[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Simulavr-devel] ATMega48 Crash
From: |
Knut Schwichtenberg |
Subject: |
Re: [Simulavr-devel] ATMega48 Crash |
Date: |
Mon, 13 Apr 2009 20:17:10 +0200 |
User-agent: |
Thunderbird 2.0.0.19 (X11/20081227) |
Michael N. Moran wrote:
> I don't see a connection. For example the HWMegaSpi member
> in the M128 and M48 are both pointers.
That's true. And yes it is definitely not related to this.
> I did not glean anything useful from the trace, though
> I have set my share of breakpoints in the AddToCycleList
> operation over the last few days :)
Okay, as you read from my error message I think I found the problem. I guess the
following: SPI-subsystem was tested with the at90s-CPUs which use the HwSpi
constructor. The M48 uses the HwMegaSpi. Currently I don't know the difference
in Hardware. In the simulator there is one. HwMegaSpi calls
core->AddToCycleList(this) while in HwSpi this call is commented out.
IMHO the CycleList contains all HW-subsystems that need an activity during a
cycle (not sure if it is a microcycle of 1ns or if it is a clock cycle). Anyway,
if the SPI is only added if the SPE bit is set it is process only than and that
brings performance. Therefore I expect the additional core->... is forgotten to
comment out.
But that does not explain while the spi-example does not generate this error.
:-(
Knut