Notes
main
main
  • Introduction
  • linuxKernel
    • tips
    • make_help
    • old linux
      • compile_linux0.11
      • TestEnvironment
      • load_setup
      • get_hard_data
    • list
    • plist
    • fifo
    • idr
    • xarray
    • rbtree
    • maple_tree
    • syscall
    • bitmap
    • page
    • page_flags
    • page_size
    • page mapcount
    • page refcount
    • folio
    • slub
      • proc_slabinfo
      • slub_theory
      • kmalloc_kfree
      • kmem_cache
      • slab_alloc
      • slab_free
      • proc_meminfo_SReclaimable_SReclaimable
    • vmalloc
    • brk
    • mmap
    • mremap
    • mprotect
    • madvise
    • read
    • write
    • shmem
    • huge_page
    • page_fault
    • rmap
    • lru
    • multi-gen-LRU
    • page_reclaim
    • page_cache
    • page_table
    • rcu
    • kvm
    • aarch64_boot
    • tracing_system
    • cache_coherence_and_memory_consistency
    • cpu_speculates
    • mmap_lock
    • per-vma_lock
    • cgroup
    • symbol
    • campact
    • page_ext
    • mempool
    • kernelstack
    • filesystem
    • io_stack
    • workingset
    • ioremap
    • sched_period
  • linuxDebug
    • openocd_openjtag
    • i2c_tools
    • objdump
    • addr2line
    • gdb_useage
    • debug_linux_kernel_via_gdb
    • debug_linux_module_via_gdb
    • early_boot
    • sequentially_execute
    • dynamic_debug
    • research_linuxKernel_by_patch
    • tracefs
    • ebpf
    • bpftrace
    • perf
    • flame_graph
    • crash
    • ASAN_HWASAN_MTE_check_mem_bug
    • page_owner
    • vmtouch
    • fio
    • benchmark
  • linuxSystem
    • common
      • system_version
      • procfs
      • proc_sys_vm
      • cmd_ps
      • makefile
      • file_descriptor
      • psi
      • ulimit
      • top
      • delay_accounting
    • ubuntu
      • custom_kernel
      • get_cmd_src
      • record_ssh_info
      • log
      • run_custom_script
      • repo
      • cockpit
      • nfs
      • tftp
      • misc
    • fedora
      • system_upgrade
      • custom_kernel
      • lvextend
      • yt-dlp
      • jellyfin
  • linuxDriver
    • i2c_peripherals_driver
    • spi_peripherals_driver
    • gpio_subsystem
    • IRQ_driver
    • blockIO_unblockIO_async
    • linux_own_driver
    • misc_device
    • input_device
    • timer
    • atomic_spinlock_semaphore_mutex
    • lcd
    • touch_screen
    • debugfs
    • v4l2
    • mmap
  • hardware
    • paging_mmu_pt
    • iommu
  • process_thread_scheduler
    • scheduler01
    • scheduler02
    • scheduler03
    • scheduler04
    • scheduler05
    • scheduler06
  • memory_management
    • mm1
    • mm2
    • mm3
    • mm4
    • mm5
  • input_output_filesystem
    • io_fs_01
    • io_fs_02
    • io_fs_03
    • io_fs_04
  • lock_and_lockup_detector
    • general_lock
    • hung_task
    • softLockup_hardLockup
    • crash_experiment
  • MIT_6.S081
    • 6.S081_Operating_System_Engineering
    • Schedule.md
    • Class
      • Overview
      • Administrivia
    • Labs
      • Tools
      • Guidance
      • startup
      • syscall
      • page_table
      • Calling_Convention
      • traps
    • xv6
      • xv6
    • References.md
  • qemu
    • qemu_buildroot
    • qemu_busybox.md
    • Serial.md
    • demo_mini2440
      • 0_compilation_error_summary
      • 1_compilation_steps
      • 2_operation_mode
      • 3_transplant_tools_libraries
      • 4_tools_use
      • reference_website
  • tools
    • getKernelSourceCodeList
    • nat
    • shell
    • translating
    • YouCompleteMe
    • cscope
    • global
    • vscode
    • vim
    • binary
    • markdown
    • draw
    • git
    • tig
    • tmux
    • mail_client
    • download_patchset_from_LKML
    • minicom
    • clash
  • other
    • interview
    • interview_c_base
    • know_dontknow
    • Stop-Ask-Questions-The-Stupid-Ways
    • How-To-Ask-Questions-The-Smart-Way
    • docker
    • buildroot
    • rv32_to_rv64
Powered by GitBook
On this page
  • DMA与Cache一致性
  • cgroup
  • 文件dirty数据的写回
  • 水位设置
  • getdelays
  • vmstat
  • 查找某关键字的命令或函数

Was this helpful?

  1. memory_management

mm5

DMA与Cache一致性

  • DMA传输外设数据到内存,内存是新数据,cache中可能是老数据

  • CPU写数据到内存,cache中是新数据,内存是老数据

CPU <---------> cache <---------> memory <---------> DMA <---------> 外设
                                    ^
                                    +---------------> DMA <---------> 外设

如上,此时存在DMA与Cache一致性问题,所以需要调用相关DMA API进行内存分配,才不会出现此问题。

Coherent DMA buffers,此API会关闭cache,memory与CPU直接互通,此API将从CMA获得内存

dma_alloc_coherent()
dma_free_coherent()

DMA streaming mapping,此API会开启cache,每一次读写操作都会自动同步cache与memory的内容

dma_map_sg()
dma_unmap_sg()
dma_map_single()
dma_unmap_single()

还有一些DMA自带iommu/smmu,可以分配不连续的内存,以及限制DMA的访问范围(protection)

因些dma_alloc_coheren()具体实现,有如下三种选择:

              dma_alloc_coheren()
          /           |             \
alloc_pages          cma            iommu(smmu)

cgroup

在linux读写文件时,它用于缓存文件的逻辑内容,从而加快对磁盘上映像和数据的访问

$ cat oom.c
#include <stdlib.h>
#include <stdio.h>
#include <string.h>

int main(int argc, char **argv)
{
    int max = -1;
    int mb = 0;
    char *buffer;
    int i;
#define SIZE 2000
    unsigned int *p = malloc(1024 * 1024 * SIZE);

    printf("malloc buffer: %p\n", p);

    for (i = 0; i < 1024 * 1024 * (SIZE/sizeof(int)); i++) {
        p[i] = 123;
        if ((i & 0xFFFFF) == 0) {
            printf("%dMB written\n", i >> 18);
            usleep(100000);
        }
    }
    pause();
    return 0;
}
$ gcc oom.c

$ swapoff -a                              ## 关闭所有swap
$ echo 1 > /proc/sys/vm/overcommit_memory ## 应用层可以随便申请内存

$ cd /sys/fs/cgroup/memory/
$ mkdir A
$ cd A
$ echo $((200*1024*1024)) > memory.limit_in_bytes ## 设置cgroup A的最大内存为200MB

$ cgexec -g memory:A ./a.out                      ## 把进程放入cgroup A执行
malloc buffer: 0x7f4d95cb6010
0MB written
4MB written
...
192MB written
196MB written
Killed

文件dirty数据的写回

  • dirty_expire_centisecs

    $ ls /proc/sys/vm/dirty_expire_centisecs

    当超过dirty_expire_centisecs时间时,内核自动将dirty page从memory写回disk

  • dirty_writeback_centisecs

    $ ls /proc/sys/vm/dirty_writeback_centisecs
  • dirty_backgroup_ratio

    $ ls /proc/sys/vm/dirty_writeback_centisecs

    当dirty page达到dirty_backgroup_ratio时,内核自动将dirty page从memory写回disk

  • dirty_ratio

    $ ls /proc/sys/vm/dirty_ratio

    当dirty page达到dirty_ratio时,内核自动将dirty page从memory写回disk,同时将应用程序暂停

    NOTE: dirty_backgroup_ratio < dirty_ratio

水位设置

  1. linux memory有三个水位设置,如下:

high:空闲内存到此点,停止回收
 |
 |
low:空闲内存到此点,开始回收内存,file-backed pages and swap
 |
 |
min:空闲内存到此点,继续回收内存并且阻塞应用程序

PE_MEMALLOC:紧急内存,可以忽略内存管理的min_free_kbytes水位进行分配

$ ls /proc/sys/vm/min_free_kbytes
$ ls /proc/sys/vm/lowmem_reserve_ratio ## 修改min:low:high比例
$ ls /proc/zoneinfo
  1. swappiness反映是否积极地使用swap空间

$ ls /proc/sys/vm/swappiness
  • swappiness = 0

    仅在内存不足的情况下(free and file-backed pages < high water mark in a zone),使用swap空间

  • swappiness = 60

    默认值

  • swappiness = 100

    内存将积极的使用swap空间

  • vfs_cache_pressure表示内核回收用于directory和inode_cache内存的倾向

getdelays

测量调度、I/O、swap、reclaim的延时

$ cd Documentation/accounting/getdelays.c
$ gcc getdelays.c -o getdelays
$ ./getdelays -d -c ./swap

vmstat

vmstat可以展现给定时间间隔的服务器的状态值,包括linux的CPU使用率,内存使用,虚拟内存交换情况,IO读写情况

$ vmstat 5  ##时间间隔5s

查找某关键字的命令或函数

$ apropos timer  ## 查找timer关键字
Previousmm4Nextinput_output_filesystem

Last updated 4 years ago

Was this helpful?