[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Device detection?



On Tue, 9 Mar 1999, Brandon Mitchell wrote:

> On Tue, 9 Mar 1999, Stephen Crowley wrote:
> 
> > On Tue, Mar 09, 1999 at 07:37:49PM +0100, Wichert Akkerman wrote:
> > > 
> > > No way, this will certainly crash lots of systems..
> > 
> > Of course, that's why I said it should only be used on devices which wont
> > lock up systems.
> 
> But how do you auto detect a device that locks up your system when you
> probe for it?  This is the reason for windows throwing you into safe mode,
> no auto detection because you probably crashed from a misguided auto
> detect.  I prefer linux's way of telling the kernel what you have.  But, I
> guess someone could make a package that just mod-probes the heck out of
> your system with every option until you crash or the modprobe succeeds.

While scanning for hardware, Windows95 keeps a log of all its scanning
activities in a file on disk (c:\detlog.txt, IIRC), for which it turns off
disk caching. That's why there is so much disk activity while Windows95 is
scanning for hardware: the file is growing rapidly and both the root
directory and the FATs have to be kept up-to-date after every write to the
file. If the machine locks up, the user hits the reset button and restarts
the installation. Setup detects a previous, failed attempt to install
Windows95 and asks if it should do a 'safe recovery'. If the user agrees,
it reads the log file to find at which point the computer locked up and
continues the scan, using the information in the log file to try to avoid
another lockup.

BTW, if you have the device you are probing for you are probably safe.
It's totally unrelated hardware that locks up the system. Probing for
hardware often involves writing data to several I/O addresses to see if
there is some kind of a response. If you are probing for a SCSI card at
I/O addresses which are occupied by a network card (or the other way
around), strange things can happen.

Remco
-- 
rd31-144: 11:30pm  up 5 days, 22:22,  8 users,  load average: 1.03, 1.07, 1.08


Reply to: