home *** CD-ROM | disk | FTP | other *** search
- Path: watnews.watson.ibm.com!kgold
- From: kgold@watson.ibm.com (K Goldman)
- Newsgroups: comp.sys.m68k
- Subject: Re: 68k memory test in assembly
- Date: 2 Apr 1996 18:17:44 GMT
- Organization: IBM T.J. Watson Research Center
- Distribution: ibm
- Message-ID: <4jrr08$cnp@watnews1.watson.ibm.com>
- References: <jharkinsDp7JMt.DCx@netcom.com>
- NNTP-Posting-Host: ovni.watson.ibm.com
-
- jharkins@netcom.com (Jim Harkins) writes:
- |> I need a RAM test for a 68360. It needs to be in assembly as it will
- |> run before the C environment is started. Does anyone have one they can
- |> send me, or am I going to need to convert an 80x86 ramtest?
- |>
-
- 1 - I can first tell you that all my 68360 diagnostics are in C. That
- includes ROM, RAM, and peripherals. There's very little (several
- registers, chips selects, and the stack pointer) that I set up in
- assembler.
-
- 2 - The RAM test I usually use is to run two write-read-verify loops.
-
- The first starts at adrs 0 (in the RAM under test) and data 0
- and writes incrementing data to incrementing addresses. After
- writes, it reads to verify.
-
- The next test does the same thing but starts with data all ones and
- decrements.
-
- This is simple to code and fast to run. Each loop is about 5 lines of
- C code. It catches all stuck address and data lines, all shorted
- address lines, many shorted data lines, and writes all RAM bits to
- both 0 and 1.
-
- You can catch the rest of the shorted data lines if you inc/dec
- by the constant 1 + (1<<n) where n is the number of address bits.
-
- Don't forget to service the watchdog timer occasionally!
-
- --
- Ken Goldman kgold@watson.ibm.com 914-784-7396
-