Stephen Gran - [EMAIL PROTECTED] wrote: > This one time, at band camp, Matthew Exon said: > >>/dev/hdb: >> setting using_dma to 1 (on) >> HDIO_SET_DMA failed: Operation not permitted > > > This is only likely to happen if the kernel module dealing with your ide > controller is unable to handle DMA. > > >>I've googled around for people who've run into similar problems, but it >>always seems to be caused by badly configured kernels. > > > And this is likely your problem as well. Please post the output of > lsmod & lspci. In assume that ide-generic is grabbing the ide controller > before the module that has DMA support is able to.
Thanks for the quick response. This provided me with enough clues to get a workaround, at least. I already had a vague idea that it was something to do with the order in which the modules were being loaded, so I was trying to compile the IDE modules into the kernel. But I had only targeted the modules shown by lsmod | grep ide. I didn't realise that there was a specific IDE driver too. I've now compiled via82cxxx into the kernel as well, and it works. Hooray! This is only a workaround though, because I'm having to compile my own kernel. I really want to use the stock Debian kernels, and these continue not to work. I don't know which package is causing the DMA problem, but I suspect it's an interaction between linux-image and initramfs-tools - I know that there were some problems with the initrd stuff in the 2.6.12 to 2.6.15 transition, so it's likely I'm seeing some residue of that. What do you think - should I take my bug to initramfs-tools? Bear in mind that this is all with stock kernels, nothing to do with compiling my own custom kernel or anything. The custom kernel is just a workaround. It's a real problem, not just me living on the bleeding edge and cutting myself :-) So there doesn't seem to be a bug in hdparm, except that the error message is very confusing. I'm not the only person to be totally confused by "Operation not permitted". It's more of a feature request, but it'd be nice if hdparm could catch this situation and print a more helpful error message, rather than just using perror. This is an upstream thing of course. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]