| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 8652.1 | Doesn't anybody know how to pick up a book anymore | NABETH::alan | Dr. File System's Home for Wayward Inodes. | Thu Jan 30 1997 15:23 | 17 |
| Yes.
Probably.
Don't know.
Yes - setld.
Yes.
See Also: SCSI(7), re(7), ra(7)
Currently, for SCSI disks, the bus and target number
are encoded in the name using "/dev/[r]rz#[a-h]" where
'#' is (bus * 8) + target-id. This will be changing
in a future release in a way that will probably make
it even more obvious what the associations are.
|
| 8652.2 | | VAXCPU::michaud | Jeff Michaud - ObjectBroker | Thu Jan 30 1997 15:28 | 12 |
| > Someone at Xerox corporation (investigating the port of some
> print software from HP-UX to Digital UNIX) ....
>> Q: Can disks be accessed for open/read/write as 'raw' devices with a
>> custom directory format?
>>
>> Q: Please provide information about the disk device naming convention,
>> how disk devices are created, and if the disk names can be
>> 'lexically' mapped to the controller ID and SCSI ID values?
Those are a couple of strange questions for someone porting
"print software". Unless they want to use an Alpha and Digital
UNIX embedded in a print server?
|
| 8652.3 | | ILLUSN::SORNSON | Are all your pets called 'Eric'? | Thu Jan 30 1997 16:07 | 9 |
| re 8652.2 by VAXCPU::michaud
> Those are a couple of strange questions for someone porting
> "print software". Unless they want to use an Alpha and Digital
> UNIX embedded in a print server?
The print utility on the host puts its data on a SCSI disk shared by
the UNIX host and the printer (these are high-speed printers that use
the shared disk to make sure the print-engine is fed at high-speed).
|
| 8652.4 | Are there more fundamental questions they should ask? | WIBBIN::NOYCE | Pulling weeds, pickin' stones | Thu Jan 30 1997 16:38 | 4 |
| Is the SCSI bus shared too, or is it a special disk
that has a SCSI port to the UNIX system and a separate
port to the printer? SCSI shared between multiple hosts
has some interesting ramifications...
|
| 8652.5 | | SMURF::DENHAM | Digital UNIX Kernel | Thu Jan 30 1997 17:15 | 9 |
| > Q: Are signals (such as SIGUSR1 and SIGTERM) reliable as specified
> by SVID, POSIX, XPG, or other standard?
Yes. All those. The POSIX's we comply with re: signals are
1003.1, 1003.1b (realtime), 1003.1c (threads).
> Q: Can shared memory blocks be locked in main memory?
Yes. With the POSIX 1003.1b mlock() function[A
|
| 8652.6 | | XIRTLU::schott | Eric R. Schott USG Product Management | Fri Jan 31 1997 14:19 | 25 |
| > Q: Can disks be accessed for open/read/write as 'raw' devices with a
> custom directory format?
yup, with appropriate privs...they should ensure they test this both
with raw disks, and raw lsm volumes...see below
>
> Q: Please provide information about the disk device naming convention,
> how disk devices are created, and if the disk names can be
> 'lexically' mapped to the controller ID and SCSI ID values?
This is a bit tricky...it can be decoding rz {partition} {unit}
I would suggest you look at sys_check...it does handling
name to scsi bus target lun and vice versa...both for
old and new device names (/dev/rzxxx) and (/dev/disk/diskxx)
You should have them do this correctly from the start.
Again, also ensure they understand lsm volumes, and that you
can't translate them back to a device due to mapping in lsm
|
| 8652.7 | is this a silly question? | BBPBV1::WALLACE | john wallace @ bbp. +44 860 675093 | Sun Feb 02 1997 14:14 | 5 |
| Re .0: By "custom directory format" do you mean the directory format
that the printer wants to see (sounds like it thinks it knows how to
read a "dedicated" on-disk file structure which may or may not be known
to UNIX's filesystems) ? If so, UNIX folks, does that change any of the
previous answers?
|