On 1/17/2007 2:16 PM, Matt Taggart wrote: > "James Love" writes... >> They use a HP-proprietary TEAC SCSI floppy disk, which appears to be an >> IDE drive with a SCSI-IO/FDC board piggy-backed on it. Any idea whether >> the TEAC drive is the one that HP won't provide the documents for? > > I hope you are able to make progress on a driver, that would be cool. But I > don't think anyone from HP is going to be able to help. :( > >> Regardless, the drive works in MKLinux, so hopefully someone knows >> something about how it works... > > Yeah it sounds like there is hope of getting it working then. > On 1/24/2007 11:28 AM, Grant Grundler wrote: > On Tue, Jan 23, 2007 at 03:51:32PM -0500, James K. Love wrote: >> In an attempt investigate further, I installed MKLinux on a 715, and the >> scsi floppy drive was supported through some combination of the MACH >> kernel and it's hosted Linux kernel. I've looked through the MKLinux >> source code, and it appears that the MACH scsi driver had some support >> for flexible disk drives (which I assume is the floppy drive), but I'm >> uncertain if this is the only missing piece to the puzzle. If anyone >> has any additional info/insight, I'd really appreciated it. > > If you can figure out what's missing between linux and MKlinux there > is a good change it could get added. The closer you get to providing > a patch, the more likely it will get included. > I finally figured some of this out. I ported the MACH flexible scsi disk code over to the 2.6 kernel, and I can now mount/read/write with my HP TEAC FC-1 drive. The modifications I made were inside the scsi device (sd) driver, which is similar to where they were inside the MACH kernel. This is obviously not ideal, since the device major is for a scsi disk, but the drive is really a floppy. This breaks some user-space floppy apps like fdformat. There really should be a new floppy driver similar to the cdrom driver, but these changes may not work with any other scsi floppy drive (are there any others?). I only have a few HP OEM TEAC revs to test with here. Bottom line... The scsi disk driver could be patched, but it may only provide partial/experimental support for this drive without extensive changes. I'm traditionally a user-space developer, so I'd appreciate any advice/thoughts before I get knee deep. Also, if someone could pass along a few links to some procedural docs about properly submitting kernel code/patches, I'd really appreciate it. Thanks, James L.