Erik Hemdal wrote:
The original question remains, though. Why do we need nodma to check the integrity of only two CD's out of the CD set, when every other CD ever used on the system is readable with DMA on? Maybe the real question is , "What's the specific failure mode going on here?"
Erik
As a suggestion to tracking down the "failure mode". If the ISO CDs are burned using the command line cdrecord...were they burned with or without the "padsize=33" option?
If without, maybe someone who is having the failure could use the same hardware but try burning the same ISO file with cdrecord with the option "padsize=33" (and without the option "-pad").
I have seen speculation that the problem is due to the dma overrunning the end of the ISO and not recovering from the "error" correctly. Adding that much pad might then bypass the occurrence of the error.
WARNING: This is ONLY SPECULATION
John D.
