OSR5.0.4: Floppy I/O Errors on Loaded System

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Lucky Leavel » Fri, 02 Oct 1998 04:00:00



OS: OSR5.0.4c, no patches installed;P-120 with 32NB RAM, 4 SCSI HDs on
Adaptec 1542B

I was trying to create Lone-TAR Airbag diskettes on this system while a
tape backup was running and several users to running COBOL runtimes and
Ingres RDBMS apps.  I kept getting floppy I/O errors (mostly read but some
writes). If I interrupted the backup, the I/O errors went away using the
same diskettes.

I have read TA 105797 and 109277.  Would oss476a help here?

Thank you,
Lucky

Lucky Leavell                      Phone: (800) 481-2393  
UniXpress - Your Source for SCO       OR: (812) 989-0174
1560 Zoar Church Road NE             FAX: (812) 366-3618

WWW Home Page:  http://www.UniXpress.com  

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Adam R. Pa » Fri, 02 Oct 1998 04:00:00




Quote:>OS: OSR5.0.4c, no patches installed;P-120 with 32NB RAM, 4 SCSI HDs on
>Adaptec 1542B

Minor nit-pick:
If there are no patches installed then its not OSR5.0.4c :)  The product
as-shipped is version 5.0.4p.  A 'c' could only refer to rs504c...

Quote:>I was trying to create Lone-TAR Airbag diskettes on this system while a
>tape backup was running and several users to running COBOL runtimes and
>Ingres RDBMS apps.  I kept getting floppy I/O errors (mostly read but some
>writes). If I interrupted the backup, the I/O errors went away using the
>same diskettes.

>I have read TA 105797 and 109277.  Would oss476a help here?

It sounds like the symptoms you're describing are similar to those that
oss476a is intended to fix - only way to find out if it would help is to
install it :)

_Adam
--

Not speaking for SCO Engineering

"Saint, n.  A dead sinner revised and edited" - Ambrose Bierce

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Bill Vermilli » Mon, 05 Oct 1998 04:00:00




Quote:>OS: OSR5.0.4c, no patches installed;P-120 with 32NB RAM, 4 SCSI HDs on
>Adaptec 1542B
>I was trying to create Lone-TAR Airbag diskettes on this system
>while a tape backup was running and several users to running COBOL
>runtimes and Ingres RDBMS apps. I kept getting floppy I/O errors
>(mostly read but some writes). If I interrupted the backup, the I/O
>errors went away using the same diskettes.
>I have read TA 105797 and 109277. Would oss476a help here?

This might help.  This is from the FAQ that Steve Dunn posts
regualary.

Note however that the value listed in the FAQ to change TO - is
15 while the SCO notes say 8.

I had the problem installing the 504c patches because of this - on
a quiescent system.

------------------ start included text ------------------------

  My floppy drive doesn't work reliably
  =====================================

  First things first - it may be a bad diskette, a bad floppy drive, or the
  cable connecting your floppy drive to its controller may be loose.

  However, if the problem gets worse if the system is under load
  (particularly, when anything is doing DMA - and that usually means a SCSI
  host adapter) and is particularly bad when trying to format a floppy, you
  may be running into a problem with the FIFO on the floppy controller.

  Briefly, older floppy controllers have a one-byte buffer, and the system
  must grab that byte before the next byte is read from the drive or else you
  have a problem. If the DMA controller (floppies use DMA) is blocked long
  enough, usually by a SCSI host adapter, the buffer gets overwritten. Newer
  controllers generally have a 16-byte buffer, but it may or may not be
  turned on by your BIOS. SCO Unix doesn't program the buffer by default, but
  the floppy driver in OSR5 can be set to use the buffer if you wish.

  Make a backup of /etc/conf/pack.d/fd/space.c (just in case). Now edit the
  file. At the bottom you will see

  int fd_enable_FIFO = 0;
  int fd_FIFOthresh = 0;

  Change them to

  int fd_enable_FIFO = 1;
  int fd_FIFOthresh = 15;

  The first one simply enables or disables programming of the FIFO; the second
  one sets the FIFO's size if you've enabled it. There's probably no reason
  to set it to anything other than 15.

  Relink and reboot to activate the change.

  Making this change can actually be counterproductive if your system is
  working properly. Some floppy drive controllers will not work properly if
  you try enabling the FIFO, and you will end up with a system with floppy
  drives which do not work. Trust me on this one.

  ____________________________________________________________________________
------------- end of included text --------------------

--

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Tony Lawrenc » Mon, 05 Oct 1998 04:00:00


Quote:

>   My floppy drive doesn't work reliably
>   =====================================
>   Make a backup of /etc/conf/pack.d/fd/space.c (just in case). Now edit the
>   file. At the bottom you will see

>   int fd_enable_FIFO = 0;
>   int fd_FIFOthresh = 0;

>   Change them to

>   int fd_enable_FIFO = 1;
>   int fd_FIFOthresh = 15;

Note that the release notes for 5.0.5 say:

 " We recommend setting the value to 13 for
most Pentium systems, although you may have to experiment to
find the best value for your system."

--

SCO ACE
SCO articles, help, book reviews: http://www.aplawrence.com

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Tony Lawrenc » Mon, 05 Oct 1998 04:00:00



> Note that the release notes for 5.0.5 say:

>  " We recommend setting the value to 13 for
> most Pentium systems, although you may have to experiment to
> find the best value for your system."

But strangely:  Those variables are NOT in the space.c file
(they certainly are on 5.0.4)!

--

SCO ACE
SCO articles, help, book reviews: http://www.aplawrence.com

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Bill Vermilli » Tue, 06 Oct 1998 04:00:00




Quote:

>>   My floppy drive doesn't work reliably
>>   =====================================
>>   Make a backup of /etc/conf/pack.d/fd/space.c (just in case). Now edit the
>>   file. At the bottom you will see

>>   int fd_enable_FIFO = 0;
>>   int fd_FIFOthresh = 0;

>>   Change them to

>>   int fd_enable_FIFO = 1;
>>   int fd_FIFOthresh = 15;

>Note that the release notes for 5.0.5 say:

> " We recommend setting the value to 13 for
>most Pentium systems, although you may have to experiment to
>find the best value for your system."

And the 5.0.4 notes say the maximum value can be 'f'.  
Is someone confusing decimal/hex here?
--

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Jean-Pierre Radle » Tue, 06 Oct 1998 04:00:00


Tony Lawrence typed (on 04Oct):
| >
| >   My floppy drive doesn't work reliably
| >   =====================================
| >   Make a backup of /etc/conf/pack.d/fd/space.c (just in case). Now edit the
| >   file. At the bottom you will see
| >
| >   int fd_enable_FIFO = 0;
| >   int fd_FIFOthresh = 0;
| >
| >   Change them to
| >
| >   int fd_enable_FIFO = 1;
| >   int fd_FIFOthresh = 15;
|
| Note that the release notes for 5.0.5 say:
|
|  " We recommend setting the value to 13 for
| most Pentium systems, although you may have to experiment to
| find the best value for your system."

Since the Release Notes don't have an index, I'm having a problem
spotting those instructions.  What page number?

--

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Tony Lawrenc » Tue, 06 Oct 1998 04:00:00



> Tony Lawrence typed (on 04Oct):
> |
> | Note that the release notes for 5.0.5 say:
> |
> |  " We recommend setting the value to 13 for
> | most Pentium systems, although you may have to experiment to
> | find the best value for your system."

> Since the Release Notes don't have an index, I'm having a problem
> spotting those instructions.  What page number?

Page 51 in the set I'm looking at, which claims to be Document version
5.0.5, 1 August 1998.

But, as I said in another post, the actual system does NOT
have these symbols in fd/space.c.

--

SCO ACE
SCO articles, help, book reviews: http://www.aplawrence.com

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Bill Vermilli » Tue, 20 Oct 1998 04:00:00




Quote:

>>   My floppy drive doesn't work reliably
>>   =====================================
>>   Make a backup of /etc/conf/pack.d/fd/space.c (just in case). Now edit the
>>   file. At the bottom you will see

>>   int fd_enable_FIFO = 0;
>>   int fd_FIFOthresh = 0;

>>   Change them to

>>   int fd_enable_FIFO = 1;
>>   int fd_FIFOthresh = 15;
>Note that the release notes for 5.0.5 say:
> " We recommend setting the value to 13 for
>most Pentium systems, although you may have to experiment to
>find the best value for your system."

Thanks.  Now for one more question.  On the original TA for the
earlier OSR5 releases, it said to try 8, and work down.  It also
noted the the maximum value was to be  'f'.  I found the TA to be
at odds with other recommendations of '15'.  Even though they are
both the same value - was this just a documentation bug?

--

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Jean-Pierre Radle » Sun, 25 Oct 1998 04:00:00


Tony Lawrence typed (on 19Oct):
|
| > Note that the release notes for 5.0.5 say:
| >
| >  " We recommend setting the value to 13 for
| > most Pentium systems, although you may have to experiment to
| > find the best value for your system."
|
| But strangely:  Those variables are NOT in the space.c file
| (they certainly are on 5.0.4)!

I bug-reported that a couple of weeks ago.

--

 
 
 

OSR5.0.4: Floppy I/O Errors on Loaded System

Post by Tony Lawrenc » Mon, 26 Oct 1998 03:00:00



> Tony Lawrence typed (on 19Oct):
> | But strangely:  Those variables are NOT in the space.c file
> | (they certainly are on 5.0.4)!

> I bug-reported that a couple of weeks ago.

Yes, you did.  

But I didn't send this again on October 19th.  But something
somewhere is regurgitating posts, and confusing the hell
out of all of us.

Is anyone making any progress on finding this damn machine
and stopping it?

--

SCO ACE
SCO articles, help, book reviews: http://www.aplawrence.com