On Thu, 01 Jun 2000, you wrote:
> Civil,
> In this case I really would like to help out with the bug fixing/devel
> cycle reporting bugs and possible fixes. When will the next ISO's be made
> available ( I am afraid of downloading/burning out dated ISOs only to
> report bugs that are fixed ).
>
> As far as the segate drives, I really had no idea there were issues with
> them. The interrupt thing I don't recall, but that doesn't mean it didn't
> happen, I could just be blocking it out :)
>
> How does linux handle newer drives? ( Maxtor DiamondMAX Plus 40 ), are
> those problems still prevelant?
>
> Best wishes,
>
> Riyad Kalla
> Java Programmer
> Game Enthusiast
>
Linuxcare in February carried info about the "Strange Timing Requirements of
WD drives" which spoke to hardware.
I have been having problems with Pentium code linux distros combined with VIA
MVP3 and MVP4 chipsets, SiS 530 Chipsets, Intel 810 Chipsets (even with
memory certified for the 810) and Intel TX chipsets in combination with
Seagate drives in the Primary Master IDE position. I have played with clocks
for the bus, the processor and so on without resolution. Finally with two
dual-trace scopes I was able to chase down a signal reflection. Use the IDE
driver in 386 code--no problem. 486 code (as in 7.0-486) the most occasional
error -- 586 code (as from 6.1 and 7.0) -- a frequent signal reflection on
one of the timing lines. I did not compile for 686 code but I found in
Linuxcare that the results are even more disturbing.
Actually the article suggested the following
386 code--no problem
486 code--occasional "lost interrupt"
586 code--frequenrt crashes
686 code--data will fry
but that was for WD drives which have their own special issues (What use is
S.M.A.R.T. when the drive logic blows off the CRC?). Also the issue was
originally discovered with a slow WD on the same channel as a Fast Maxtor.
I know of no issues affecting Maxtor drives. That doesn't mean there aren't
any. I know that WD should NEVER be combined on the same IDE channel with
Maxtor because the two can and will under certain configurations destroy each
other with timing chatter. I also know that I should stay away from Seagate
or put them in as slaves on the secondary channel if I have certain chipsets
and Pentium code.
The point is, with Stampede, Enoch, Mandrake, and possibly other distros
going stock to advanced processor coding, the hardware better be up to
specifications. Hardware manufacturers have had it easy--the relaxed
requirements for meeting 386 coding and a buggy OS that could take the blame
for many/most problems (Windows). Now, the jig is up. There is real
competition in the OS area, and some of that competition expects the hardware
to (gasp) MEET the specifications.
OK the newer drives--those I have tested seem OK and I don't have an
exhaustive lab here. My testing has been with Quantum and IBM IDE drives--I
don't buy Fujitsu, Maxtor, JTS, Seagate, or WD any longer because the stats
tell me to stay away from them (Calculated MTBFs much lower than
Quantum/IBM). I also know that the cheap little fans, which last 15-24
months before needing replacement, extend drive life incredibly, perhaps
tripling it.
And yes, Mandrake seems to have more newbies than most--the flash and pizazz
as well as the greater convenience of installation seems to draw them. I
worked with one who was upset because his eMachine had trouble with XFree86
(SiS 6326 Chipset) under Mandrake 6.1. Well, he was sure it was all
Mandrake's fault so he bought RH. SAME PROBLEM. He then decided Caldera had
to be easier.... SAME PROBLEM. So he came back to me and I said either a)
be willing to hack minimally on XF86Config or b) buy SuSE which had a better
driver. He DLed the driver from SuSE and helps other newbies with Mandrake
these days.
Civileme
