[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: CMD64x IDE driver and DMA [was: Re: Installation Failing Sun V100]



Jurij Smakov wrote:

Hi,

On Thu, 6 Jan 2005, Dieter Jurzitza wrote:

Hi guys,
this is an issue I rose around October timeframe with my U60, this is what
SuSE does and this is what makes total sense (to me):
start with ide=nodma as default setting, check what your system can cope with
and activate dma later. I have seen severe filesystem corruption due the
malconfiguration regarding dma.


I understand that you can turn DMA on and off on the running system using hdparm (perhaps the /proc/ide interface as well), but how do you make sure (other than empirically) that it will work properly once activated? I presume that you are talking about the 2.4.x kernel, have you ever tried 2.6? According to Joshua Kwan its IDE driver works somewhat better.

I honestly do not understand why people do (debian sarge does) it the other way round. A stable system should be the primary target, after that the dma
configuration may easily be adjusted.


Perhaps because this opinion was not voiced before?

The debian installer could come with some tests that check for the proper dma
configuration interactively.


As turning the DMA off while booting requires passing the command-line arguments to the kernel, installer does not have any control over it whatsoever. The only thing we could do (short of hacking the driver in the kernel) is to document this workaround in the installer's manual for sparc. Any other ideas?

I think that proper documentation and FAQ's that relate to bugs is a great idea. I realize that there is a bug tracker, but maybe a better idea would be to have a simple FAQ page that notes each of this little nuances. I'm thinking I will have to switch back to i386 hardware until this stuff gets fixed. Thanks
Brandon



Reply to: