Re: SC1200 failure in 2.6.23 and 2.6.24-rc1-git10

["Denys Fedoryshchenko" <denys@visp.net.lb>]

Does it work as kernel parameter?

I tried libata_dma_mask=0x4 and to set 0xf or 0xff - doesn't help. How to 
disable DMA in libata, if it is compiled in kernel?

On Thu, 8 Nov 2007 01:30:53 +0100, Bartlomiej Zolnierkiewicz wrote
> On Thursday 08 November 2007, Denys Fedoryshchenko wrote:
> > 2.6.24-rc2 not working very well
> > 
> > 
> > dmesg
> > [   12.386395] Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2
> > [   12.405579] ide: Assuming 33MHz system bus speed for PIO modes; 
override 
> > with idebus=xx
> > [   12.430441] SC1200: IDE controller (0x100b:0x0502 rev 0x01) at  PCI 
slot 
> > 0000:00:12.2
> > [   12.454070] SC1200: not 100% native mode: will probe irqs later
> > [   12.471947]     ide0: BM-DMA at 0xfc00-0xfc07, BIOS settings: hda:pio, 
> > hdb:pio
> > [   12.493873]     ide1: BM-DMA at 0xfc08-0xfc0f, BIOS settings: hdc:pio, 
> > hdd:pio
> > [   12.515810] Probing IDE interface ide0...
> > [   12.528810] Clocksource tsc unstable (delta = -497423729 ns)
> > [   12.545888] Time: pit clocksource has been installed.
> > [   12.563379] hda: SanDisk SDCFH-1024, CFA DISK drive
> > [   12.578340] hda: applying conservative PIO "downgrade"
> > [   12.593869] hda: host max PIO4 wanted PIO255(auto-tune) selected PIO1
> > [   12.594006] hda: MW DMA 2 mode selected
> > [   12.594297] ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
> > [   12.608778] Probing IDE interface ide1...
> > [   12.623192] hda: max request size: 128KiB
> > [   12.635322] hda: 2001888 sectors (1024 MB) w/1KiB Cache, CHS=1986/16/
63, 
> > DMA
> > [   12.657134]  hda:<4>hda: dma_timer_expiry: dma status == 0x21
> > [   12.865846] hda: DMA timeout error
> > [   12.876092]  ide_dma_end dma_stat=21 err=1 newerr=0
> > [   12.890753] hda: dma timeout error: status=0x58 { DriveReady 
SeekComplete 
> > DataRequest }
> > [   12.914977] ide: failed opcode was: unknown
> > [   12.927743] hda: DMA disabled
> > [   12.937035] ide0: reset: success
> > [   12.948324]  hda1
> > 
> > Mounting taking long time on 1GB card cause of DMA issues. In dmesg i am 
not 
> > sure about timestamp showing few seconds, in real life it took about 2 
> > minutes.
> 
> Please try booting with "hda=nodma".
> 
> It could be a hardware problem (CF adapter without DMA lines).
> 
> Thanks,
> Bart


--
Denys Fedoryshchenko
Technical Manager
Virtual ISP S.A.L.