Name
Ioremap_nocache-map memory to CPU Space
Synopsis
Void _ iomem * ioremap_nocache (unsigned longPhys_addr, Unsigned longSize);
Arguments
-
Phys_addr
-
Physical address to be mapped
-
Size
-
Size of the resource to be mapped
Description
Ioremap_nocache performs a series of platform-related operations so that the CPU can be accessed through Io functions such as readb/readw/readl/writeb/writew/writel.
The returned address cannot be directly accessed as a virtual address..
[Translator's note: In the use process of the translator, there is no direct access from a virtual address, maybe not on some platforms. The translator's platform is x86 and ixp425]
This version of ioremap ensures that the CPU in these instances is not buffered, just like the existing buffer rules on the PCI bus. Note: There are still other buffers and caches on many bus. In some special drivers, the author should read data during PCI writing.
This is useful for some control registers in areas that do not require composite write or buffer read.
The returned ing address must be released using iounmap.
It has been plagued by the high-speed buffer of the CPU. It is hard to find this function and finally solves the problem.
Used to ensure that data is written to the memory every time