Hi, David Niklas wrote: > Well there is a kernel dev mailing list.
You mean Linux Kernel Mailing List ? That's a nice place to get your head bitten off. At least if you have not done all reasonable research about your problem and invested some ingeniosity. Beware. Just thinking about posting to LKML can have severe conesquences: In december 2005 i prepared a petition, to ask for CD TAO burning in the kernel after yet another flame war between cdrecord's author and the kernel eminencies. The research yielded that there was a half-dead library project called libburn. In august 2006 i found myself as main developer of a fork. Not actually a fork, because the old branch did not move on. I never finished that petition. For the multi-SG_IO problem one would have to find out where the change to the worse was applied, by whom and for what reason. >From a short expedition into the antique ex-Linux SCSI and IDE drivers of Hurd, i know that it's all queue driven. A nightmare to follow in the code. today's code is probably much more advanced and artful. Have a nice day :) Thomas