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

Re: Kernel 2.4.14 slow due to softirq daemon?

On Thu, 20 Dec 2001, Roger Larsson wrote:
> > I found out what causes the slowdown - it is the high
> > frequency (15360 Hz) TIMER1 interrupt, used in serial port driver. As soon
> I think there is a new IRQ frequency limiter in newer kernels.
> (author Ingo Molnar
>  "[announce] [patch] limiting IRQ load, irq-rewrite-2.4.11-B5")
> This is to protect from unclearable IRQs from a defective unit.
> (the high frequency seams close to the limit)

But there is nothing defective about it, it's a high frequency timer by 

You shouldn't use that config option (the flush_dma_fast). It's a "hack" 
for improving throughput in dual-way legacy serial port applications (like 
bluetooth) where you need to check for incoming bytes faster than the DMA 
hardware flushes them to memory. By necessity a timeconsuming process.. 
just like Mikael also wrote.

I think the code for the flush_dma_fast can be made less CPU intensive, 
but there should be almost no reason to use it at all anyway.