home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
OS/2 Shareware BBS: 10 Tools
/
10-Tools.zip
/
waithang.zip
/
MAT015.NTE
< prev
next >
Wrap
Text File
|
2002-07-16
|
3KB
|
49 lines
******************************************************************************
Subject: MAT015 problem
In talking with the debugger about the fix that was made for the system he
had in debug analysis, the specific wire added would be unique for that
system model only. For other manufacturers that are having the MAT015 HANG
problem, the fix would depend on their systems and the way they attached
the processor chip, and how the NPX lines were connected.
The 486 spec that the debugger had indicated that the C14 pin was defined
as the floating point error line on the chip. The BUS controller NPX error
line is supposed to be connected to the chip socket for the C14 pin.
In that system model, the BUS Controller NPX error line was connected to
the A13 pin. A fix was applied as described by hardware developer to have
a wire added from pin A13 to pin C14. The MAT015 test passed when the
system was tried with the hardware fix in place. The full compatability
testcase bucket was run and all tests passed.
Other system manufacturers would have to look at their boards to determine
what type of fix would be required for their systems. There is a program
that was written by the debugger that was used by to debug the problem with
the system model called WAITHANG. This has only enough code to re-create
the MAT015 problem, with an INT3 added for debugging. There is also a
WAITHANG2 with is the same without the INT3. I will send these to you
along with the source code.
If they are getting a MAT015 error of "bad indicator" the same WAITHANG program
can be used to debug the system board similar to the "hang" problem. In the
code, after the floating point store from BP+10, check BP+10, it should be
080009455373A92F4 (really small floating point negative number).
On a system that gets the MAT015 "bad indicator" message, the value of BP+10
is a floating point positive 1 instead of the REALLY REALLY small negative
number that it should have been.
******************************************************************************
Subject: MAT015
Once again, a manufacturer has called inquiring where we are on the
problem he is having with this test case. What I found out today is that
the test does not appear to be hung or failing. The log file is supposed to
have 3 lines of output but in this case, it only has the first line. The
system accepts either a 486CPU or a Pentium 60/66. If he puts a 486 CPU
in it, MAT015 runs fine and creates the correct log file. If he puts a
Pentiun 60 or 66 CPU in it, MAT015 only has one line in it but the overall
SNFMVDM tests run to completion with no indication of failures.
******************************************************************************