Factually, QubATA driver when started from ROM will take 16128 bytes of RAM to copy its code from ROM to RAM (for faster access) and 1024 Bytes to build the driver linkage block (Directory Driver Definition Block) - This is a fixed size - And after it will try to link in the first partition by loading its FAT in RAM (size depends on partition size and on block size used for partition). Also QDOS will allocate a physical definition block for each win drive (256 bytes each for QubATA).Silvester wrote:It is not a bug, it is due to v2.17c Eprom handling of RES_128.
The original v1.xx Eprom checks RAM size, if <=128K it just links in the FLP driver, if >128K it links in everything. The problem is v2.17c just links in everything, regardless of RAM size found.
But even if the v2.17c code is modified (ie. FLP only if RES_128), you may still have problem if QubATA driver uses enough of the 128K RAM to upset things similarly.
Simplest solution is to use Minerva where you can selectively boot 128K and ignore add-ons (leaving MDVs only).
I took a look on my Tetroid and found that QubATA init is really done after a res_128 boot. So it grabs memory as after a normal boot.
I will be able to modify the QubATA init routine to don't start when RAM is <128 as there is no big chance to load any FAT under 128 KB of whole RAM system. However I found that FLP driver is also started after res_128 boot. So, deactiving QubATA will not be enough to retrieve initial TC V1 behavior.
Alain