[t10] deferred errors in host & device

Pat LaVarre LAVARRE at iomega.com
Wed May 14 09:20:22 PDT 2003


* From the T10 Reflector (t10 at t10.org), posted by:
* "Pat LaVarre" <LAVARRE at iomega.com>
*
> > The devices had
> > BETTER NOT ALLOW DEFERRED ERRORS.
>
> Hosts force devices to defer errors by forcing devices
> to defer operations, most egregiously by setting an 
> Immed bit, but also in other ways.

Offline kindly I'm reminded of hosts that require bus activity as a heartbeat.

These hosts see too much time between the last bit of Data Out and the first bit of Status In as a failure.

That host design for timeouts by definition requires write-behind in the device.  Because most devices found in the development of a massively distributed host did write-behind, now all devices that follow must also write-behind.

Last I checked, the host folk had adopted no public standard for timeouts, which leads to this kind of creativity in hosts forcing devices to write-behind.

> ...

Offline kindly I'm also reminded of hosts that apply PDT x00 HDD timeouts to PDT x05 DVD/CD devices.

With hosts like that massively distributed, a polite policy against write-behind in HDD that caused only an acceptable degradation in speed may turn out to be impractical for DVD/CD.

Pat LaVarre

*
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org




More information about the T10 mailing list