2.4.18-pre9-ac4 fails to boot

2.4.18-pre9-ac4 fails to boot

[Michal Jaegermann]

Just a note that I tried to boot 2.4.18-pre9-ac4 (plus Alpha specific
adjustments) on Alpha-Nautilus box.  Got stuck in that with:

..........
FAT: bogus logical sector size 0
FAT: bogus logical sector size 0
Kernel panic: VFS: Unable to mount root fs on 03:00

I have no problems booting 2.4.18-rc1 (with the same adjustments :-).
Anybody knows by any chance what this is about before I will start
digging?

  Michal

PS. "Adjustments" mentioned above are required or I will not boot at all
with any current kernel but this is still WIP.

2.4.18-pre9-ac4 filesystem corruption

[Kristian]

Hello.

I've seen filesystem corruption using -ac4 with ext2 although I'm not using a SIS chipset. So I really recommend using not this patch.

PS: This may have nothing in common with your problem cause I encountered these corruption on Intel.

Yesterday:

kernel: init_special_inode: bogus imode (177777) 
kernel: init_special_inode: bogus imode (177777) 
kernel: init_special_inode: bogus imode (167777) 
kernel: init_special_inode: bogus imode (177777) 
last message repeated 2 times 
kernel: init_special_inode: bogus imode (177767) 
kernel: init_special_inode: bogus imode (177767) 
kernel: init_special_inode: bogus imode (137777) 
kernel: init_special_inode: bogus imode (167777) 
kernel: init_special_inode: bogus imode (177777) 
kernel: init_special_inode: bogus imode (177757) 
kernel: init_special_inode: bogus imode (177677) 
kernel: init_special_inode: bogus imode (177777) 
last message repeated 2 times 
kernel: init_special_inode: bogus imode (177377) 
kernel: init_special_inode: bogus imode (177777) 

after fsck-ing:

kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (20) 
kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (10) 
kernel: init_special_inode: bogus imode (0) 
last message repeated 6 times 
kernel: init_special_inode: bogus imode (20) 
kernel: init_special_inode: bogus imode (0) 
last message repeated 3 times 

Today:

kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (35623) 
kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (30070) 
kernel: init_special_inode: bogus imode (30061) 
kernel: init_special_inode: bogus imode (0) 
kernel: init_special_inode: bogus imode (35623) 
kernel: init_special_inode: bogus imode (0) 

I'm using -rc2 now and compiling 5 kernels at the same time and haven't seen any of those problems yet.

*Kristian

Michal Jaegermann <michal@harddata.com> wrote:
> ..........
> FAT: bogus logical sector size 0
> FAT: bogus logical sector size 0
> Kernel panic: VFS: Unable to mount root fs on 03:00

  :... [snd.science] ...:
 ::
 :: http://www.korseby.net
 :: http://gsmp.sf.net
  :..........................:

Re: 2.4.18-pre9-ac4 filesystem corruption

[Alan Cox]

> I've seen filesystem corruption using -ac4 with ext2 although I'm not using
> a SIS chipset. So I really recommend using not this patch.

The SiS patch is only changing anything if the SiS vode is in use.

Precisely what chipset, what IDE, what ide cable (40/80 pin) and drives
do you have. What hdparm commands are you using if any ?

Alan

Re: 2.4.18-pre9-ac4 filesystem corruption

[Kristian]

Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> > I've seen filesystem corruption using -ac4 with ext2 although I'm not using
> > a SIS chipset. So I really recommend using not this patch.
> 
> The SiS patch is only changing anything if the SiS vode is in use.
> 
> Precisely what chipset, what IDE, what ide cable (40/80 pin) and drives
> do you have. What hdparm commands are you using if any ?

No hdparm settings. 40pin cable and this drive:

$ dmesg|grep hda
    ide0: BM-DMA at 0x10a0-0x10a7, BIOS settings: hda:DMA, hdb:DMA
hda: WDC AC24300L, ATA DISK drive
hda: 8421840 sectors (4312 MB) w/256KiB Cache, CHS=557/240/63, UDMA(33)

$ lspci
00:00.0 Host bridge: Intel Corp. 440BX/ZX - 82443BX/ZX Host bridge (rev 02)
00:01.0 PCI bridge: Intel Corp. 440BX/ZX - 82443BX/ZX AGP bridge (rev 02)
00:0e.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone]
00:0f.0 Token ring network controller: IBM 16/4 Token ring UTP/STP controller (r
ev 05)
00:14.0 ISA bridge: Intel Corp. 82371AB PIIX4 ISA (rev 02)
00:14.1 IDE interface: Intel Corp. 82371AB PIIX4 IDE (rev 01)
00:14.2 USB Controller: Intel Corp. 82371AB PIIX4 USB (rev 01)
00:14.3 Bridge: Intel Corp. 82371AB PIIX4 ACPI (rev 02)
01:00.0 VGA compatible controller: Matrox Graphics, Inc. MGA G200 AGP (rev 01)

Before you ask: I'll test memory later just to be sure.

*Kristian

  :... [snd.science] ...:
 ::
 :: http://www.korseby.net
 :: http://gsmp.sf.net
  :..........................:

Re: 2.4.18-pre9-ac4 filesystem corruption

[Alan Cox]

> No hdparm settings. 40pin cable and this drive:
> 
> $ dmesg|grep hda
>     ide0: BM-DMA at 0x10a0-0x10a7, BIOS settings: hda:DMA, hdb:DMA
> hda: WDC AC24300L, ATA DISK drive
> hda: 8421840 sectors (4312 MB) w/256KiB Cache, CHS=557/240/63, UDMA(33)
> 
> 00:14.1 IDE interface: Intel Corp. 82371AB PIIX4 IDE (rev 01)

PIIX and the WDC drive is supposed to be past the range that had the
nasty UDMA DMA bugs.

> Before you ask: I'll test memory later just to be sure.	
 
Ok

Re: 2.4.18-pre9-ac4 filesystem corruption

[Kristian]

Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> 
> PIIX and the WDC drive is supposed to be past the range that had the
> nasty UDMA DMA bugs.
> 
> > Before you ask: I'll test memory later just to be sure.	
>  
> Ok

memtest86 completed successfully.

I'll test with -rc2-ac1 for ext2 corruption again.

*Kristian

  :... [snd.science] ...:
 ::
 :: http://www.korseby.net
 :: http://gsmp.sf.net
  :..........................:

Re: 2.4.18-pre9-ac4 filesystem corruption

[Alan Cox]

> memtest86 completed successfully.
> I'll test with -rc2-ac1 for ext2 corruption again.

Thanks. If you do see it can you test with ide=nodma as well and see what
that does. Andre will probably also want to know how long your IDE cables
are 8)

Re: 2.4.18-pre9-ac4 filesystem corruption

[Kristian]

Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> > memtest86 completed successfully.
> > I'll test with -rc2-ac1 for ext2 corruption again.
> 
> Thanks. If you do see it can you test with ide=nodma as well and see what
> that does. Andre will probably also want to know how long your IDE cables
> are 8)

Booting just normal with -rc2-ac1:

Directly after boot appears this messages again:
	init_special_inode: bogus imode (70141) 
I'd had to run e2fsck -f /dev/hda5 (/) on it:
	Entry 'par2' in /dev (8166) has deleted/unused inode 9337.
But former errors occured on other partitons as well (/dev/hda(5-8)).
The cable used:
	+----------+----------------------+
	|          |                      |
	|<--16cm-->|<--------40cm-------->|
	+----------+----------------------+
    controller  WDC AC24300L          LTN301 CDROM

Booting with ide=nodma:

The corruption hasn't appeared yet.

I've also tried dd if=/dev/hda(1,5-8) of=/dev/null bs=4k without error-messages in the logs.

So I may switch back to -pre9-ac3 again and see if it happens with that kernel.

*Kristian

  :... [snd.science] ...:
 ::
 :: http://www.korseby.net
 :: http://gsmp.sf.net
  :..........................:

Re: 2.4.18-pre9-ac4 filesystem corruption

[Kristian]

Ok. Forgive me my own stupidity. Putting that cable in wrong order is a fault I will not make again. I really must ask myself why I haven't noticed it the last 13 years... ;-)


  :... [snd.science] ...:
 ::
 :: http://www.korseby.net
 :: http://gsmp.sf.net
  :..........................: