Date: Sat, 27 Feb 1999 14:13:23 +0100 From: Stefan Esser <se@mi.uni-koeln.de> To: Dennis <dennis@etinc.com> Cc: Ted Faber <faber@isi.edu>, hackers@freebsd.org, Stefan Esser <se@freebsd.org> Subject: Re: PCI Card Failures in 3.1 - FYI Message-ID: <19990227141323.A4953@dialup124.mi.uni-koeln.de> In-Reply-To: <199902261546.KAA20180@etinc.com>; from Dennis on Fri, Feb 26, 1999 at 10:55:12AM -0500 References: <199902261546.KAA20180@etinc.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 1999-02-26 10:55 -0500, Dennis <dennis@etinc.com> wrote: > Also, we dont call pci_map_port(), so I doubt that is the > problem. The mapping seems to work (when we map the > registers into memory we can read them and they have > the correct default values). the problem is that is cant > seem to dma to memory...its impossible to tell if it > can read however. You know about the bus-master enable bit in the command register ? There had been problems with cards doing bogus transfers if that feature was enabled by default, so the BIOS may leave it OFF. You are responsible for setting it in your driver, after the chip has been reset/initialzed and can be assumed under control ;-) The call to pci_map_port() exists for symmetry reasons. It does return the port address, but at least on the i386 does not have to setup any mappings. That's much different from pci_map_mem(), which establishes an VM mapping of the requested physical addresses and since there might be systems that provide only indirect access, to I/O ports, the same may be true for those architectures ... Regards, STefan To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19990227141323.A4953>