I have the same problem. And my guess is that it has to do something with the new south bridge over PCI floppy controllers. It's the same problem with the onboard PCI ac97 sound support. I think that the hardware and kernel resources gets overloaded or miss configured. Maybe there are also BIOS bugs in the ACPI support. It's a new thing for the motherboard envelopers and BIOS programmers (about only 6 or 8 years). So on some motherboards it could work well. So I think the kernel should get a ACPI correction layer so an AML, ASL debugger.
ACPI wiki: http://en.wikipedia.org/wiki/Advanced_Configuration_and_Power_Interface or more detailed in german: http://de.wikipedia.org/wiki/Advanced_Configuration_and_Power_Interface I did test it with fdformat and superformat on different drives also with jumpers and both prints different errors. Sometimes fdformat did get 1% formated and did break then with an ioctl error. I also think and it seams so too that data reaches the drive in a non continous way with very large delays between. Maybe a buffered DMA to PCI support for the controller could solve this problem and the module don't have to write to the drive directly, only to that buffer. Reading is no problem, the system does mount FDs and files can also be opened but writing to FDs fails. Datasheet for the Southbridge AMD SB600 is at: http://support.amd.com/us/Embedded_TechDocs/42119_sb600_ds_pub_3.07.pdf And how to handle it (should be handled from BIOS) is at: http://support.amd.com/us/Embedded_TechDocs/46157_sb600_bdg_pub.3.00.pdf like many other resources also at: http://www.coreboot.org/Datasheets It's ironic, because I just wanted to make a BIOS update because of its ACPI errors and the only way it works for me is the floppy drive (BIOS internal burn programm). Now it could be the BIOS ACPI setup itself why it doesn't work. Damn! Bastl -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
