I am doing porting work; hence I am not referring to any existing architecture. The reason for my questions was to understand how and where does Linux insures that userspace addresses do not conflict with kernelspace addresses within a process address space ? In other words, when a userspace virtual address mapping is created in a process address space, how and where does Linux insures that it does not conflict with addresses returned by kmalloc() ? Searching the Internet returned that kmalloc() always return an address in low-memory, and that it is always a 1-to-1 mapping to physical memory; while userspace virtual address mapping reside in high-memory; however, I could not find where in the Linux kernel the low-memory or high-memory mapping range is defined. Any help understanding how Linux insures that userspace addresses do not conflict with kernelspace addresses would be much appreciated ? On Thu, Nov 14, 2019 at 3:55 PM Jonathon Reinhart <jonathon.reinhart@xxxxxxxxx> wrote: > > What architecture are you referring to? > > "High memory" and "Low memory" apply to x86_32 (see arch/x86/Kconfig), > but not to x86_64, where all physical memory is directly usable by the > kernel (see Documentation/x86/x86_64/mm.rst). > > On Thu, Nov 14, 2019 at 4:30 PM William Tambe <tambewilliam@xxxxxxxxx> wrote: > > > > How to tell whether an address is from low-memory or high-memory ? > > > > And are addresses in low-memory always a 1-to-1 mapping to physical memory ?
