>Ok I picked roughly similar numbers. Can you tell me where you fished the
21
>out from and why (just curious)
>
>I'll probably change the 2.2.15 ones to match the SuSE ones. Mine are
similar
>values and we might as well just have one default size of extra devices
>
>Alan
Maybe(?) there is another approach... Please forgive my naivet�.
The real problem is that fact that anything not serviced by the boot SCSI
driver is subject to the SD_EXTRA_DEVS limit. For example, if my box has
an
Adaptec on the motherboard (w/ boot device attached), anything on the
Symbios/Qlogic/XX adapters would be subject to SD_EXTRA_DEVS.
What kind of code gymnastics would be required to scan _all_ the SCSI
drivers installed before saying "this is all the devices we found"?
Extend the initial scan to include the buses attached to additional
adapters.
With that kind of function, users would still have to reboot to config all
devices, but its a start. Then SD_EXTRA_DEVS would apply only to those
"extra" devices added between reboots...
In any event, just about any number you choose is going to be too small as
the momentum builds towards FC and Linux clusters. Personally I'd like to
boot off of my internal SCSI bus and be able to access at least one full
FC-AL, but SD_EXTRA_DEVS = 128 would probably cause memory prices to rise.
Another blue sky idea: Could SD_EXTRA_DEVS be added SCSI lilo bootparms?
Users already have to add a 'max_scsi_luns' statement to lilo.conf to
access serious storage.
Again, I'm not intimately familiar with the SCSI drivers (yet), so please
humor the ramblings of the latest idiot in the village.
Regards,
Steve
-
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to [EMAIL PROTECTED]
