* IGEL DOC2000 flash driver performance
@ 1999-11-10 1:02 Jon Burford
0 siblings, 0 replies; 2+ messages in thread
From: Jon Burford @ 1999-11-10 1:02 UTC (permalink / raw)
To: mtd
[-- Attachment #1: Type: text/plain, Size: 1292 bytes --]
I have been using the IGEL flash driver on a Linux 2.2.12 system now for several months. One very annoying "feature" is that it causes the entire OS to hang for 2-5 seconds after mounting or moderate write operations. I talked with an M-Systems tech and although they did not write the driver, he thought it might be due to the garbage collection running in the DOC2000 firmware. When a block becomes some configurable percentage utilized, the garbage collector tries to coallesce the used portion of this block with others to decrease block fragmentation and increase wearability. His theory was that the driver would request space for a particular write operation and if garbage collection was necessary, the write operation would block. My problem is that when this happens and the blocking lasts for > 1.5 seconds, my watchdog refresh process can't pet the watchdog and the system resets. I have tried making the watchdog process a real-time thread with RTLinux, but the RTLinux beta11 is proving to be a little unstable for my blood right now. I am wondering if anyone has encountered this problem and/or may have some advice for me. Does the MTD DOC2000 driver suffer from this behavior and can it be patched into a 2.2.12 kernel? Any other thoughts?
TIA!
Regards,
Jon
[-- Attachment #2: Type: text/html, Size: 1845 bytes --]
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: IGEL DOC2000 flash driver performance
@ 1999-11-10 8:42 David Woodhouse
0 siblings, 0 replies; 2+ messages in thread
From: David Woodhouse @ 1999-11-10 8:42 UTC (permalink / raw)
To: Jon Burford; +Cc: mtd
jburford@xsilogy.com said:
> <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML>
> <HEAD> <META content="text/html; charset=iso-8859-1"
> http-equiv=Content-Type> <META content="MSHTML 5.00.2614.3401"
> name=GENERATOR> <STYLE></STYLE> </HEAD> <BODY bgColor=#ffffff> <DIV>
> <FONT face=Arial size=2>I have been using the IGEL flash driver on a
> Linux 2.2.12 system now for several months. One very annoying
> "feature" is that it causes the entire OS to hang for 2-5 seconds
> after mounting or moderate write operations.
Urgh. HTML.
> I am wondering if anyone has encountered this problem and/or may
> have some advice for me. Does the MTD DOC2000 driver suffer
> from this behavior and can it be patched into a 2.2.12 kernel?
> Any other thoughts?</FONT></DIV>
The bad news is: yes, the MTD DoC2000 driver does suffer from this behaviour.
It disables interrupts for the whole time it's dealing with the flash, even
when it's folding Virtual Unit Chains.
The good news is that there's no real need for it to do so, and hence it won't
be like that for long. I know exactly how I'm going to deal with it, but I
just haven't done it yet. I want to be 100% sure that my NFTL support is
correct before I make life more complex by enabling threadedness in the driver.
And yes, it can be patched into a 2.2.12 kernel - I've only used it as
modules, but with a little hacking you can compile it into the kernel. I know
that people on this list have done so - prod them to provide patches.
Remember that it's still only beta though.
---- ---- ----
David Woodhouse David.Woodhouse@mvhi.com Office: (+44) 1223 810302
Project Leader, Process Information Systems Mobile: (+44) 7976 658355
Axiom (Cambridge) Ltd., Swaffham Bulbeck, Cambridge, CB5 0NA, UK.
finger dwmw2@ferret.lmh.ox.ac.uk for PGP key.
To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~1999-11-10 8:42 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
1999-11-10 1:02 IGEL DOC2000 flash driver performance Jon Burford
-- strict thread matches above, loose matches on Subject: below --
1999-11-10 8:42 David Woodhouse
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox