Ciekawy wpis o 'naprawie' tzw 'BAD DMA' softwareowo:
https://atariscne.org/news/index.php/at ... n-software
I jeszcze tu:
https://www.pouet.net/prod.php?which=104445
For those not in the know. Earlier Atari STe dma chips had some different behaviour in certain cases than the later chips. For a long time these earlier chips were deemed 'bad' dma. Only recently due to decapping and analysis of the chip the community found out the subtle hardware difference between the two versions. As a result, by using the 'bad' dma chip version in an other way (in code) it turns out the chip behaves as intended. This release contains a binary patch for one of the better known hard disk drivers for the Atari, such that this hard disk driver now also works correctly with the 'bad' dma. And thus, we achieve DMA bliss!
https://www.chzsoft.de/site/hardware/ne ... stigation/
This investigation has found the root cause of the bad DMA phenomenon - hopefully ending years of speculation and guesswork:
* The errors while writing to an ACSI disk in some STEs with the old C025913 DMA IC are caused by borderline timing of the RDY signal between DMA and GSTMCU, which in turn causes some bytes not to be written to the disk. This corrupts the disk's contents.
* The newer C398739 DMA IC is not just better by chance but does indeed contain a specific fix for this problem: a change in the timing of the RDY signal.
* The previously published workarounds (replacing the CPU, replacing the power supply, changing pull-up resistors), while potentially beneficial to a particular machine, are just a short-term fix for a much deeper problem
* There is a potential software workaround, which should be further investigated and validated.