From: Thomas S. <tho...@we...> - 2009-10-19 15:07:52
|
Konrad Rzeszutek Wilk wrote: > I am not familiar with libpciaccess, but I was wondering why that library > cannot realize it failed in its endeavour and use other means to accomplish > its goals? The thing is that the mmap() will succeed! Only the memory region will not be set up as write combining when PAT is disabled. So userspace would have to find out it PAT is enabled in kernel and accordingly set up MTRR entries. And currently I don't know of a kernel interface that would tell userspace if PAT is enabled. Kind regards, Thomas |