Skip to contentSkip to conversation

Disabling ATAPI retry?

DISCUSSION [email protected] 4 participants 5 messages Mar 6, 2003 – Mar 8, 2003

lore
Share & tools
lore
#0 lore raw

I am attempting to copy data off a damaged partition on an ATAPI hard drive onto a new drive using 'dd conv=sync,noerror' since my Deskstar recently blew up.

My main problem is the fact that when it hits a bad sector, the drive goes into retry mode for a sizeable portion of time before coming back with an error. This means that as a rough estimate, it is going to take a month to copy the partition.

Is there a boot parameter or a runtime command that can tell the linux IDE driver not to automatically retry on error.

Sorry if this is an FAQ, but I could not find the information.

Kelley Cook

#1 lore raw
Quoted 2 lines · On Thu, 2003-03-06 at 20:58, [email protected] wrote:

Is there a boot parameter or a runtime command that can tell the linux IDE driver not to automatically retry on error.

There isn't. You can always build a kernel set not to, but even then it takes the drive firmware a sizeable time to retry a block. If its an IBM you might want to try the ibm tools on them if you can get them. They seem to have vanished from the face of the earth when IBM dumped its disk business

#2 lore raw
Quoted 6 lines · Alan Cox
Quoted 2 lines · On Thu, 2003-03-06 at 20:58, [email protected] wrote:

Is there a boot parameter or a runtime command that can tell the linux IDE driver not to automatically retry on error.

There isn't. You can always build a kernel set not to, but even then it takes the drive firmware a sizeable time to retry a block.

Hrm. Is this something that is likely to be introduced at some point?

My interest lies in the fact that I use a Linux based DVD player which locked up for around twenty minutes the other night trying to read a dozen bad blocks on a DVD with a single scratch...

Being able to reduce that time lag to a minimum, even if it left the hardware delays, would be great -- even if it was only for ATAPI devices.

The biggest slowdown was the kernel retrying each block a number of times, then performing a full ATAPI bus reset before giving up.

Daniel

Signature · 3 lines
Why could one never do a natural thing without having to
screen it behind a structure of artifice?
        -- Edith Wharton,  _House of Mirth_
#3 lore raw
Quoted 9 lines · Daniel Pittman
Quoted 6 lines · Alan Cox
Quoted 2 lines · On Thu, 2003-03-06 at 20:58, [email protected] wrote:

Is there a boot parameter or a runtime command that can tell the linux IDE driver not to automatically retry on error.

There isn't. You can always build a kernel set not to, but even then it takes the drive firmware a sizeable time to retry a block.

Hrm. Is this something that is likely to be introduced at some point?

I take patches. Otherwise its not going to happen until the bugs are all sorted in 2.4 and 2.5, the current chipset stuff is all dealt with, the initialisation mess is resolved, ide-cs unload bugs are fixed and so on.

Alan

#4 lore raw

IBM^H^H^HHitachi disk tools (was Re: Disabling ATAPI retry?)

In reply to Alan Cox
Quoted 5 lines · Alan Cox

There isn't. You can always build a kernel set not to, but even then it takes the drive firmware a sizeable time to retry a block. If its an IBM you might want to try the ibm tools on them if you can get them. They seem to have vanished from the face of the earth when IBM dumped its disk business

Actually, IBM dumped its disk business to Hitachi, and that's where the tools went too:
http://www.hgst.com/hdd/support/download.htm

They say "HITACHI" instead of "IBM" but otherwise they're the same tools with the same user interface and same functionality.

-Barry K. Nathan <[email protected]>