home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.bsd
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!agate!tfs.com!tfs.com!julian
- From: julian@tfs.com (Julian Elischer)
- Subject: Re: [386bsd] Queries for the new scsi driver.
- Message-ID: <1993Jan6.200235.4322@tfs.com>
- Organization: TRW Financial Systems
- References: <3813@tansei1.tansei.cc.u-tokyo.ac.jp>
- Date: Wed, 6 Jan 1993 20:02:35 GMT
- Lines: 171
-
- In article <3813@tansei1.tansei.cc.u-tokyo.ac.jp> mhiroshi@tansei.cc.u-tokyo.ac.jp (H. Murakami) writes:
- >To: comp.unix.bsd
- >Subject: [386bsd] Queries for the new scsi driver.
- >
- >I tried the Julian new scsi driver.
- >
- >However, before I use the aha1542 specific new-scsi driver,
- >I was able to use two partition on one disk drive for the swap like:
- >
- > config "386bsd" root on as0 swap on as0b and as0e
- >
- >And both as0b and as0e were working fine even they are on the identical drive.
- >
- >After I put the new scsi driver,
- >in the configuration file for the kernel,
- >I removed many unused drivers to save the kernel size
- >by commenting out like:
- >
- > #options NFS
- > #disk wd1 at wd0 drive 1
- > #device we0 at isa ....
- > #device ne0 at isa ..
- > #device ec0 at isa ..
- > #device is0 at isa ..
- > #pseudo-device ddb
- >
- >for the swap partition I made the line like:
- >
- > config "386bsd" root on sd0 swap on sd0b and sd0e
- >
- >My current problem with this julian new scsi driver system is,
- >
- >the kernel built does not add the partition sd0e for the swap area.
- >The "swapon /dev/sd0e" makes error message saying that
- >
- > adding swap: failed! (unchanged)
- > swapon: /dev/sd0e Device not configured
-
- did you modify the devices.i386 file in i386/conf?
-
- >
- >Note, I have the line in the file /etc/fstab like:
- >
- > /dev/sd0e - swap sw - -
- >
- >This kind of trouble never happened when I used the old scsi driver.
- >What is the reason that I cannot assign /dev/sd0e for swap ?
-
- my guess is that you haven't done the full configuration..
-
- >
- >You might think that I should not use two swap reagion on one drive,
- >however I have a good reason for that.
- >Sometimes, I just use /dev/sd0b of size 50MB for the swap, and
- >/dev/sd0e of size 1.0GB for the file system by using newfs
- >to work with many files.
- >And sometimes I would like to use both /dev/sd0b and /dev/sd0e to
- >enable the large virtual memory as large as a giga bytes.
- >(However I find many trouble to extend the virtual memory of that
- >size or a few hundred meg, I still making struggle for that.)
-
- that is a good reason..
- I have done similar.
- >
- >BTW.
- > What is the advantage of the new scsi driver
- > compared the old scsi drivers ?
- > Was the old scsi driver not using DMA transfer
- > but program data transfer?
-
- both use dma,
- however the new drivers are modular, allowing the easy addition of extra
- devices. They also support multiple different types of controllers.
- The performance is much the same (disk writes get 20% faster), but
- the ease of modification and tayloring makes it easier
- to add new devices.
- the following adapteres 'fit into' the system.
-
- aha1542
- aha1742
- bt742a
- tmc1670 (mostly works)
- wd7000 (apparently mostly works)
-
- I also have drivers for:
- scsi disk
- scsi tape
- scsi cdrom
- scsi media changer
- scsi (kodak) scanner (can't release)
- scsi (AEG) character recognition engine (can't release)
- scsi (CALERA) character recognition engine (can't release)
-
- All of the device drivers can run on any of the adapters without change
- and I sometimes run them on multiple DIFFERENT cintrollers on the same bus
- at the same time:
- e.g.
- 2 disk and 1 tape on a bt742
- + 1 disk and i AEG on a aha1742
- + 1 tape on a bt742
-
- all at the same time.
- The disk are sd0,sd1,sd2
- the tapes are st0 and st1
- and the AEG is bll0 and bll1 (it has 2 lu's)
-
- try do that with the old scsi driver....
-
-
- > It seems the data transfer rate measured by the dd command
- > /usr/bin/time dd if=/dev/rsd0h of=/dev/null ibs=512x88
- > did not speed up compared with that of old scsi driver.
- > (comparision is made with only one swap partition system.
- > rsd0h is about 300MB large to reduce the effect of the caching.)
- > The jumper of the adaptec 1542b card is to use the syncronos
- > negotiation mode.
- > CPU is 486DX 33MHz, AMI bios 256K cache 16MB Dram,
- > IDA card + adaptec 1542B.
- > The scsi disk is Fujitsu's 1.6GB, 11mS, 10MB/Sec peak, SCSI-2 disk.
- > It has 512bytes/sector, 88 sectors/track, and 90 rotation/sec means,
- > it the data transfer is made ideally, its sustained transfer rate
- > is about 4MB/sec, but dd give the number around 1MB/sec.
-
- If you use a EISA card, you will achieve around 3.2MB/sec. (with 16k reads)
- the aha1542 just doesn't seem that fast. I use the bustek 742 and the adaptec
- 1742 for faster throughput. (On EISA only though)
-
- >PS.
- > What is the meaning of the message given at the boot time by the
- > new scsi driver;
- >
- > aha0 reading board settings, 150 nSEC ok, use 200 nSEC
- > ** aha0 probing for scsi devices **
- > dev0,lu0: type 0(direct),fixed 'FUJITSU M2652S-512 008P' scsi2
- > sd0: 1651MB, cyls 1944, heads 20, secs 87, bytes/sec 512
- >
- > Why it drops from 150 nS to 200nS (and what is the meaning of 150nS)?
- > Does this relate to the disk trasfer rate?
- >
- > Hiroshi Murakami
- > mhiroshi@tansei.cc.u-tokyo.ac.jp
- > hiroshi@teine.chem2.hokudai.ac.jp
-
- 150ns is the first (highest) speed it could get the dma to work at.
- it is only a small DMA test, so it might be marginal at that speed, so
- it backs off to the next slower speed. The adaptec can actually use a 100nSec
- setting but it crashes my machine when I try it so I start at 150Nsec.
- You might make it start it's speed check at 100Nsec. If it works
- then it will back down to 150Nsec rather than 200Nsec.
-
- you may then try 'hard-wiring' to 100Nsec and stopping it from
- backing down, but be on the look out for errors, as the test
- at 100Nsec only transfers some 64 bytes (I think) so just because it
- succeeds, doesn't mean that it will be reliable.
-
- >
- >PS2.
- > What is the virtue of ddb ?
- >Sorry for the buggy English.
- better than my japanese!
-
-
- +----------------------------------+ ______ _ __
- | __--_|\ Julian Elischer | \ U \/ / On assignment
- | / \ julian@tfs.com +------>x USA \ in a very strange
- | ( OZ ) 2118 Milvia st. Berkeley CA. \___ ___ | country !
- +- X_.---._/ USA+(510) 704-3137(wk) \_/ \\
- v
-
-
-
-