For what it's worth, I use the -u1 flag on both my DVD-ROM and my
DVD-RW. Why? Well, it was driving me nuts that anytime I would copy
large amounts of data from the DVD-ROM or burn a DVD-R that the data
rate on my modem went to almost zero. Turns out that since the COM port
interrupts aren't considered as "important" as the IDE interrupts, that
the modem was simply getting no "interrupt" time. Or at least this is
how I understood what I could find via google. Unmasking the irq on both
devices completely fixed the problem. Occassionally when I switch
kernels, I think about checking to see if the setting is still required,
but I haven't tried it recently.

Just my 2c.

Sean

On Sat, 2003-11-01 at 16:57, Hemmann, Volker Armin wrote:
> On Saturday 01 November 2003 21:17, Dennis Freise wrote:
> > On Thu, 30 Oct 2003 19:13:22 +0100
> > "Hemmann, Volker Armin" <[EMAIL PROTECTED]> wrote:
> >
> > Hi Volker,
> >
> > > hdparm -d1 -c1 -u1 should be save.
> >
> > Are you sure that -u1 is safe? I've read the hdparm man page,
> > but I (obviously) didn't understand it in full - is -u1 safe for
> > _all_ chipsets now (got some sort of CMD640 chipset...) ?
> > If it is, what can I expect from it concerning speedup ?
> 
> as man hdparm says:
>  -u     Get/set interrupt-unmask flag for the drive.   A  setting  of  1
>               permits  the driver to unmask other interrupts during processing
>               of a disk interrupt, which greatly improves Linux's  responsive-
>               ness and eliminates "serial port overrun" errors.
> 
> the chipsets mentioned there are from the stone age of ide-controllers. If you 
> have one of them, you do not need to worry about dma etc ;o) And man hdparm 
> further says, that the kernels 2.0.13 and later incorporate an apropriate bug 
> fix. If you do not believe man, don't use it.
> 
> I am using this options since ages without probs but with other chipsets. (SiS 
> 735, 746, hpt366 ,  via kt133 (I do not know the southbridge anymore, AFAIR 
> an udma 66 type.))
> 
> Gl�ck Auf
> Volker


--
[EMAIL PROTECTED] mailing list

Reply via email to