Race conditions on PCI with Intel processors

Question about which tools to use, bugs, the best way to implement a function, etc should go here. Don't forget to see if your question is answered in the wiki first! When in doubt post here.
Post Reply
rdos
Member
Member
Posts: 3306
Joined: Wed Oct 01, 2008 1:55 pm

Race conditions on PCI with Intel processors

Post by rdos »

It seems like I have some tricky race-conditions with PCI-busmasters like AHCI discs and ethernet cards. I know the drivers work as they should because they work with AMD-processors. It seems like there is similar symptoms on at least 3 different modern machines with multicore Intel-processors. It seems like the hardware announces the presence of new data with an interrupt, but then the data read-out from memory isn't correct.

What difference between Intel and AMD processors (or possibly chipsets) could account for such problems?
Top
rdos
Member
Member
Posts: 3306
Joined: Wed Oct 01, 2008 1:55 pm

Re: Race conditions on PCI with Intel processors

Post by rdos »

At least on one of the machines the AHCI driver apparently takes a physical address above 4G and truncates it into a address below 4G and passes it to the SATA-controller. Not good. :oops:
Top
Post Reply

Return to “OS Development”