can not cross a page boundary. This fixes a problem when a dump(8) buffer
just crossed a page boundary and bus_dmamap_load() would coalesce the
the overlap onto the first segment. The NEXTPTR value would get set to
-1 when starting the DMA transfer and the residual data for unaligned
transfers would be copied to physical address 0x1ffffffc. On the R3000,
this would result in corrupted data, and on the R4000 would usually result
in a hard system hang.
both the primary and the secondary caches. MachFlushDCache() will not
invalidate the correct secondary cache lines for KSEG2 addresses.
R4000 systems should now be able to boot and run.
unaligned transfers and adjust the physical address to align the transfer.
If the buffer end just crossed the page boundary, the computation of the
NEXTPTR physical address resulted in using -1. The cleanup at the end of
the DMA transfer would try to copy the residual data to physical address
0x1ffffffc. This would silently corrupt data on the R3000 and usually
would hang the R4000.
.De's were really .Ds's. Problems were causing output not to be rendered-
number of output pages more than doubled after these fixes, and made much more
sense.
floppies as long as they fit, but they are the first to be sacrificed
if space is short, and get moved to a seperate rescue floppy.
This means that the default fdset and the 2.88M image have them,
but others (the "small" and "tiny" floppies) do not.
Sysinst is also back on the "tiny" image, and tested to be working
within 4M.
enabling and disabling swap. Enabling swap is currently only
done by the i386 port on systems with <= 8M of physical memory.
If a user re-enters the install procedure through the main menu,
and the target disk has an active swap partition, try to disable it,
warning the user that this might lead to 'out of swap' problems,
making a restart necessary. This should not happen very often.
Partly based on comments by Simon Burge.
leo