Advanced Systems Programming - Lesson 12: Linux Memory Issues

Linux Memory Issues An introduction to some low-level and some high-level memory management concepts Some Architecture History • 8080 (late-1970s) 16-bit address (64-KB) • 8086 (early-1980s) 20-bit address (1-MB) • 80286 (mid-’80s) 24-bit address (16-MB) • 80386 (late-’80s) 32-bit address (4-GB) • 80686 (late-’90s) 36-bit address (64-GB) • Core2 (mid-2000s) 40-bit address (1-TB) ‘Backward Compatibility’ • Many buyers resist ‘early obsolescence’ • New processors need to run old programs • Early design-decisions leave their legacy • 8086 could run recompiled 8080 programs • 80x86 can still run most 8086 applications • Win95/98 could run most MS-DOS apps • But a few areas of incompatibility existed Linux must accommodate legacy • Legacy elements: hardware and firmware • CPU: reset-address and interrupt vectors • ROM-BIOS: data area and boot location • Display Controllers: VRAM & video BIOS • Support chipsets: 15MB ‘Memory Window’ • DMA: 24-bit memory-address bus • SMP: combined Local and I/O APICs Other CPU Architectures • Besides IA-32, Linux runs on other CPUs (e.g., PowerPC, MC68000, IBM360, Sparc) • So must accommodate their differences – Memory-Mapped I/O – Wider address-buses – Non-Uniform Memory Access (NUMA) Nodes, Zones, and Pages • Nodes: to accommodate NUMA systems • However 80x86 doesn’t support NUMA • So on 80x86 Linux uses just one ‘node’ • Zones: to accommodate distinct regions • Three ‘zones’ on 80x86: – ZONE_DMA (memory below 16-MB) – ZONE_NORMAL (from 16-MB to 896-MB) – ZONE_HIGHMEM (memory above 896-MB) Zones divided into Pages • 80x86 supports 4-KB page-frames • Linux uses an array of ‘page descriptors’ • Array of page descriptors: ‘mem_map[]’ • physical memory is ‘mapped’ by CPU How 80x86 Addresses RAM • Two-stages: ‘segmentation’ plus ‘paging’: – First: logical address  linear address – Then: linear address  physical address • CPU employs special caches: – Segment-registers contain ‘hidden’ fields – Paging uses ‘Translation Lookaside Buffer’ Logical to Linear selector GDTR segment-register operand-offset virtual address-space memory segment global descriptor table base-address and segment-limit descriptor Segment Descriptor Format Limit[ 15..0 ]Base[ 15..0 ] Base[ 23..16 ]Base[ 31..24 ] Limit [19..16 ] 31 0 Linear to Physical physical address-space offsettable-index linear address CR3 dir-index page frame page directory page table CR3 and CR4 • Register CR3 holds the physical address of the current task’s page-directory table • Register CR4 was added in the 80486 so software would have the option of “turning on” certain advanced new CPU features, yet older software still could execute (by just leaving the new features “turned off”) Example: Page-Size Extensions • 80586 can map either 4KB or 4MB pages • With 4MB pages: middle table is omitted • Entire 4GB address-space is subdivided into 1024 4MB-pages Demo-module: ‘cr3.c’ creates a pseudo-file showing the values in CR3 and in CR4 Linear to Physical physical address-space offset linear address CR3 dir-index page frame page directory 4-MB page-frames PageTable Entry Format Frame Address 031 Frame attributes 1112 Some Frame Attributes: P : (present=1, not-present=0) R/W : (writable=1, readonly=0) U/S : (user=1, supervisor=0) D : (dirty=1, clean=0) A : (accessed=1, not-accessed=0) S : (size 4MB = 1, size 4KB = 0) Visualizing Memory • Our ‘pgdir.c’ module creates a pseudo-file that lets users see a visual depiction of the CPU’s current ‘map’ of ‘virtual memory’ • Virtual address-space (4-GB) – subdivided into 4MB pages (1024 pels) – Text characters: 16 rows by 64 columns Virtual Memory Visualization • Shows which addresses are ‘mapped’ • Display granularity is 4MB • Data is gotten from task’s page-directory • Page-Directory location is in register CR3 • Legend: ‘-’ = frame not mapped ‘3’ = r/w by supervisor ‘7’ = r/w by user Assigning memory to tasks • Each Linux process has a ‘process descriptor’ with a pointer inside it named ‘mm’: struct task_struct { pid_t pid; char comm[16]; struct mm_struct *mm; /* plus many additional fields */ }; struct mm_struct • It describes the task’s ‘memory map’ • Where’s the code, the data, the stack? • Where are the ‘shared libraries’? • What are attributes of each memory area? • How many distinct memory areas? • Where is the task’s ‘page directory’? Demo: ‘mm.c’ • It creates a pseudo-file: ‘/proc/mm’ • Allows users to see values stored in some of the ‘mm_struct’ object’s important fields Virtual Memory Areas • Inside ‘mm_struct’ is a pointer to a list • Name of this pointer is ‘mmap’ struct mm_struct { struct vm_area_struct *mmap; /* plus many other fields */ }; Linked List of VMAs • Each ‘vm_area_struct’ points to another struct vm_area_struct { unsigned long vm_start; unsigned long vm_end; unsigned long vm_flags; struct vm_area_struct *vm_next; /* plus some other fields */ }; Structure relationships VMA VMA VMA VMA VMA mm_struct task_struct *mm *mmap Linked list of ‘vm_area_struct’ structures The ‘process descriptor’ for a task Task’s mm structure Demo ‘vma.c’ module • It creates a pseudo-file: /proc/vma • Lets user see the list of VMAs for a task • Also shows the ‘pgd’ field in ‘mm_struct’ EXERCISE • Compare our demo to ‘/proc/self/maps’ In-class exercise #2 • Try running our ‘domalloc.cpp’ demo • It lets you see how a call to the ‘malloc()’ function would affect an application list of ‘vm_area_struct’ objects • NOTE: You have to install our ‘vma.ko’ kernel-object before running ‘domalloc’