This message is from the T13 list server.

> Pat, please, here we are talking about how the host side hardware or software 
>handles the possible difference between the size of the application I/O buffer in 
>memory and the amount of data that a device transfers on the I/O interface.

Yes.

An AtapiPio device can ask to copy In an odd count of bytes In, an AtapiDma device 
cannot.


> This is not a T13 PIO or DMA data transfer problem.

It is T13 that has denied AtapiDma devices the privilege of asking to copy In an odd 
count of bytes.


> Host software or hardware that rounds data transfers up or down to some 
>byte/word/dword boundary is not a T13 I/O interface problem.

T13 Atapi Pio protocol tells Microsoft (whoops, I mean the host software) when not to 
do this.  T13 Atapi Dma protocol doesn't.

How is that not a T13 problem?  Pio works, Dma doesn't.  Who but T13 cares about the 
distinction between Pio and Dma?


> I can clearly see now that you are confusing

Now that you feel you understand me, can you tell me again why I'm wrong to want Atapi 
Dma to be as capable of counting bytes as is Atapi Pio?


> ...

I'm not sure agree with the rest as worded, but I'd like to try deferring those 
seemingly deeper issues until we get my mis/understanding straight on whether AtapiDma 
devices can or cannot ask to copy In odd counts of bytes.

Pat LaVarre


>>> Hale Landis 04/14/02 12:38 PM >>>
This message is from the T13 list server.


On Sun, 14 Apr 2002 09:10:27 -0600, Pat LaVarre wrote:
>This message is from the T13 list server.

>"Residual byte length is the number of bytes not
>transferred to, or received from, the target SCSI
>device.

>And the examples:
>"For example, if the ASPI buffer length for a SCSI
>Inquiry command is set for 100 bytes, but the target
>only returns back 36 btes, this makes for a residual
>length of 64 bytes.

>"Another example, if the ASPI buffer length for a
>SCSI write command is set for 514 bytes, but the
>target only takes 512 bytes, this makes for a
>residual length of 2 bytes.

Pat, please, here we are talking about how the host side hardware or software handles 
the possible difference between the size of the application I/O buffer in memory and 
the amount of data that
a device transfers on the I/O interface.

The concept of the host side hardware or software reporting this value has been with 
us since the 1950's.  Unfortunately, the PCI bus ATA host adapters designers at Intel 
apparently don't understand the need to report this value.  That still does not make 
this a T13 problem, unless you want the T13 "host adapter standard" to address it?

This is not a T13 PIO or DMA data transfer problem.  An ATA or ATAPI device will 
attempt to transfer the amount of data as specified by the current command.  The 
interface works only because a device transfers that amount of data (or less for read 
commands) or the device signals an error on the command.

I can clearly see now that you are confusing the actions of host side software and 
hardware with how you think ATA/ATAPI PIO and DMA work.  Host software or hardware 
that rounds data transfers up or down to some byte/word/dword boundary is not a T13 
I/O 
interface problem.



*** Hale Landis *** www.ata-atapi.com ***

Reply via email to