Shachar Kaufman wrote: > With your helpful guidance I've been able to advance. > > Eventually I came into a new problem: > The platform I'm working on has a memory mapped "window" and cannot have > more than one HW surface accessible to the CPU. My plan is to have one > large HW buffer which devmem will manage (actually the core surface pool > will manage) as a linear chunk of memory in which it can define as many > non-layer-tied surfaces as it needs. In addition, I have a surface pool > manager for layer-tied surfaces, where each will be allocated a buffer. > Both types of HW surfaces would be accessible to the accelerator, but > only one HW buffer can be mapped to the CPU accessible window at any one > point in time. > > So, before the CPU access any surface (for a software fallback rendering > operation for instance) I need to be given a notice, to point the HW > mapping window to the appropriate buffer. > > Does DFB support my case? Is this what "surface lock" is for?
Yes, the pool receives the Lock() request before the CPU does a read/write and it receives the Unlock() afterwards. -- Best regards, Denis Oliver Kropp .------------------------------------------. | DirectFB - Hardware accelerated graphics | | http://www.directfb.org/ | "------------------------------------------" _______________________________________________ directfb-users mailing list directfb-users@directfb.org http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
