LinuxSA Mailing list archives

  From: Fraser Farrell <fraser@trilobytes.com.au>
  To  : <linuxsa@linuxsa.org.au>
  Date: Fri, 13 Apr 2001 07:06:43 GMT

A cure to "no DRQ after issuing WRITE"

THE PROBLEM
A message like this one appears on all of your 
console displays (logged in or not), X terminals, 
console applications, etc – disrupting anything 
that happens to be on screen:

   no DRQ after issuing WRITE
   ide0 reset: success

Because That Message appears hundreds 
of times per day; you're soon reduced to a 
quivering wreck of frustration, desperately 
reading HOW-TOS and manpages and DejaNews and 
zillions of mail archives. But all you can 
find is thousands of pleas from other poor 
sods with the same problem, and hundreds of 
replies saying "the hard disk is faulty".

But more often than not, your IDE hard disk 
is either brand new, or it was running that 
other popular operating system for months 
or years before you rescued it.


SO WHY IS IT HAPPENING?
The kernel hackers reading this know the 
answer already (and can probably recite 
the relevant source code from memory ;-) 

But in simple terms, this message is Linux's 
way of telling you that it tried to WRITE to 
a batch of disk sectors at once; but the drive 
controller didn't respond with the IDE 
"standard reply" to this multi-sector write.

Some newer drives – such as many Seagate 
models – extend their capabilities by 
providing extra info in their response to a 
WRITE. Or they may be waiting for another 
batch of data before actually doing the WRITE 
and issuing the response. 

An antique drive may also have trouble because 
its controller doesn't respond in a "standard 
IDE" way. This one probably won't affect you 
unless you're a Linux minimalist trying to 
run everything from a 20MB Conner... ;-)

Anyway, whenever Linux accesses these drives, 
it detects what it thinks is a hardware glitch, 
tries to reset the drive controller, and issues 
That Message.

Your drive ISN'T broken, but it's speaking 
in an accent. Linux needs a bit of help to 
understand what the drive controller is saying. 


A CURE
You need to use "hdparm" to interrogate the 
disk controller, and then to tell Linux how to 
access the drive in future. To save you from 
reading the manpage (recommended, hdparm has 
a ton of features...) here's the procedure that 
gets rid of That Message in most cases:

 (1) log in as root

 (2) hdparm -i /dev/hda

This interrogates your first IDE hard disk, 
and displays a pile of info about your disk 
and its controller. The interesting bit is 
the value of "MaxMultSect", which describes 
your disk's multi-sector read/write capability.
You can do hdparm -i /dev/hdb to find out about 
your second IDE hard disk, and so on.
As an example, my server's second drive 
(a 2GB Seagate) returns a value of 32 for 
MaxMultSect.

 (3) hdparm -mX /dev/hda

where X is the MaxMultSect value reported 
in step (2). This tells Linux about your 
disk's multi-sector read/write capability.
Using my second drive as an example again:
     hdparm -m32 /dev/hdb

If this stops That Message from appearing, then

 (4) Add the hdparm -mX... command to your startup 
script (typically /etc/rc.local) so that this 
cure is automatically applied if you reboot.

However, if That Message continues to appear, 
you may need to specify other hdparm settings.
Or you may really have a faulty hard drive or 
controller.

But for most sufferers, the above procedure 
should alleviate much wailing and gnashing 
of teeth. It's helped me on many occasions.


cheers,
Fraser Farrell       fraser@trilobytes.com.au

- - - - - - - - - - - - - - - - - - - - - - - -
(in deference to our idiotic new copyright laws)
Permission is granted to forward this entire 
message to anyone who might be interested, and 
to store it in mailing archives, and to display 
it on publically-accessible websites.

-- 
LinuxSA WWW: http://www.linuxsa.org.au/  IRC: #linuxsa on irc.linux.org.au
To unsubscribe from the LinuxSA list:
  mail linuxsa-request@linuxsa.org.au with "unsubscribe" as the subject