 1-byte wide.
- >All configuration data transfers are single-byte transfers. Since
- >I am designing a 32-bit slave, I only monitor -BE(0-3) (and do not
- >monitor -SHBE, A1, and A0). Which of -BE(0-3) is asserted when
- >POS register 0 is being accessed?
-
- Again see MC spec pg. 1-40. For single-byte transfers, the
- appropriate -BE to use is address-mod-4. Thus, POS 0 would be
- -BE0 (assuming POS 0 is addressed xxx0).
-
- (The above answers are courtesy of some engineering contacts)
-
-
- --
- internet: konopik@austin.ibm.com | Brad Konopik
- uunet: uunet!austin.ibm.com!konopik | IBM AIX Porting Center
- ibmvnet: KONOPIK at AUSTIN | Austin, Texas
-