Re: linux interrupt handling problem

Jes Sorensen (Jes.Sorensen@cern.ch)
10 Nov 1999 09:45:13 +0100


>>>>> "Roman" == Roman Zippel <zippel@fh-brandenburg.de> writes:

Roman> Hi, On 9 Nov 1999, Jes Sorensen wrote:

>> We already have this feature, you run the critical interrupt as
>> SA_INTERRUPT, the rest as normal sti() ones. The 7 priorities don't
>> buy you anything.

Roman> We never had that feature in a working version. And even if it
Roman> would work you would limit yourself to only two interrupt
Roman> sources, for everything more advanced you can forget about good
Roman> perfomance.

You don't need any more sources that that, all it does is to add
complexity and hide what is really going on.

Roman> Another reason is that the driver writer had to decide how
Roman> "important" his driver is and set the SA_INTERRUPT for this. As
Roman> soon as you want to get more than one driver working properly
Roman> together, you get problem with this system.

Rubbish, the current case we are talking about is serial performance
vs the rest of the system. Run the serial as high priority and the
rest as slow - you don't need this for mouse, disk, ethernet etc.

Jes

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/