On Sun, 13 Jun 1999 [EMAIL PROTECTED] wrote:
> Windows 3 uses the BIOS.  I don't know about the "or greater" windoze.
> I have never used any.

        I used I guess every windoze version from 3.0 to the latest
95.  I refused to install 98... political decision related with that
netscape trial.  And I'm not a pro in this area but the way I see it
m$ uses to send a lozy update as a 'new improved system'.  I checked
different versions (builds) of 95 and mostly there are new drives
sent as a new os.  osr2 has some dlls from that 4.0.950 or even from
95 beta.
 
> >     And here are the numbers:
> >     - first I redid that autodetection from bios.  And I faced
> > three options:
> >
> opt   size    cyls    head    precomp lands   sect    mode
> >
> 1(y)  8447    1027    255     0       16382   63      LBA
> ^^^^ I don't like nor trust this.  I could swallow 256 heads, but 255
> make me gag.  The hardware registers allow for 16 heads and 64k
> cylinders and 255 sectors per track (starting with 1).  Or 262144k
> sectors in LBA mode.  (That's right, 128g bytes.)  The BIOS calling
> sequence provides for 1024 cylinders, 64 heads, and 63 sectors.  Okay,
> the top 2 bits of head # are undefined, so it could go to 256.  But why
> use only 255?  That just makes the mapping arithmatic harder.  If there
> is a defined LBA mode for the BIOS I don't know of it.  More likely
> there is a private understanding between hardware manufacturers and
> Micro$oft.  There is also an unofficial standard that uses the top 2
> bits of the head number for cylinder number bits 10 and 11, for 4096
> "cylinders".

        I don't know what to say. :(
 
> >     The secondary master is my old drive which works perfectly.
> > Here are some things that puzzle me.  So the first field of the
> > description should be that LBA/normal/large thing.  What is CHS?
> > Cyls/Heads/Sectors?  Than the second field.  I'm guessing Mode 3
> 
> ^^^^^^^^^^^^^^^^^^  yep. :-)

        But why does the string CHS apeared in the same field as
LBA?  CHS should be followed by a drive description.  While LBA is
the description.  It's like mixing apples with the apple juice.
 
> > means the third PIO mode.  There are 4 PIO modes standard and a few
> > others.  UDMA 2 is related with PIO?  Sizes are accurate.
> I don't know what the U is.  DMA is direct memory access, meaning the

        U means ultra.  I check it out some time ago.

> channel does the IO to a block of physical memory.  In PC's, it is used
> mostly for the floppy(!).  With PIO, all data goes to and from the hd 
> through the CPU.  (This is why a good modem has a fifo :-). 

        UDMA... it's at 33 or at 66.  But never saw 2.  Maybe this
means UDMA/66.  Not sure.  I'm just guessing.
 
> >     -third and last this is what Linux autodetects... it can be 
> > seen with dmesg.
> > hda: <the drive id which seems ok> 8063MB w/416kB Cache, LBA,
> > CHS=16383/16/63
> This is what linux gets by asking the drive what it is without using the
> BIOS.  There is an identify drive command defined in the ANSII spec for
> ATA (IDE) drives.  LBA here means that the drive is capable of using it,
> and if it sees bit 6 of the drive/head register set on any IO, it will
> interpret the other registers as a zero-based block addres rather than
> CHS.  8063 MB, where 1 MB = 1048576 bytes, not 1000000.

        I know that.  But I also knew that bios did the same
(1m=2^20octets)
  
> >     How come the size is 8063?  None of the options were like
> > that.  I don't know about the cache.  But my 'normal' drive is now
> > LBA?  And at the same time the CHS are correct.  Afaik in case the
> > drive is seen uncorrectly I can define the C/H/S.  But nothing more.
> If the drive is capable of the hardware LBA, linux will use it

        Makes sense.  So it's not saying the drive is in LBA but
that it can be.

> By now you should have dosfsck, and bpe.  I still don't trust the
> partition boot record in hda1.

        Yes.  I'll try them tomorrow night I think
 
> I see your old drive is on hdc.  If you have a CMD640 ide interface chip
> that the IDE driver doesn't detect, this may scramble data on both
> drives.  If you don't suffer from a CMD640, of course it's better to
> have the drives on different channels.

        No.  The other drive never had a problem.  And this is
another disturbing fact.  So the old drive is a 560M... 535M
actually samsung drive.  I can consider it a no name.  And still it
is about 5 years old.  And in this 5 years it suffered quite a lot.
I used it as removable media in the freezing winter or in summer.  I
did a lot of repartitioning.  I formated it a lot.  And there is no
problem, besided that lost clusters that appear in dos... that is a
system fault, not hardware related.  At the same time my 8g quantum
with so many copyrighted standards written on its wrapping gives me
a headache.
 
> Lawson

        Raider
--
                ``Liberate tu-temet ex inferis''

Reply via email to