From: Rene B. <re...@we...> - 2002-08-08 20:36:05
|
On 2002.08.08 21:49 Ken Tyler wrote: > > Thats interesting. This does not happen on my machine, maybe its > relatet > > by some other hardware you use. But for freeing some of the dma memory > > pci_free_consistent is used, I'll "ifdef" it out and replace it by some > > other. > > I think its because I have no PCI or PCI_PERMEDIA mcompiled in. That could be the reason. > ___dma_getp has an #ifdef CONFIG_AMIGA that forces the use of > __get_free_pages (via a #define) instead of pci_get_consistent. > > ___dma_freep always uses pci_get_consistent. I thought it was not necessary for now to remove pci_free_consistent for Amiga, because its simply a wrapper for free_pages(). But this is no problem. > > Can you send me a dmesg? What version of kernel do you use and have you > > > compiled the 53c770 driver as module? Are you using my last update on > > "apus_setup.c"? > > A monolithic 2.4.18 kernel, I don't have your latest cache patch - I'll > get it. > > As for dmesg I don't have one at the moment because I added the needed > free_page call in ___dma_freep and I'm running the gcc 2.95.2 compiled > 2.4.18 test kernel at this instant. I can give you a dmesg when I return > to my normal + 53c770 config. > > The driver doesn't seem to be initialising or at least not finding the > drive in the test kernel. I'm wondering while ___dma_freep is called, but you don't see the chip initialising... Ciao, Renè |