Sysgen neither copies nor doesn't copy any specific components. Basically, the DR standard Sysgen for CP/M 2.2 copies the contents of the first two tracks (52 sectors of 128-bytes each) to and from a buffer at 900H. It doesn't know or care what components are present in those tracks, it just does a blind copy: 52 sectors (128 bytes per sector, two tracks of 26 sectors each) to and from memory at address 900 hex. [This address was different [lower] in some earlier versions of CP/M]. Now that was the standard Digital Research SYSGEN. If you are dealing with an OEM system or any system that used any flavor of 5.25" media, then all bets are off. In such systems the entire methodology for system generation, booting and copying the boot code were entirely up to the OEM and could be extremely different from the DR standard (which was both simplistic and quite restrictive). On my own implementations for the Z-100, I put the OS and the BIOS onto the data area as files OS.SYS and BIOS.SYS (which did not have to be either contiguous or located in special areas). (For MP/M-86, the OS consisted of about a half-dozen different modules.) What I put into the system tracks was an OS loader program that could actually read files using the directory and FCBs. And for writing and copying that, I created a program called "WRITLDR" (write loader). The loader had a CP/M-like boot sector and bios of it's own, but since the loader itself was relatively small, the loader's bios could be relatively huge, allowing for the complexity required to deal with systems with a large number of bootable disk formats.