Go Back   nV News Forums > Linux Support Forums > General Linux

Newegg Daily Deals

Reply
 
Thread Tools
Old 01-31-07, 07:42 AM   #61
stevemca@optush
Registered User
 
Join Date: Jan 2007
Posts: 3
Default Re: nForce 4 corrupting data written to HDD

Is there an update to this disk corruption issue? I have been plagued by what seems to be the same issue as others have described above. That is, only a handful of bytes out of every GB or so being corrupted. As with everybody else, no errors, nothing in the logs.

My config is Gigabyte M55PlusS3(rev 1.0), 4GB RAM, Athlon64x2 4600, Fedora Core 6.

I have tried all of the suggestions in this thread all to no avail.

I upgraded to the latest kernel (2.6.19-1.2895) a few days ago and have noticed the issue has reduced by an order of magnitude. I'm afraid it doesn't help very much as I am running a heavily used Postgresql database that suffers each day from either corrupted tables, corrupted indexes or a system crash (presumably because some critical system data has been written incorrectly).

I understand the problem is under investigation. Is there any outcome from that investigation? Any idea when an update (presumably one is required) will be available for FC6?
stevemca@optush is offline   Reply With Quote
Old 02-01-07, 05:20 PM   #62
calestyo
Christoph Anton Mitterer
 
Join Date: Dec 2006
Location: München, Germany
Posts: 48
Send a message via ICQ to calestyo Send a message via AIM to calestyo Send a message via MSN to calestyo Send a message via Yahoo to calestyo
Default Re: nForce 4 corrupting data written to HDD

Quote:
Originally Posted by stevemca@optush
Is there an update to this disk corruption issue?
Not yet, any updates are published in the lkml thread and/or the bug-tracker at kernel.org

Quote:
Originally Posted by stevemca@optush
I have tried all of the suggestions in this thread all to no avail.
Have your tried the iiommu=soft or BIOS memhole mapping=disabled workarounds?

Quote:
Originally Posted by stevemca@optush
I upgraded to the latest kernel (2.6.19-1.2895) a few days ago and have noticed the issue has reduced by an order of magnitude.
Cannot believe that. As far as I know nothing has changed in 2.6.19 (the vanilla tree).
Although some things changed in 2.6.20-rcX... (have a look at the lkml thread for more information).

Chris.
calestyo is offline   Reply With Quote
Old 02-02-07, 02:26 AM   #63
stevemca@optush
Registered User
 
Join Date: Jan 2007
Posts: 3
Default Re: nForce 4 corrupting data written to HDD

Quote:
Not yet, any updates are published in the lkml thread and/or the bug-tracker at kernel.org
Thanks

Quote:
Have your tried the iiommu=soft or BIOS memhole mapping=disabled workarounds?
Yes seems to make no difference

Quote:
Cannot believe that. As far as I know nothing has changed in 2.6.19 (the vanilla tree).
Although some things changed in 2.6.20-rcX... (have a look at the lkml thread for more information).
Understand, but that's what I'm seeing. Immediately after building the box simply loading the database via a SQL copy and then doing an integrity check (via vacuumdb) would show page header issues or index problems. After upgrading to 2.6.19 (but also moving the database to two WD 36GB Raptors (SATA 1??)) the problems reduced substantially to the point where I could run the database for 24hrs before the dreaded page header faults hit. Perhaps it was moving the database to the raptors that alleviated the issue more than the kernel upgrade.


Steve
stevemca@optush is offline   Reply With Quote
Old 02-02-07, 08:35 AM   #64
calestyo
Christoph Anton Mitterer
 
Join Date: Dec 2006
Location: München, Germany
Posts: 48
Send a message via ICQ to calestyo Send a message via AIM to calestyo Send a message via MSN to calestyo Send a message via Yahoo to calestyo
Default Re: nForce 4 corrupting data written to HDD

Quote:
Originally Posted by stevemca@optush
Yes seems to make no difference
Steve
Than it seems that you suffer from a different error (or the one reportet by Kurtis Raader)...

Chris.
calestyo is offline   Reply With Quote
Old 02-07-07, 06:15 AM   #65
stevemca@optush
Registered User
 
Join Date: Jan 2007
Posts: 3
Default Re: nForce 4 corrupting data written to HDD

Quote:
Than it seems that you suffer from a different error (or the one reportet by Kurtis Raader)...
It seems I was a little hasty in my workaround testing. Not sure what happened but my original testing with iommu=soft didn't produce anything positive. Based on your comments I went back and tried again. Lo and behold (cross my fingers) since adding iommu=soft I have been running for 3 days without an error.

Perhaps I made a typo in editing grub.

Performance impact with this switch hasn't been too bad. Maybe 3-5% in increased runtime for my jobs. Still it would be nice to get a fix and I still worry if some of my O/S has been damaged whilst installing the O/S with buggy disk writes.
stevemca@optush is offline   Reply With Quote
Old 02-07-07, 07:05 AM   #66
calestyo
Christoph Anton Mitterer
 
Join Date: Dec 2006
Location: München, Germany
Posts: 48
Send a message via ICQ to calestyo Send a message via AIM to calestyo Send a message via MSN to calestyo Send a message via Yahoo to calestyo
Default Re: nForce 4 corrupting data written to HDD

Quote:
Originally Posted by stevemca@optush
Still it would be nice to get a fix
Yes, as far as I know AMD and Nvidia is still investigating in this issue. You might track the bug at kernel.org's bugzilla for news.

Quote:
Originally Posted by stevemca@optush
and I still worry if some of my O/S has been damaged whilst installing the O/S with buggy disk writes.
Unfortunately this would have been possible. Perhaps it would be best to reinstall it.

Chris.
calestyo is offline   Reply With Quote
Old 02-07-07, 11:14 AM   #67
netllama
NVIDIA Corporation
 
Join Date: Dec 2004
Posts: 8,763
Default Re: nForce 4 corrupting data written to HDD

Yes, all involved parties are still investigating the issue, and pursuing a root cause.
netllama is offline   Reply With Quote
Old 04-12-07, 12:15 PM   #68
netllama
NVIDIA Corporation
 
Join Date: Dec 2004
Posts: 8,763
Default Re: nForce 4 corrupting data written to HDD

To provide a status update, the root cause of the problem appears to have been determined, and a kernel patch is being prepared for approval by Andi Kleen. Once approved, it will be posted to the LKML.
netllama is offline   Reply With Quote

Old 04-12-07, 12:22 PM   #69
calestyo
Christoph Anton Mitterer
 
Join Date: Dec 2006
Location: München, Germany
Posts: 48
Send a message via ICQ to calestyo Send a message via AIM to calestyo Send a message via MSN to calestyo Send a message via Yahoo to calestyo
Default Re: nForce 4 corrupting data written to HDD

Quote:
Originally Posted by netllama
To provide a status update, the root cause of the problem appears to have been determined, and a kernel patch is being prepared for approval by Andi Kleen. Once approved, it will be posted to the LKML.
Great!! Please provide us also some in-detail information about the problem an how it is solved.

Thanks and best wishes,
Chris.
calestyo is offline   Reply With Quote
Old 04-13-07, 06:12 AM   #70
chunkey
#!/?*
 
Join Date: Oct 2004
Posts: 662
Default Re: nForce 4 corrupting data written to HDD

+1 to where's the patch?
chunkey is offline   Reply With Quote
Old 04-13-07, 11:27 AM   #71
netllama
NVIDIA Corporation
 
Join Date: Dec 2004
Posts: 8,763
Default Re: nForce 4 corrupting data written to HDD

As I stated, the patch will be posted to the LKML once Andi Kleen has accepted it.
netllama is offline   Reply With Quote
Old 04-16-07, 11:49 AM   #72
netllama
NVIDIA Corporation
 
Join Date: Dec 2004
Posts: 8,763
Default Re: nForce 4 corrupting data written to HDD

I've attached a kernel patch (pci-gart.c.patch.gz), which is believed to resolve this problem, without any noticeable performance impact. The patch should be applied to $SRC/arch/x86_64/kernel/pci-gart.c

For those running RHEL-4.4-x86_64, Redhat has also provided the following kernel RPM which includes this same patch:
http://people.redhat.com/coldwell/ke....EL.x86_64.rpm

Please note that you should *NOT* boot with the iommu kernel parameter when you are using this patch, as doing so would mask the bug (and potential fix). Please feel free to redistribute this patch, however note that it has not yet been accepted into any official kernel tree/branch.

Testing & feedback would be appreciated.

thanks,
Lonni
Attached Files
File Type: gz pci-gart.c.patch.gz (392 Bytes, 228 views)
netllama is offline   Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


Similar Threads
Thread Thread Starter Forum Replies Last Post
Maintain Your Privacy by Manually Accepting and Rejecting "Cookies" (nV News) MikeC Open Forum 2 02-02-13 08:15 PM
Verizon's shared data plans won't save solo users much money News Archived News Items 0 06-12-12 11:40 AM
Verizon announces 'Share Everything' plans ' the future of mobile data (sort of News Archived News Items 0 06-12-12 11:40 AM
New Paper: MPI-ACC ' An Integrated Approach to Data Movement in Accelerators News Archived News Items 0 06-02-12 04:00 AM

All times are GMT -5. The time now is 04:22 AM.


Powered by vBulletin® Version 3.7.1
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Copyright ©1998 - 2014, nV News.