Mailing List Archive

Support open source code!


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: tlug: Sparc Linux



On Thu, Jan 14, 1999 at 02:01:36PM +0900, Chris Sekiya wrote:

> > I can do most of the build okay, but in the final stages of setup, the
> > SCSI chain hangs, then resets itself.  After this point nothing seems
> > to work correctly.
> 
> During the build of the coda userspace utilities?

Sorry, I should have been more specific.  Everything compiles okay.  But
when the Coda utilities are then used to build the filesystem partitions,
things go pear-shaped.

> What does the partition table for that drive look like?  What else is on
> that SCSI line?

I'm not at the machine right now, but the last setup I used had something
like 3 megs in /dev/hdb1 and 120k in /dev/hdb2, with the rest of the disk
as unformatted nothing.  In initial tests, I had two of these 4-gig
external drivers plus an HP DAT tape drive on the same line.  Someone with
SCSI experience pointed out that the total cable length was probably over
spec, so I cut back to just that single external drive on the line, and
checked that the cable came from a reputable maker (according to a
Reliable Source:).  Things improved a little, but a little isn't enough of
course.

> > ... as a general question,
> > are there grounds for believing that the Linux SCSI drivers in RH
> > Linux 5.1 are flakey?
> 
> Yes.  Compiling the 2.0.x esp.c with egcs is a Bad Thing.  Symptoms
> include, but are not limited to:
> 
> * An aborted SCSI command during boot-time device detection.  Depending on
>   whether your sparc contains a 53C90 or a 53C90A, the box may or may not
>   successfully reset the SCSI controller after a five-second pause.
> * Random command timeouts under slight load.

These are EXACTLY the symptoms I've been getting.  Thank you for this
information, Chris.  Let me know if there's anything I can bring you from
Nagoya when I finally do get up there for a meeting.

> RedHat "fixed" this by compiling esp.c with gcc-2.7, but compiled the rest
> of the kernel with egcs.  This is another Bad Thing.
>
> Me, I'd grab kernel 2.1.130 from one of the kernel mirrors and compile it
> myself.  RedHat did a serious number on their sparc kernel -- it's not
> really 2.0, but it's not really 2.1 either.  I don't trust it.

Thank you for this, it is VERY helpful.  Would the compiler on RH 4.1
Sparc Linux be gcc, not egcs?  If so, we have a 4.1 box to hand and I may
be in the clear. How can you distinguish egcs from gcc on a running box
--- the top-level binary for both is called gcc, isn't it? 

Cheers,
-- 
-x80
Frank G Bennett, Jr         @@
Faculty of Law, Nagoya Univ () email: bennett@example.com
Tel: +81[(0)52]789-2239     () WWW:   http://rumple.soas.ac.uk/~bennett/

-------------------------------------------------------------------
Next Nomikai: 14 January 1999, 19:30 Tengu TokyoEkiMae 03-3275-3691
*** it will will be Jan 14 (Thu), as Jan 15 (Fri) is a natl holiday
Next Technical Meeting: Feb 13 (Sat), 12:30    ace: Temple Univ.
-------------------------------------------------------------------
more info: http://tlug.linux.or.jp                     Sponsor: PHT


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links