[Simh] RP vs RM header operations

[Rob Doyle]

On 6/27/2019 11:27 AM, Bob Supnik wrote:
> As I suspected, RP class drives (except the RP07) have 4 words of 
> header, and RM class drives have 2. This is confirmed by the RSTS/E disk 
> initialization code. RSTS/E can't format an RP07, so I can't verify how 
> those functions ought to work. RSTS/E seems to think it has 2 words of 
> header, though.
> 
> Any implementation of the write/write check/read header functions needs 
> to take care of the header words properly. Assume, for example, that 
> SALV is trying to write an entire RP06 track (20 sectors/track). It 
> would specify a Massbus <word> count (in 18b words) of 20 * (4 + 256) = 
> 5200 words. If write track plus header just defaults to being a normal 
> write, then 80 words will spill over to the start of the next track. 
> That might not matter if the next track is promptly overwritten, but on 
> the last track, the extra words will generate an addressing error, and 
> SALV will (probably ) fail.
> 
> For an RM, the count needs to be 30 * (2 + 256) = 7740 words. Again, if 
> the two extra words aren't accounted for, the last track write will 
> overflow and generate an error.
> 
> RSTS/E formats both RPs and RMs the same way: cylinder (plus 16b flag) 
> in the first word, track/sector in the second. For the RP, the third and 
> fourth are left as 0. There are bad block flags reserved in the header, 
> but all SimH disks are perfect. ;)
> 
> tl;dr. If you're going to implement these functions as other than stubs, 
> they better be done right, or other kinds of errors will surface.
> 
> /Bob
> 

Referring to 
<http://www.bitsavers.org/pdf/dec/disc/rp04_rp05_rp06/EK-RP056-MM-01_maint_Dec75.pdf> 
Figure 2-6:

As a clarification, it looks to me like the sector header data "words" 
are always 16-bit, regardless if the disk pack is formatted for 16-bit 
data or 18-bit data.  I suppose that makes sense so that either type of 
system can read the sector header correctly.

Rob