Bug 27247

Summary:	Panic on install - "page fault syncing discs"
Product:	Base System	Reporter:	mark <mark>
Component:	i386	Assignee:	Søren Schmidt <sos>
Status:	Closed FIXED
Severity:	Affects Only Me
Priority:	Normal
Version:	Unspecified
Hardware:	Any
OS:	Any

Description mark 2001-05-10 15:10:03 UTC

6.5GB ATA33 IBM HDD, on Biostar M7MIA mobo including the VIA IDE controller identified by the boot process as a "VIA 82C686 ATA66 Controller", althoguh it may be the 686B chip not the 686. Full spec included below.

Hard drive partitioned as 4.5GB Win2K/FAT32 partition, rest empty awaiting a FreeBSD install. I've checked out the hard drive with the IBM disk testing utility and all tests passed fine (I've even tried doing a "low" level format on it using the IBM wipe utility,
to no avail), also windows 2000 installed without problem.

I have been booting from 4.1-release install CD, set up kernel as usual knocking out all the unused hardware and sorting out any conflicts. Picked express setup, created a FreeBSD partition, sliced it as per my requirements (100MB root, 260MB swap, 300MB /var, and the rest for /usr), picked the minimal distribution (althoguh picking any of the others has the same/similar effect), install begins. This is all much as I've done before with no problems before the new motherboard & hard drive re-format.

At a seemingly random point while either extracting/installing the distribution, or while the "Remaking all devices.. Please Wait!" message is on screen shortly thereafter, the computer pauses for a fair time (probably around two to five minutes, not always the same though) then displays the messages...

panic: page fault syncing discs... 374 374 374 374 374 374 (repeat 374 a total of 20
times)
giving up on 328 buffers
Uptime: (whatever)
Automatic reboot in 15 seconds....

I've a strong suspicion this is related to the hard drive controller, althoguh I've not been able to find any problems reported anywhere around the net with it, so that assumption could be wrong.

Also, on one occasion while trying to install, it did manage to complete the install, and it rebooted successfully once, but after a few minutes of use I had a "read command timeout" error, followed by a seemingly endless loop of the OS reseting the drive controller then it failing again, reseting again, failing again, and so on, only solved by
a hard reset. On restarting FreeBSD wouldn't even boot, crashing out with a "BTX halted" error and a load of hex (which I'm afraid I didn't think to note down, there was a lot of it) and no error codes or messages aside from this.

HW Spec:
Biostar M7MIA motherboard, using AMD761 north bridge and VIA 686B south bridge
128MB DDR PC2100 memory (under-clocked at 100/200MHz to match FSB of processor)
AMD Duron 750MHz CPU
IBM Deskstar 6.5GB ATA33 Hard Disk Drive (partitioned as 4.5GB FAT32 Win2K, rest FreeBSD)
SMC 1211TX-WL 10/100 WOL NIC
ATI Radeon VE 32MB DDR Graphics
12x ARTEC DVDROM
3.5" floppy drive, 1.44MB (wow)

Fix:

No idea, I'm afraid.
How-To-Repeat: Put in the install CD and attempt to install.

Comment 1 mark 2001-05-11 01:29:42 UTC

Follow up:

Managed to get it to complete install again, and boot, twice. First time it booted up with
a "read command timeout" error then a loop of reseting devices, write error, resetting
devices, etc, after the UDMA startup sequence. After this the filesystem got ruined and it
would not boot again. Second time though I managed to boot and login, it failed to start
up the DMA mode on startup so didn't go all haywire. Placed

/sbin/sysctl -w hw.atamodes="pio,dma"

into /etc/rc.conf which has allowed me to boot every time successfully and the DMA related
error messages have gone. However, this obviously comes at the cost of abysmal hard drive
performance.

Hope this helps to narrow down where the problem lies.

Mark

Comment 2 mark 2001-05-11 02:20:24 UTC

Follow up 2:

Having run the system for about a half hour, it stopped working again, with 

ad0: read command timeout
ata0: resetting devices .. done
ad0: WRITE command timeout
ata0: resetting devices .. done
ad0: read command timeout
ata0: resetting devices .. done

eventually ending in a 

/: bad dir ino 12611 .. mangled entry

then rebooting itself after a panic which i didn't catch before it rebooted i'm afraid.

so, i'm back where I started. guess I was just lucky that it worked for a bit.

Regards,
Mark

Comment 3 Jens Schweikhardt freebsd_committer

2002-08-10 15:59:51 UTC

State Changed
From-To: open->feedback

There has been some work on the ATA code. Does this problem 
still persist in a recent 4-STABLE?

Comment 4 mh_lists 2002-08-10 16:21:28 UTC

Began working from later version of FreeBSD (4.3-releaase onwards) and
installed fine and has been working for many months without problems of this
nature. Assume the bug is now fixed?

Regards,
Mark

Comment 5 leonids 2003-01-31 00:12:13 UTC

Looks like it's doing it again. I went to upgrade from 4.5 to 4.7 today.
And, I got the same EXACT error, at the same exact point. The only different
was mine said panic: page fault syncing discs... 1 1 1 1 1 1 1 1 instead. My
hardware is the following:

Mobo:  Intel CA810e
HD:      ad1: 14324MB <QUANTUM FIREBALLlct10 15> [29104/16/63] at ata0-slave
UDMA66
            ad2: 14324MB <QUANTUM FIREBALLlct10 15> [29104/16/63] at
ata1-master UDMA66

Just as I side note, in 4.5, on boot I get the following error:

Mounting root from ufs:/dev/ad1s1a
ad1: READ command timeout tag=0 serv=0 - resetting
ata0: resetting devices .. done
ad1: READ command timeout tag=0 serv=0 - resetting
ata0: resetting devices .. done
ad1: READ command timeout tag=0 serv=0 - resetting
ata0: resetting devices .. done
ad1: READ command timeout tag=0 serv=0 - resetting
ad1: trying fallback to PIO mode
ata0: resetting devices .. done


Leo S.

Comment 6 Kris Kennaway freebsd_committer

2003-07-12 13:27:40 UTC

Responsible Changed
From-To: freebsd-bugs->sos

Assign to the ATA maintainer for comment

Comment 7 Søren Schmidt freebsd_committer

2003-07-14 12:28:45 UTC

State Changed
From-To: feedback->open

A complete dmesg would have been nice, as I cannot tell if you have a 
master device on channel 0 since the disk is *slave* which wont 
work well with fireballs..

Comment 8 Søren Schmidt freebsd_committer

2003-09-08 20:05:00 UTC

State Changed
From-To: open->closed

feedback timeout