On Tue, 2007-04-17 at 19:00 +0530, Prasad Joshi wrote: > Hi All, > > I am confused with the ZONEs and HIGH memory. > > > Memory within ZONE_NORMAL is directly mapped by the kernel into the > upper region of the linear address space. This is the physical memory > directly accessible by the kernel and is related to the > kernel/userspace split in the linear address space marked by > PAGE_OFFSET. > > > Why ZONE_NORMAL is directly mapped into upper region of the linear > address space? I think here PAGE_OFFSET (3GB) indicates the upper > region of the linear address space ? So that the kernel can _always_ access mem within ZONE_NORMAL. So, there is a possibility that a page in ZONE_NORMAL is mapped in userspace to (allocated to a process). So, then that page frame then has two mappings one for the process and the other for the kernel itself. > > > ZONE_HIGHMEM is the remaining available memory in the system and is > not directly mapped by the kernel. > > > Generally, > ZONE_DMA First 16MiB of memory > ZONE_NORMAL 16MiB – 896MiB > ZONE_HIGHMEM 896 MiB – End > > Now, the previous para says ZONE_NORMAL is at PAGE_OFFSET (3GB) but, > this table indicates it at 16 MB ? > > Because the address space usable by the kernel (ZONE_NORMAL) is > limited in size, the kernel has support for the concept of high > memory. Two thresholds of high memory exist on 32-bit x86 systems, one > at 4GiB and a second at 64GiB. The 4GiB limit is related to the amount > of memory that may be addressed by a 32-bit physical address. To > access memory between the range of 1GiB and 4GiB, the kernel 896 MB to 4GB actually... > temporarily maps pages from high memory into ZONE_NORMAL with kmap(). correct... > > > What is the size of the ZONE_NORMAL? I think it is 1GB - 128 bytes. 1GB - 128MB ;) .. > > This is what I have concluded > 1. ZONE_NORMAL is from 16MB - 896MB. ==> address space usable by > the kernel Thats the physical memory, which is mapped from PAGE_OFFSET + 16MB. > 2. High memory is from 1GB-4GB ==> to access this part of memory > kernel has to temporarily map pages from high memory into ZONE_NORMAL. > So, ZONE_HIGHMEM must be from 1GB-4GB and not from 896 MiB – End HIGHMEM is 896MB - 4GB (if u have 4GB ram). Which is mapped runtime by the kernel if needed using kmap/kmap_atomic. The virtual addresses for these temporary mappings for the kernel will always be greater than PAGE_OFFSET + 896MB, i.e in the last 128Mb of the virtual address space available. > 3. I am not getting at all how kernel/userspace split is related to > ZONE_NORMAL and ZONE_HIGHMEM? It is related in a way, if you have 2G/2G user/kernelspace split, then you can have more physical memory in ZONE_NORMAL, i.e it will be 2GB - 128MB if HIGHMEM enabled. This way the kernel is able to access more physical memory directly (since ZONE_NORMAL is mapped directly), than having to go through the overhead of kmap()/kmap_atomic(). > > I think I have put so many questions, please educate me. Thanks in > advance. > cheers, sandeep -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
