Re: panic: biodone2 already

[Lars Nordlund <lars.nordlund%hem.utfors.se@localhost>]

On Thu, 11 Sep 2008 15:20:11 +0200
Juergen Hannken-Illjes <hannken%eis.cs.tu-bs.de@localhost> wrote:

> On Mon, Sep 01, 2008 at 07:22:15PM +0200, Lars Nordlund wrote:
> > Hi
> > 
> > Got this panic while writing a DVD+RW last night. I am running NetBSD
> > 4.99.72/amd64. Kernel message hand-copied from screen..
> > 
> > panic: biodone2 already
> 
> Lars,
> please check if you see this or a related panic with the diff attached.

Hi again

While testing stuff this week I noticed that my wd0 disk was slower
than usual. It was only doing 5-10MiB/s instead of the more normal
50MiB/s. It turned out to be caused by a BIOS setting change I made
some weeks ago while hunting for some 2000 lost sectors. Anyway,
changing SATA/RAID mode from 'disabled' to 'AHCI' gave back the
proper disk performance.

Since I am writing the DVDs from an ISO on wd0 I wanted to
re-test with and without your patch. Without patch I got a hang
within "half a DVD-RW". With your patch the computer has been able to
write DVD-RWs all day long.

The 'dmesg diff' is:

===> lars@laphroaig:~ %diff dmesg.before dmesg.after
7c7
< NetBSD 4.99.72 (LAPHROAIG) #17: Thu Sep 11 20:20:59 CEST 2008
---
> NetBSD 4.99.72 (LAPHROAIG) #1: Sat Sep 13 13:55:10 CEST 2008
14c14
< ( )
---
>     ( )
74d73
< mskc is not split
77d75
< msk is not split
119d116
< ahc is not split
136,143c133,141
< piixide0 at pci0 dev 31 function 2
< piixide0: Intel 82801H Serial ATA Controller (ICH8) (rev. 0x02)
< piixide0: bus-master DMA support present
< piixide0: primary channel configured to native-PCI mode
< piixide0: using ioapic0 pin 19 for native-PCI interrupt
< atabus4 at piixide0 channel 0
< piixide0: secondary channel configured to native-PCI mode
< atabus5 at piixide0 channel 1
---
> ahcisata1 at pci0 dev 31 function 2: vendor 0x8086 product 0x2821
> ahcisata1: interrupting at ioapic0 pin 19
> ahcisata1: AHCI revision 1.1, 6 ports, 32 command slots, features 0xef20e0c0
> atabus4 at ahcisata1 channel 0
> atabus5 at ahcisata1 channel 1
> atabus6 at ahcisata1 channel 2
> atabus7 at ahcisata1 channel 3
> atabus8 at ahcisata1 channel 4
> atabus9 at ahcisata1 channel 5
147,154d144
< piixide1 at pci0 dev 31 function 5
< piixide1: Intel 82801H Serial ATA Controller (ICH8) (rev. 0x02)
< piixide1: bus-master DMA support present
< piixide1: primary channel wired to native-PCI mode
< piixide1: using ioapic0 pin 19 for native-PCI interrupt
< atabus6 at piixide1 channel 0
< piixide1: secondary channel wired to native-PCI mode
< atabus7 at piixide1 channel 1
167c157
< timecounter: Timecounter "TSC" frequency 2133418240 Hz quality 3000
---
> timecounter: Timecounter "TSC" frequency 2133449120 Hz quality 3000
185a176
> ahcisata1 port 4: device present, speed: 3.0Gb/s
190a182,186
> wd0 at atabus8 drive 0: <SAMSUNG HD300LJ>
> wd0: drive supports 16-sector PIO transfers, LBA48 addressing
> wd0: 279 GB, 581418 cyl, 16 head, 63 sec, 512 bytes/sect x 586070255 sectors
> wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 7
> wd0(ahcisata1:4:0): using PIO mode 4, DMA mode 2, Ultra-DMA mode 6 
> (Ultra/133) (using DMA)
204,209d199
< wd0 at atabus6 drive 0: <SAMSUNG HD300LJ>
< wd0: drive supports 16-sector PIO transfers, LBA48 addressing
< wd0: 279 GB, 581418 cyl, 16 head, 63 sec, 512 bytes/sect x 586070255 sectors
< wd0: 32-bit data port
< wd0: drive supports PIO mode 4, DMA mode 2, Ultra-DMA mode 7
< wd0(piixide1:0:0): using PIO mode 4, Ultra-DMA mode 6 (Ultra/133) (using DMA)
222d211
< pad is not split


Basically, piixide turns into ahcisata.

To sum up a long and rambling mail; ahcisata+patch seems to
do some good.

Or perhaps shorter time is now spent in the 'wd0 interrupt
handler', effectively shrinking the window when some
dvd-writing-bug may stomp upon the bufq pointers:s..?

Hmm..

Best regards,
        Lars Nordlund