A Couple of minor DASD related changes

[SE20225]

Is there any plan or wish for providing support for the Extended Addressing Volumes
'EAV' in Hercules? Not that I have any demanding use for them but more for my
self-education. As long as you do not fill them with used tracks, I expect that
the compression mechanism will work OK as long as the max file size is not
exceeded. I am using DASDINIT64.

While trying to make minor changes to allow much higher cylinder numbers for 3390
I found one defect and one rather minor/cosmetic mishap.

1. I found that Hercules does not fully validate that the begin and end extent
   CCHH values specigied for the DX (0X63 CCW opcode) are not fully validated.
   I could specigy a head number greater than the actual number of heads in a track
   and most likely also a too high cylinder number. The documentation clearly
   specifies that the data should be valid track addresses for the device used.

The change is very simple:
In CKDDASD.C the DefineExtent routine:
After the ckdxbcyl to ckdxehead variables are filled, add the following:

/* Verify that the extent is described via VALID DASD track addresses */
if (bcyl  >= dev->ckdcyls   ||
    ecyl  >= dev->ckdcyls   ||
    bhead >= dev->ckdheads  ||
    ehead >= dev->ckdheads )
{
    ckd_build_sense(dev, SENSE_CR, 0, 0, FORMAT_0, MESSAGE_4);
    *unitstat = CSW_CE | CSW_DE | CSW_UC;
    return false;
}

2. The response to the RDC (0x64) CCW will indicate the presence of alternate
   tracks when the underlying DASDINITed volume was created with alternate tracks
   also when it is attached via a 2105 or 2107 CU. These 'emulated' volumes
   never have alternate tracks.

In DASDTAB.C the CKD the dasd_build_ckd_devchar routine
Modify the statements that indicate alternate tracks to:

    /* Emulated 3390 volumes do not have alternate tracks                    */
//    if (altcyls > 0)
    if (altcyls > 0 &&
       !(cu->devt == 0x2105 || cu->devt == 0x2107)
    )
    {
        store_hw(devchar+28, cyls - altcyls);   // Alternate cylinder & tracks
        store_hw(devchar+30, altcyls * ckd->heads);
    }

Anders Edlund andersedlund@telia.com