Re: Linux 2.4.21pre3-ac2

[John Stoffel]

Alan> I'm seeing enough other -ac specific errors to be fairly sure
Alan> its not just hardware in the current -ac tree case. I don't know
Alan> what the common factor is yet - it 'works for me' which makes it
Alan> hard to pin down

I'm now running -ac3 and it seems more stable.  I've got a Dual Xeon
550 Mhz Dell with 768mb of ECC memory and a Matrox G450 card.  Nice
stable system usually.

Alan> Guess #1 is reverting mm/shmem.c. Guess #2 is reverting the buffer cache
Alan> changes. Guess #3 is new IDE + highmem and Guess #4 is quota related (are
Alan> people seeing the problem with quota disabled ?)

I'm not running with quotas at all, and I've seen the complete lockup
under -ac1 and -ac2.  I'll see about stressing -ac3 with some
filesystem stuff to see what happens.  

Here are the options I have enabled for IDE on my system:
    
    CONFIG_PARIDE=m
    CONFIG_PARIDE_PARPORT=m
    CONFIG_PARIDE_PD=m
    CONFIG_PARIDE_PCD=m
    CONFIG_PARIDE_PF=m
    CONFIG_PARIDE_PT=m
    CONFIG_PARIDE_PG=m
    CONFIG_IDE=y
    CONFIG_BLK_DEV_IDE=y
    CONFIG_BLK_DEV_IDEDISK=y
    CONFIG_IDEDISK_MULTI_MODE=y
    CONFIG_BLK_DEV_IDECD=y
    CONFIG_BLK_DEV_IDETAPE=m
    CONFIG_BLK_DEV_IDEFLOPPY=m
    CONFIG_BLK_DEV_IDESCSI=m
    CONFIG_BLK_DEV_IDEPCI=y
    CONFIG_IDEPCI_SHARE_IRQ=y
    CONFIG_BLK_DEV_IDEDMA_PCI=y
    CONFIG_IDEDMA_PCI_AUTO=y
    CONFIG_BLK_DEV_IDEDMA=y
    CONFIG_IDEDMA_AUTO=y
    CONFIG_BLK_DEV_IDE_MODES=y

Maybe I can pull the buffer cache and/or SHMEM changes.  

John