rimtech
October 9th, 2006, 06:08 PM
I am a recent gentoo convert, that is, i used to use gentoo but no more, now I'm an ubuntu boy...
So, a while back I was using gentoo and I installed it on my Sub Blade 100... but lo and behold... it simply would not boot with DMA support. Every single boot would have to be appended with ide=nodma. So... after a long time of running in PIO mode only which is very very slow, and after I had recently discovered ubuntu on my desktop, I decided to give it ago on my blade 100... and so i did, but... again, I had to use ide=nodma on each boot...
I came to my own conclusion and from reading various documents on the internet/google that apparently this system has a specific IDE controller.... the ALI15X3, of which there is support in the linux kernel but on sparc64 systems there is data corruption when DMA is enabled.
More research has yielded me further results:
http://www.ussg.iu.edu/hypermail/linux/kernel/0307.2/0620.html
Apparently whoever posted this had discovered a solution for 2.4 kernels..., why hasn't this made it's way into the kernel source tree and WHY oh WHY doesn't ubuntu have this patch for the 2.6 version, and more specifically the sparc64 kernel?
What would be really great is if an avid developer were to read this and decide to modify the ALI15X3 driver for sparc64 kernels to encorporate this change, as it is clearly a kernel defect. This is my dream.
OR...
What would be nice also is if someone would post how i can do it myself, considering i am not a developer and have no idea how to turn a 2.4 kernel patch into a 2.6 kernel patch.
Kindest regards to whoever reads this....
:)
So, a while back I was using gentoo and I installed it on my Sub Blade 100... but lo and behold... it simply would not boot with DMA support. Every single boot would have to be appended with ide=nodma. So... after a long time of running in PIO mode only which is very very slow, and after I had recently discovered ubuntu on my desktop, I decided to give it ago on my blade 100... and so i did, but... again, I had to use ide=nodma on each boot...
I came to my own conclusion and from reading various documents on the internet/google that apparently this system has a specific IDE controller.... the ALI15X3, of which there is support in the linux kernel but on sparc64 systems there is data corruption when DMA is enabled.
More research has yielded me further results:
http://www.ussg.iu.edu/hypermail/linux/kernel/0307.2/0620.html
Apparently whoever posted this had discovered a solution for 2.4 kernels..., why hasn't this made it's way into the kernel source tree and WHY oh WHY doesn't ubuntu have this patch for the 2.6 version, and more specifically the sparc64 kernel?
What would be really great is if an avid developer were to read this and decide to modify the ALI15X3 driver for sparc64 kernels to encorporate this change, as it is clearly a kernel defect. This is my dream.
OR...
What would be nice also is if someone would post how i can do it myself, considering i am not a developer and have no idea how to turn a 2.4 kernel patch into a 2.6 kernel patch.
Kindest regards to whoever reads this....
:)