Hiya,

OK, this sounds absolutely crazy but I have to ask if anyone here has experienced a similar problem...

I recently built a new system containing an Athlon XP 2200 processor and A7V333 Socket A Motherboard with KT333 chipset. Well, I installed SoftICE and it seems to work great until I attempt to perform a backtrace using the bprw 'module name' T command. At this point 1 of 2 things happen... 1) It locks the entire system up irrecoverably 2) The trace simply fails reporting that the 'backtrace buffer is empty' when attempting a 'show' command. The internal winice variable containing the total number of instructions logged likewise remains at 0. More often, scenario 1 happens though Strangely, an ordinary range breakpoint works fine (bpr 'address' rw). Needless to say, I am running the same version of SoftICE I've used on Win98 for the past 2 years and the only thing that has changed is the system I am running it on. I am curious if my new motherboard or processor could have anything to do with this problem ??? Could it have something to do with the way it handles paging? And more importantly, how can I fix this...??? I am really at a loss here and will welcome any suggestions.

Thanks,
Clandestiny

Hi Clandestiny,

I was running Windows 98 on a 333 mhz system with 64 megs of ram and using Softice 3.25. The bpr I set would always lockup the system and the only way out was to re-boot.

I also ran the same version of windows and the same version of Softice on a 650 amd with 128 megs of ram and the same thing happened.

Because of this I've never used the bpr because I just figured it wasn't going to work.

I thought I heard someone say that this only happens when you use an amd processor but I'm not sure.

Pappy28

I got the same result when using BPR in Win98SE with AMD processor. DriverStudio v2.7

I vaguely remember some discussion about this subject on ElicZ's board.

I doubt I am going to tell you something that you don't know but. . .

In theory, the internal workings of a CPU should be transparent to the applications runing in them, thanks to a layer of interrupts and BIOS and other software devices that make the comunication between code and CPU abstract and formalized. Apps that code very close to the bare metal, like a Kernel level Debugger, may not be as protected and may run into trouble with the internal workings of the CPU.

One of the aspects in which AMD differs substantially from Pentium ids the PipeLine and Branch Prediction devices, which, if you consider carefully, could collide with the BPR T instruction. BUT I do not know that for a fact.

A possible solution is to find a registered user of SoftIce (There has to be at least one, don't you think?) tosubmit your question to Numega technical support, or, if any Numega coder reads in this board ( I would not be surprised), to kindly help Clandestiny with his problem.

Did you try, in Winice.dat turning

Pentium = off

And

Mapping non-exiting pages = Off?

Look in Using SoftIce.pdf, trouble shooting area.