RE: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

RE: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

[David Brodbeck]

> -----Original Message-----
> From: joe briggs [mailto:jbriggs@briggsmedia.com]

> I am glad that you mentioned ext3 because while I curse 
> ReiserFS, I really 
> don't think that it is part of the problem. Definately 
> PCI-dma related, but 
> does onboard IDE (i.e., my system disk) use DMA in the same 
> way that a PCI 
> adapter such as Promise does? 

I'm not a hardware expert, but since on-board IDE controllers usually appear
as PCI devices to the OS, I assume they do DMA the same way.

Re: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

[joe briggs]

Is this true??? Does onboard IDE controllers appear and work the same way as 
IDE or PCI controllers with respect to IRQ, DMA, and memory access?

On Wednesday 23 April 2003 03:22 pm, David Brodbeck wrote:
> > -----Original Message-----
> > From: joe briggs [mailto:jbriggs@briggsmedia.com]
> >
> > I am glad that you mentioned ext3 because while I curse
> > ReiserFS, I really
> > don't think that it is part of the problem. Definately
> > PCI-dma related, but
> > does onboard IDE (i.e., my system disk) use DMA in the same
> > way that a PCI
> > adapter such as Promise does?
>
> I'm not a hardware expert, but since on-board IDE controllers usually
> appear as PCI devices to the OS, I assume they do DMA the same way.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

-- 
Joe Briggs
Briggs Media Systems
105 Burnsen Ave.
Manchester NH 01304 USA
TEL/FAX 603-232-3115 MOBILE 603-493-2386
www.briggsmedia.com

Re: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

[Alan Cox]

On Mer, 2003-04-23 at 23:13, joe briggs wrote:
> Is this true??? Does onboard IDE controllers appear and work the same way as 
> IDE or PCI controllers with respect to IRQ, DMA, and memory access?

They appear to. Most of them nowdays are wired directly to the
north/southbridge link for performance.

Re: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

[joe briggs]

That's what I thought, which is why I suspected that the underlying problem 
might be related to the scope and completeness of the hardware arbitration.  
Specifically, if a PCI board is DMA'ing a block of data to main memory, how 
is the operation of the direct-connect IDE device impacted?  When most PCI 
devices transfer data to main memory, don't they arbitrate the bus, become 
master, and blast it accross in chunks?  Doesn't onboard IDE (legacy?) use 
the DMA controller (8237 derivate I gues) to transfer?

On Wednesday 23 April 2003 06:11 pm, Alan Cox wrote:
> On Mer, 2003-04-23 at 23:13, joe briggs wrote:
> > Is this true??? Does onboard IDE controllers appear and work the same way
> > as IDE or PCI controllers with respect to IRQ, DMA, and memory access?
>
> They appear to. Most of them nowdays are wired directly to the
> north/southbridge link for performance.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

-- 
Joe Briggs
Briggs Media Systems
105 Burnsen Ave.
Manchester NH 01304 USA
TEL/FAX 603-232-3115 MOBILE 603-493-2386
www.briggsmedia.com

Re: [Motion] Re: IDE corruption during heavy bt878-induced interr upt load [LKM]

[Alan Cox]

On Iau, 2003-04-24 at 11:31, joe briggs wrote:
> Specifically, if a PCI board is DMA'ing a block of data to main memory, how 
> is the operation of the direct-connect IDE device impacted?  When most PCI 
> devices transfer data to main memory, don't they arbitrate the bus, become 
> master, and blast it accross in chunks?  Doesn't onboard IDE (legacy?) use 
> the DMA controller (8237 derivate I gues) to transfer?

Onboard legacy mode IDE doing PCI DMA is doing it directly. How onboard
and southbridge IDE interact is a complicated field and chipset
dependant.