***************************************************************** * * * NOTE: IF YOU USE THE DISKSPARE.DEVICE FOR YOUR * * PROGRAMS YOU HAVE TO COPY THIS DOC.FILE TOO !!!! * * * * Note: * * I can only support your program in future releases * * of the device, if you send me a FREE copy. * * * ***************************************************************** TABLE OF CONTENTS diskspare.device/CMD_Remove diskspare.device/CMD_GetGeometry diskspare.device/CMD_RawRead diskspare.device/CMD_RawWrite diskspare.device/CMD_Eject diskspare.device/Open The other commands work like the corresponding trackdisk cmds. KNOWN BUGS: * The value of dg_CylSectors returned by GetGeoMetry is wrong. Fixed in V2.0. * The Device doesn't support ETD_Cmds correctly. Fixed in V2.1. * The Device returns an Error IOERR_BADADDRESS if the adress of IO_DATA is odd (PFS only). Fixed in V2.1. * If remove is called twice, the System crashes. Fixed in V2.1. * Error in CloseDevice: Maybe a RemPort on a Port which wasn't added -> Crash (caused by register mistake) Fixed in V2.1 * Error in TD_FORMAT: Doesn't do an Update but flushes the trackbuffer. (PFS only) Fixed in V2.2 * It was possible to read/write beyond the last track. Fixed in V3.0 * Error in CMD_READ: Trackbuffer not cleared although disk changed. (PFS only) Fixed in V3.0 * Enforcer hit in initport. Fixed in V3.0 * Error in WaitIO: under some strange conditions, there might be write errors (signalbits not cleared! V3.0 only) Fixed in V3.1 NAME Remove FUNCTION If you are not the FastFileSystem, and I guess you aren't, NEVER call Remove ! It's no use calling Remove !! The best thing is to forget that there's a cmd called Remove !!! IF you call this forgotten cmd and the DD<>HD autoswitch is enabled (Bit 2 in Flags is set), the system WILL crash !!!! If you want to add a diskchange-int, use the addchangeint/remchangeint cmd instead. Why was this patch necessary ? Because there's a restriction in the FFS: It checks, if the devicename is 'trackdisk.device' and if it isn't, GetGeometry won't be called. I tried to fix the disksize in the dosenvec-struct and all programs like FixDisk, DiskSalv, Format etc worked fine. BUT the FFS said 'Not a DOS Disk'... SEE ALSO Includes/exec/alerts.i BUGS If Remove was called twice, the System crashed. Fixed in V2.1. NAME GetGeometry FUNCTION Works like trackdisk. There are two possibilities for Sectors per Track: 12 sectors/track and 24 sectors/track. You have to check this after each diskchange ! New for V3.0: You get always the right disksize. The HD mode always works (with HD drives). Please check version number. SEE ALSO AutoDocs/trackdisk.doc BUGS There was a bug in the first release (V1.6): The value returned in dg_CylSectors was wrong. I forgot an 'ext.l', so the contents of the high order word was rubbish. Fixed in V2.0. NAME RawRead FUNCTION No Function ! Returns IOERR_NOCMD. Use trackdisk device for rawread. NAME RawWrite FUNCTION No Function ! Returns IOERR_NOCMD. Use trackdisk device for rawwrite. NAME Eject FUNCTION No Function ! Returns IOERR_NOCMD. Use trackdisk device to eject (if you've got a drive which ejects). NAME OpenDevice FUNCTION Opens the device with a drive-specific set of flags. There's something special about the flags: If the device wasn't opened yet, everything works as usual. If the device was already opened (via Mount) your flags will be ignored, but the device won't return an error. (Some well known programs crash when an OPENFAIL was returned ??? They obviously don't use the fssm_Flags when opening AND don't check for errors ?¿?) The active flags are returned in IO_ACTUAL. So you can check whether the flags fit your needs or not. Why ? If the user has mounted the device with HighCyl = 79 then it's his decision not to use it and you HAVE TO accept that. It's the users Amiga not yours. The minimum you get is: heads: 2 tracks: 160 sectors/track: 12/24 bytes/sector: 512 # of sectors: 1920/3840 total bytes: 983040/1966080 SEE ALSO AutoDocs/exec.doc/opendevice FLAGS BIT 0 0: disable diskchange for DFX: 1: enable diskchange for DFX: Works only when mounting the device with the FastFileSystem. When you open the device and don't call the forgotten cmd the diskchange-int of the corresponding trackdisk drive won't be affected. BIT 1 0: HighCyl is Cyl. 79 1: HighCyl is Cyl. 81 When set to '1' trackdisk WILL be patched everytime a track is read/written from/to disk! BIT 2 0: DD and HD mode are now available. That's new. Check the device version. 1: Autodetection of DD/HD-disks for the FastFileSystem, too. It is patched. Call GetGeometry to get the right disksize. See above. The patch for the ffs is in the remove cmd. So if you open without bit 1 set and don't call the forgotten cmd NOTHING is patched !! Version 2.2: BIT 3 0: Use any Mem for trackbuffer 1: Try to use Mem from $600000 - $A00000 If there's no memory at this location any other mem is used. Why ? I've got an A2000 with A2630 (which has it's mem at $200000-$5FFFFF) and a 16bit Ramcard at $600000 - $A00000. So why should I use the 32bit ram (which is prefered when calling allocmem) for a trackbuffer ? This bit may be useful for A3/4000 with 16bit ramboards too. Version 3.0: BIT 3 isn't used. Version 3.0: BIT 4 0: Normal write mode. 1: Verify write mode. Each write will be verified by the device itself. This nearly half the write speed but you're sure the data on disk is ok. Useful with cheap disks. BIT 5-31 are currently unused and RESERVED. NOTES HD-Mode: To check if the drive is a highdensity one you may call GetUnitID from the disk.resource. Here are the Results: No Disk in any Drive -> returns 0x00000000 Any Disk in a DD Drive -> returns 0x00000000 A DD-Disk in a HD Drive -> returns 0x00000000 A HD-Disk in a HD Drive -> returns 0xAAAAAAAA No Drive connected -> returns 0xFFFFFFFF As you can see, it's not possible to check if the drive is a HD one, if there's no HD Disk in it. Don't care about that. Simply call GetGeometry to get the right disksize. BUT it may be differently for each disk (DD/HD). AddChangeInt: If you've already added a ChangeInt with the trackdisk.device, you needn't add one with diskspare.device because it's directly passed through to the trackdisk.device. Inhibit: It's a good idea to inhibit the corresponding trackdisk drive (DFX) when accessing the disk via diskspare.device. Also check for mounted diskspare drives. Don't rely on the names DS0: to DS3: when you inhibit the DOS. The device itself can't inhibit the DOS because a task isn't allowed to call a dos.lib function. General: The main purpose of this device is to be mounted with the FastFileSystem. So always remember that you may not be the only accessor !!