6. Virtualization


This section contains information about updates made to Red Hat Enterprise Linux suite of Virtualization tools.

6.1. Feature Updates

  • 已​​​​​​​更​​​​​​​新​​​​​​​ blktap(blocktap)用​​​​​​​户​​​​​​​空​​​​​​​间​​​​​​​工​​​​​​​具​​​​​​​套​​​​​​​件​​​​​​​,提​​​​​​​供​​​​​​​监​​​​​​​控​​​​​​​ blktap 备​​​​​​​份​​​​​​​的​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​的​​​​​​​传​​​​​​​输​​​​​​​统​​​​​​​计​​​​​​​功​​​​​​​能​​​​​​​。​​​​​​​
  • 添​​​​​​​加​​​​​​​了​​​​​​​对​​​​​​​ Intel 扩​​​​​​​展​​​​​​​页​​​​​​​面​​​​​​​表​​​​​​​(EPT)性​​​​​​​能​​​​​​​的​​​​​​​支​​​​​​​持​​​​​​​,改​​​​​​​进​​​​​​​了​​​​​​​在​​​​​​​支​​​​​​​持​​​​​​​ EPT 的​​​​​​​硬​​​​​​​件​​​​​​​中​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​的​​​​​​​性​​​​​​​能​​​​​​​。​​​​​​​
  • 已​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​添​​​​​​​加​​​​​​​了​​​​​​​用​​​​​​​户​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​的​​​​​​​ e1000 网​​​​​​​络​​​​​​​设​​​​​​​备​​​​​​​模​​​​​​​拟​​​​​​​,只​​​​​​​支​​​​​​​持​​​​​​​ ia64 构​​​​​​​架​​​​​​​中​​​​​​​的​​​​​​​ Windows 2003 客​​​​​​​户​​​​​​​端​​​​​​​。​​​​​​​要​​​​​​​使​​​​​​​用​​​​​​​ e1000 模​​​​​​​拟​​​​​​​,则​​​​​​​必​​​​​​​须​​​​​​​使​​​​​​​用​​​​​​​ xm 命​​​​​​​令​​​​​​​。​​​​​​​
  • 已​​​​​​​将​​​​​​​ virtio 的​​​​​​​驱​​​​​​​动​​​​​​​程​​​​​​​序​​​​​​​、​​​​​​​KVM 中​​​​​​​的​​​​​​​ I/O 虚​​​​​​​拟​​​​​​​化​​​​​​​平​​​​​​​台​​​​​​​从​​​​​​​内​​​​​​​核​​​​​​​ 2.6.27 回​​​​​​​迁​​​​​​​到​​​​​​​红​​​​​​​帽​​​​​​​企​​​​​​​业​​​​​​​版​​​​​​​ Linux 5.3。​​​​​​​这​​​​​​​些​​​​​​​驱​​​​​​​动​​​​​​​程​​​​​​​序​​​​​​​将​​​​​​​使​​​​​​​ KVM 客​​​​​​​户​​​​​​​端​​​​​​​达​​​​​​​到​​​​​​​更​​​​​​​高​​​​​​​水​​​​​​​平​​​​​​​的​​​​​​​ I/O 性​​​​​​​能​​​​​​​。​​​​​​​各​​​​​​​种​​​​​​​用​​​​​​​户​​​​​​​空​​​​​​​间​​​​​​​组​​​​​​​件​​​​​​​,比​​​​​​​如​​​​​​​:anaconda、​​​​​​​kudzu、​​​​​​​lvm、​​​​​​​selinux 和​​​​​​​ mkinitrd 已​​​​​​​被​​​​​​​更​​​​​​​新​​​​​​​至​​​​​​​可​​​​​​​支​​​​​​​持​​​​​​​ virtio 设​​​​​​​备​​​​​​​。​​​​​​​
  • 本​​​​​​​地​​​​​​​ Linux 内​​​​​​​核​​​​​​​自​​​​​​​动​​​​​​​支​​​​​​​持​​​​​​​ vmcoreinfo,但​​​​​​​要​​​​​​​在​​​​​​​ dom0 中​​​​​​​设​​​​​​​置​​​​​​​ dkump,则​​​​​​​需​​​​​​​要​​​​​​​ kernel-xen-debuginfo 软​​​​​​​件​​​​​​​包​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​发​​​​​​​行​​​​​​​本​​​​​​​中​​​​​​​,已​​​​​​​修​​​​​​​改​​​​​​​了​​​​​​​内​​​​​​​核​​​​​​​和​​​​​​​监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​,且​​​​​​​现​​​​​​​在​​​​​​​支​​​​​​​持​​​​​​​ vmcoreinfo 本​​​​​​​地​​​​​​​读​​​​​​​取​​​​​​​和​​​​​​​写​​​​​​​入​​​​​​​ kdump。​​​​​​​需​​​​​​​要​​​​​​​使​​​​​​​用​​​​​​​ kdump 进​​​​​​​行​​​​​​​ debug 或​​​​​​​者​​​​​​​其​​​​​​​它​​​​​​​在​​​​​​​ dom0 域​​​​​​​中​​​​​​​进​​​​​​​行​​​​​​​调​​​​​​​查​​​​​​​的​​​​​​​用​​​​​​​户​​​​​​​现​​​​​​​在​​​​​​​可​​​​​​​以​​​​​​​不​​​​​​​安​​​​​​​装​​​​​​​ debuginfo 或​​​​​​​者​​​​​​​ debuginfo-common 软​​​​​​​件​​​​​​​包​​​​​​​。​​​​​​​
  • 全​​​​​​​虚​​​​​​​拟​​​​​​​红​​​​​​​帽​​​​​​​企​​​​​​​业​​​​​​​版​​​​​​​ Linux 5 客​​​​​​​户​​​​​​​端​​​​​​​在​​​​​​​使​​​​​​​用​​​​​​​模​​​​​​​拟​​​​​​​磁​​​​​​​盘​​​​​​​和​​​​​​​网​​​​​​​络​​​​​​​设​​​​​​​备​​​​​​​时​​​​​​​会​​​​​​​有​​​​​​​亚​​​​​​​优​​​​​​​化​​​​​​​性​​​​​​​能​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,已​​​​​​​包​​​​​​​含​​​​​​​了​​​​​​​ kmod-xenpv 软​​​​​​​件​​​​​​​包​​​​​​​以​​​​​​​简​​​​​​​化​​​​​​​在​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​磁​​​​​​​盘​​​​​​​和​​​​​​​网​​​​​​​络​​​​​​​的​​​​​​​使​​​​​​​用​​​​​​​。​​​​​​​
    在​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​使​​​​​​​用​​​​​​​这​​​​​​​些​​​​​​​驱​​​​​​​动​​​​​​​程​​​​​​​序​​​​​​​会​​​​​​​显​​​​​​​著​​​​​​​提​​​​​​​高​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​性​​​​​​​能​​​​​​​和​​​​​​​功​​​​​​​能​​​​​​​。​​​​​​​对​​​​​​​ Netfront 和​​​​​​​ blockfront 驱​​​​​​​动​​​​​​​程​​​​​​​序​​​​​​​的​​​​​​​ bug 修​​​​​​​复​​​​​​​可​​​​​​​立​​​​​​​刻​​​​​​​被​​​​​​​内​​​​​​​核​​​​​​​软​​​​​​​件​​​​​​​包​​​​​​​识​​​​​​​别​​​​​​​并​​​​​​​同​​​​​​​步​​​​​​​了​​​​​​​。​​​​​​​
  • 客​​​​​​​户​​​​​​​端​​​​​​​现​​​​​​​在​​​​​​​可​​​​​​​以​​​​​​​使​​​​​​​用​​​​​​​ 2MB 返​​​​​​​回​​​​​​​页​​​​​​​面​​​​​​​内​​​​​​​存​​​​​​​表​​​​​​​,它​​​​​​​可​​​​​​​提​​​​​​​高​​​​​​​系​​​​​​​统​​​​​​​性​​​​​​​能​​​​​​​。​​​​​​​

6.2. Resolved Issues

6.2.1. All Architectures

  • 关​​​​​​​闭​​​​​​​一​​​​​​​个​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​可​​​​​​​能​​​​​​​导​​​​​​​致​​​​​​​ dom0 在​​​​​​​一​​​​​​​段​​​​​​​时​​​​​​​间​​​​​​​内​​​​​​​停​​​​​​​止​​​​​​​响​​​​​​​应​​​​​​​。​​​​​​​对​​​​​​​有​​​​​​​大​​​​​​​内​​​​​​​存​​​​​​​(比​​​​​​​如​​​​​​​ 12GB 以​​​​​​​上​​​​​​​)的​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​感​​​​​​​到​​​​​​​延​​​​​​​迟​​​​​​​了​​​​​​​几​​​​​​​秒​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,虚​​​​​​​拟​​​​​​​内​​​​​​​核​​​​​​​允​​​​​​​许​​​​​​​使​​​​​​​用​​​​​​​ pre-emptible 关​​​​​​​闭​​​​​​​大​​​​​​​的​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​,这​​​​​​​样​​​​​​​就​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • crash was unable to read the relocation address of the hypervisor from a vmcore file. Consequently, opening a Virtualized kernel vmcore file with crash would fail, resulting in the error:
    crash: cannot resolve "idle_pg_table_4"
    
    In this update, the hypervisor now saves the address correctly, which resolves this issue.
  • 之​​​​​​​前​​​​​​​,半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​最​​​​​​​多​​​​​​​只​​​​​​​能​​​​​​​有​​​​​​​ 16 个​​​​​​​磁​​​​​​​盘​​​​​​​设​​​​​​​备​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,已​​​​​​​将​​​​​​​这​​​​​​​个​​​​​​​限​​​​​​​制​​​​​​​提​​​​​​​高​​​​​​​到​​​​​​​ 256 个​​​​​​​磁​​​​​​​盘​​​​​​​设​​​​​​​备​​​​​​​。​​​​​​​
  • 为​​​​​​​ kdump 内​​​​​​​核​​​​​​​保​​​​​​​留​​​​​​​的​​​​​​​内​​​​​​​存​​​​​​​不​​​​​​​正​​​​​​​确​​​​​​​,导​​​​​​​致​​​​​​​崩​​​​​​​溃​​​​​​​转​​​​​​​储​​​​​​​不​​​​​​​可​​​​​​​用​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,内​​​​​​​存​​​​​​​保​​​​​​​留​​​​​​​是​​​​​​​正​​​​​​​确​​​​​​​,这​​​​​​​样​​​​​​​就​​​​​​​可​​​​​​​生​​​​​​​成​​​​​​​正​​​​​​​确​​​​​​​的​​​​​​​崩​​​​​​​溃​​​​​​​转​​​​​​​储​​​​​​​。​​​​​​​
  • 在​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​附​​​​​​​加​​​​​​​一​​​​​​​个​​​​​​​有​​​​​​​特​​​​​​​定​​​​​​​名​​​​​​​称​​​​​​​的​​​​​​​磁​​​​​​​盘​​​​​​​(例​​​​​​​如​​​​​​​ /dev/xvdaa、​​​​​​​/dev/xvdab、​​​​​​​/dev/xvdbc 等​​​​​​​等​​​​​​​)会​​​​​​​导​​​​​​​致​​​​​​​在​​​​​​​该​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​出​​​​​​​现​​​​​​​一​​​​​​​个​​​​​​​崩​​​​​​​溃​​​​​​​的​​​​​​​ /dev 设​​​​​​​备​​​​​​​。​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,以​​​​​​​便​​​​​​​在​​​​​​​向​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​添​​​​​​​加​​​​​​​有​​​​​​​这​​​​​​​些​​​​​​​名​​​​​​​称​​​​​​​的​​​​​​​磁​​​​​​​盘​​​​​​​时​​​​​​​可​​​​​​​在​​​​​​​该​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​中​​​​​​​生​​​​​​​成​​​​​​​一​​​​​​​个​​​​​​​正​​​​​​​确​​​​​​​的​​​​​​​ /dev 设​​​​​​​备​​​​​​​。​​​​​​​
  • 之​​​​​​​前​​​​​​​回​​​​​​​路​​​​​​​设​​​​​​​备​​​​​​​的​​​​​​​数​​​​​​​目​​​​​​​是​​​​​​​限​​​​​​​制​​​​​​​在​​​​​​​ 4。​​​​​​​而​​​​​​​这​​​​​​​限​​​​​​​制​​​​​​​了​​​​​​​在​​​​​​​有​​​​​​​超​​​​​​​过​​​​​​​ 4 个​​​​​​​网​​​​​​​络​​​​​​​接​​​​​​​口​​​​​​​的​​​​​​​系​​​​​​​统​​​​​​​中​​​​​​​创​​​​​​​建​​​​​​​桥​​​​​​​接​​​​​​​的​​​​​​​能​​​​​​​力​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,netloop 驱​​​​​​​动​​​​​​​程​​​​​​​序​​​​​​​现​​​​​​​在​​​​​​​可​​​​​​​根​​​​​​​据​​​​​​​需​​​​​​​要​​​​​​​创​​​​​​​建​​​​​​​额​​​​​​​外​​​​​​​的​​​​​​​回​​​​​​​路​​​​​​​设​​​​​​​备​​​​​​​。​​​​​​​
  • 当​​​​​​​创​​​​​​​建​​​​​​​并​​​​​​​销​​​​​​​毁​​​​​​​虚​​​​​​​拟​​​​​​​网​​​​​​​络​​​​​​​设​​​​​​​备​​​​​​​时​​​​​​​会​​​​​​​出​​​​​​​现​​​​​​​竞​​​​​​​态​​​​​​​环​​​​​​​境​​​​​​​。​​​​​​​在​​​​​​​有​​​​​​​些​​​​​​​情​​​​​​​况​​​​​​​下​​​​​​​,特​​​​​​​别​​​​​​​是​​​​​​​处​​​​​​​于​​​​​​​高​​​​​​​负​​​​​​​载​​​​​​​状​​​​​​​态​​​​​​​时​​​​​​​,这​​​​​​​会​​​​​​​导​​​​​​​致​​​​​​​虚​​​​​​​拟​​​​​​​设​​​​​​​备​​​​​​​不​​​​​​​响​​​​​​​应​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​会​​​​​​​检​​​​​​​查​​​​​​​虚​​​​​​​拟​​​​​​​设​​​​​​​备​​​​​​​状​​​​​​​态​​​​​​​以​​​​​​​防​​​​​​​止​​​​​​​出​​​​​​​现​​​​​​​竞​​​​​​​态​​​​​​​环​​​​​​​境​​​​​​​。​​​​​​​
  • 如​​​​​​​果​​​​​​​让​​​​​​​应​​​​​​​用​​​​​​​程​​​​​​​序​​​​​​​一​​​​​​​直​​​​​​​运​​​​​​​行​​​​​​​则​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​遇​​​​​​​到​​​​​​​ virt-manager 中​​​​​​​的​​​​​​​内​​​​​​​存​​​​​​​泄​​​​​​​露​​​​​​​。​​​​​​​因​​​​​​​此​​​​​​​,应​​​​​​​用​​​​​​​程​​​​​​​序​​​​​​​将​​​​​​​不​​​​​​​断​​​​​​​消​​​​​​​耗​​​​​​​更​​​​​​​多​​​​​​​资​​​​​​​源​​​​​​​,这​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​导​​​​​​​致​​​​​​​内​​​​​​​存​​​​​​​不​​​​​​​足​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​修​​​​​​​复​​​​​​​了​​​​​​​泄​​​​​​​露​​​​​​​问​​​​​​​题​​​​​​​,也​​​​​​​就​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • the crash utility could not analyze x86_64 vmcores from systems running kernel-xen because the Red Hat Enterprise Linux hypervisor was relocatable and the relocated physical base address is not passed in the vmcore file's ELF header. The new --xen_phys_start command line option for the crash utility allows the user to pass crash the relocated base physical address.
  • Paravirtual Frame Buffer (PVFB) 不​​​​​​​能​​​​​​​捕​​​​​​​获​​​​​​​并​​​​​​​处​​​​​​​理​​​​​​​所​​​​​​​有​​​​​​​鼠​​​​​​​标​​​​​​​设​​​​​​​备​​​​​​​。​​​​​​​因​​​​​​​此​​​​​​​当​​​​​​​与​​​​​​​有​​​​​​​ Virtual Machine Console 的​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​互​​​​​​​动​​​​​​​时​​​​​​​滚​​​​​​​轮​​​​​​​不​​​​​​​会​​​​​​​工​​​​​​​作​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​,现​​​​​​​在​​​​​​​可​​​​​​​正​​​​​​​确​​​​​​​处​​​​​​​理​​​​​​​滚​​​​​​​轮​​​​​​​鼠​​​​​​​标​​​​​​​事​​​​​​​件​​​​​​​,这​​​​​​​样​​​​​​​就​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • 在​​​​​​​含​​​​​​​有​​​​​​​大​​​​​​​量​​​​​​​ CPU 的​​​​​​​机​​​​​​​器​​​​​​​中​​​​​​​使​​​​​​​用​​​​​​​虚​​​​​​​拟​​​​​​​化​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​导​​​​​​​致​​​​​​​监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​在​​​​​​​安​​​​​​​装​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​时​​​​​​​崩​​​​​​​溃​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • 在​​​​​​​返​​​​​​​回​​​​​​​ CPUID 产​​​​​​​品​​​​​​​线​​​​​​​数​​​​​​​值​​​​​​​ 6 的​​​​​​​ Intel 处​​​​​​​理​​​​​​​器​​​​​​​中​​​​​​​,在​​​​​​​ kernel-xen 中​​​​​​​只​​​​​​​启​​​​​​​用​​​​​​​了​​​​​​​一​​​​​​​个​​​​​​​性​​​​​​​能​​​​​​​计​​​​​​​数​​​​​​​器​​​​​​​注​​​​​​​册​​​​​​​。​​​​​​​因​​​​​​​此​​​​​​​,只​​​​​​​有​​​​​​​计​​​​​​​数​​​​​​​器​​​​​​​ 0 提​​​​​​​供​​​​​​​示​​​​​​​例​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​

6.2.2. x86 Architectures

  • On systems with newer CPU's, the CPU APIC ID differs from the CPU ID. Consequently, the virtualized kernel was unable to initialize CPU frequency scaling. In this update, the virtualized kernel now retrieves CPU APIC ID from the hypervisor, allowing CPU frequency scaling to be initialized properly.
  • 当​​​​​​​运​​​​​​​行​​​​​​​ x86 半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​时​​​​​​​,如​​​​​​​果​​​​​​​某​​​​​​​个​​​​​​​进​​​​​​​程​​​​​​​访​​​​​​​问​​​​​​​无​​​​​​​效​​​​​​​内​​​​​​​存​​​​​​​,它​​​​​​​就​​​​​​​会​​​​​​​在​​​​​​​一​​​​​​​个​​​​​​​回​​​​​​​路​​​​​​​中​​​​​​​运​​​​​​​行​​​​​​​而​​​​​​​不​​​​​​​是​​​​​​​获​​​​​​​得​​​​​​​ SEGV 信​​​​​​​号​​​​​​​。​​​​​​​这​​​​​​​是​​​​​​​由​​​​​​​于​​​​​​​监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​中​​​​​​​采​​​​​​​用​​​​​​​的​​​​​​​ execshield 检​​​​​​​查​​​​​​​方​​​​​​​法​​​​​​​中​​​​​​​的​​​​​​​一​​​​​​​个​​​​​​​漏​​​​​​​洞​​​​​​​所​​​​​​​致​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​

6.2.3. ia64 Architecture

  • A xend bug that previously caused guest installation failures is now fixed.
  • evtchn 事​​​​​​​件​​​​​​​频​​​​​​​道​​​​​​​设​​​​​​​备​​​​​​​缺​​​​​​​少​​​​​​​锁​​​​​​​定​​​​​​​和​​​​​​​内​​​​​​​存​​​​​​​屏​​​​​​​障​​​​​​​。​​​​​​​这​​​​​​​导​​​​​​​致​​​​​​​ xenstore 无​​​​​​​法​​​​​​​响​​​​​​​应​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • 非​​​​​​​统​​​​​​​一​​​​​​​内​​​​​​​存​​​​​​​访​​​​​​​问​​​​​​​(NUMA)信​​​​​​​息​​​​​​​不​​​​​​​是​​​​​​​由​​​​​​​ xm info 命​​​​​​​令​​​​​​​显​​​​​​​示​​​​​​​。​​​​​​​因​​​​​​​此​​​​​​​,会​​​​​​​返​​​​​​​回​​​​​​​每​​​​​​​个​​​​​​​节​​​​​​​点​​​​​​​的​​​​​​​不​​​​​​​正​​​​​​​确​​​​​​​ node_to_cpu 值​​​​​​​ no cpus。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​
  • 之​​​​​​​前​​​​​​​,在​​​​​​​使​​​​​​​用​​​​​​​ VT-i2 技​​​​​​​术​​​​​​​的​​​​​​​处​​​​​​​理​​​​​​​器​​​​​​​的​​​​​​​硬​​​​​​​件​​​​​​​虚​​​​​​​拟​​​​​​​机​​​​​​​(HVM)中​​​​​​​创​​​​​​​建​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​将​​​​​​​失​​​​​​​败​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​个​​​​​​​更​​​​​​​新​​​​​​​中​​​​​​​解​​​​​​​决​​​​​​​了​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​。​​​​​​​

6.2.4. x86_64 Architectures

  • On systems with newer CPU's, the CPU APIC ID differs from the CPU ID. Consequently, the virtualized kernel was unable to initialize CPU frequency scaling. In this update, the virtualized kernel now retrieves CPU APIC ID from the hypervisor, allowing CPU frequency scaling to be initialized properly.

6.3. Known Issues

6.3.1. All Architectures

  • 使​​​​​​​用​​​​​​​虚​​​​​​​拟​​​​​​​内​​​​​​​核​​​​​​​时​​​​​​​将​​​​​​​无​​​​​​​法​​​​​​​访​​​​​​​问​​​​​​​ Diskette 驱​​​​​​​动​​​​​​​器​​​​​​​介​​​​​​​质​​​​​​​。​​​​​​​要​​​​​​​解​​​​​​​决​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,请​​​​​​​使​​​​​​​用​​​​​​​附​​​​​​​带​​​​​​​ USB 的​​​​​​​ diskette 驱​​​​​​​动​​​​​​​器​​​​​​​。​​​​​​​
    请​​​​​​​注​​​​​​​意​​​​​​​:diskette 驱​​​​​​​动​​​​​​​器​​​​​​​介​​​​​​​质​​​​​​​在​​​​​​​其​​​​​​​它​​​​​​​非​​​​​​​虚​​​​​​​拟​​​​​​​内​​​​​​​核​​​​​​​中​​​​​​​工​​​​​​​作​​​​​​​良​​​​​​​好​​​​​​​。​​​​​​​
  • In live migrations of paravirtualized guests, time-dependent guest processes may function improperly if the corresponding hosts' (dom0) times are not synchronized. Use NTP to synchronize system times for all corresponding hosts before migration.
  • 重​​​​​​​复​​​​​​​地​​​​​​​在​​​​​​​两​​​​​​​台​​​​​​​主​​​​​​​机​​​​​​​间​​​​​​​迁​​​​​​​移​​​​​​​一​​​​​​​个​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​导​​​​​​​致​​​​​​​其​​​​​​​中​​​​​​​一​​​​​​​个​​​​​​​主​​​​​​​机​​​​​​​ panic。​​​​​​​如​​​​​​​果​​​​​​​某​​​​​​​个​​​​​​​主​​​​​​​机​​​​​​​在​​​​​​​将​​​​​​​某​​​​​​​个​​​​​​​客​​​​​​​户​​​​​​​机​​​​​​​迁​​​​​​​移​​​​​​​出​​​​​​​系​​​​​​​统​​​​​​​后​​​​​​​至​​​​​​​将​​​​​​​其​​​​​​​迁​​​​​​​移​​​​​​​回​​​​​​​该​​​​​​​系​​​​​​​统​​​​​​​前​​​​​​​重​​​​​​​启​​​​​​​,就​​​​​​​不​​​​​​​会​​​​​​​出​​​​​​​现​​​​​​​ panic 的​​​​​​​情​​​​​​​况​​​​​​​。​​​​​​​
  • 在​​​​​​​使​​​​​​​用​​​​​​​多​​​​​​​虚​​​​​​​拟​​​​​​​ CPU 引​​​​​​​导​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​时​​​​​​​,运​​​​​​​行​​​​​​​ Windows 2008 或​​​​​​​者​​​​​​​ Windows Vista 作​​​​​​​为​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​时​​​​​​​格​​​​​​​式​​​​​​​化​​​​​​​磁​​​​​​​盘​​​​​​​可​​​​​​​导​​​​​​​致​​​​​​​崩​​​​​​​溃​​​​​​​。​​​​​​​要​​​​​​​解​​​​​​​决​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,请​​​​​​​在​​​​​​​格​​​​​​​式​​​​​​​化​​​​​​​时​​​​​​​使​​​​​​​用​​​​​​​单​​​​​​​一​​​​​​​虚​​​​​​​拟​​​​​​​ CPU 引​​​​​​​导​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​。​​​​​​​
  • 通​​​​​​​过​​​​​​​ virt-manager 创​​​​​​​建​​​​​​​的​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​有​​​​​​​时​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​使​​​​​​​鼠​​​​​​​标​​​​​​​无​​​​​​​法​​​​​​​在​​​​​​​整​​​​​​​个​​​​​​​屏​​​​​​​幕​​​​​​​中​​​​​​​移​​​​​​​动​​​​​​​。​​​​​​​要​​​​​​​解​​​​​​​决​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,请​​​​​​​使​​​​​​​用​​​​​​​ virt-manager 为​​​​​​​该​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​配​​​​​​​置​​​​​​​一​​​​​​​个​​​​​​​ USB 平​​​​​​​板​​​​​​​设​​​​​​​备​​​​​​​。​​​​​​​
  • 在​​​​​​​有​​​​​​​ 128 或​​​​​​​者​​​​​​​更​​​​​​​多​​​​​​​ CPU 的​​​​​​​系​​​​​​​统​​​​​​​中​​​​​​​,必​​​​​​​须​​​​​​​将​​​​​​​最​​​​​​​大​​​​​​​ CPU 数​​​​​​​目​​​​​​​限​​​​​​​制​​​​​​​在​​​​​​​ 128 以​​​​​​​下​​​​​​​。​​​​​​​请​​​​​​​使​​​​​​​用​​​​​​​ maxcpus=126 监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​参​​​​​​​数​​​​​​​将​​​​​​​监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​限​​​​​​​制​​​​​​​为​​​​​​​ 126。​​​​​​​
  • 全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​修​​​​​​​正​​​​​​​由​​​​​​​于​​​​​​​暂​​​​​​​停​​​​​​​域​​​​​​​和​​​​​​​取​​​​​​​消​​​​​​​暂​​​​​​​停​​​​​​​域​​​​​​​所​​​​​​​造​​​​​​​成​​​​​​​的​​​​​​​时​​​​​​​间​​​​​​​损​​​​​​​失​​​​​​​。​​​​​​​在​​​​​​​暂​​​​​​​停​​​​​​​和​​​​​​​取​​​​​​​消​​​​​​​暂​​​​​​​停​​​​​​​事​​​​​​​件​​​​​​​间​​​​​​​可​​​​​​​正​​​​​​​确​​​​​​​跟​​​​​​​踪​​​​​​​时​​​​​​​间​​​​​​​是​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​内​​​​​​​核​​​​​​​的​​​​​​​优​​​​​​​势​​​​​​​之​​​​​​​一​​​​​​​。​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​已​​​​​​​在​​​​​​​上​​​​​​​游​​​​​​​进​​​​​​​行​​​​​​​了​​​​​​​讨​​​​​​​论​​​​​​​,提​​​​​​​议​​​​​​​更​​​​​​​换​​​​​​​计​​​​​​​时​​​​​​​器​​​​​​​,一​​​​​​​百​​​​​​​年​​​​​​​全​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​有​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​的​​​​​​​计​​​​​​​时​​​​​​​器​​​​​​​。​​​​​​​目​​​​​​​前​​​​​​​,这​​​​​​​个​​​​​​​编​​​​​​​码​​​​​​​还​​​​​​​在​​​​​​​上​​​​​​​游​​​​​​​开​​​​​​​发​​​​​​​中​​​​​​​,并​​​​​​​将​​​​​​​会​​​​​​​在​​​​​​​红​​​​​​​帽​​​​​​​企​​​​​​​业​​​​​​​版​​​​​​​ Linux 的​​​​​​​稍​​​​​​​后​​​​​​​的​​​​​​​版​​​​​​​本​​​​​​​中​​​​​​​可​​​​​​​用​​​​​​​。​​​​​​​
  • 重​​​​​​​复​​​​​​​迁​​​​​​​移​​​​​​​半​​​​​​​虚​​​​​​​拟​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​可​​​​​​​能​​​​​​​会​​​​​​​导​​​​​​​致​​​​​​​在​​​​​​​ dom0 控​​​​​​​制​​​​​​​台​​​​​​​中​​​​​​​出​​​​​​​现​​​​​​​ bad mpa 信​​​​​​​息​​​​​​​。​​​​​​​在​​​​​​​这​​​​​​​种​​​​​​​情​​​​​​​况​​​​​​​下​​​​​​​,监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​可​​​​​​​能​​​​​​​也​​​​​​​会​​​​​​​ panic。​​​​​​​
    要​​​​​​​防​​​​​​​止​​​​​​​监​​​​​​​控​​​​​​​程​​​​​​​序​​​​​​​内​​​​​​​核​​​​​​​ panic,请​​​​​​​在​​​​​​​出​​​​​​​现​​​​​​​ bad mpa 信​​​​​​​息​​​​​​​时​​​​​​​重​​​​​​​启​​​​​​​迁​​​​​​​移​​​​​​​的​​​​​​​客​​​​​​​户​​​​​​​端​​​​​​​。​​​​​​​
  • 当​​​​​​​设​​​​​​​置​​​​​​​绑​​​​​​​定​​​​​​​到​​​​​​​ dom0 的​​​​​​​接​​​​​​​口​​​​​​​时​​​​​​​,默​​​​​​​认​​​​​​​ network-bridge 基​​​​​​​本​​​​​​​可​​​​​​​能​​​​​​​导​​​​​​​致​​​​​​​绑​​​​​​​定​​​​​​​的​​​​​​​网​​​​​​​络​​​​​​​接​​​​​​​口​​​​​​​在​​​​​​​ unavailable 和​​​​​​​ available 之​​​​​​​间​​​​​​​切​​​​​​​换​​​​​​​。​​​​​​​即​​​​​​​我​​​​​​​们​​​​​​​通​​​​​​​常​​​​​​​所​​​​​​​说​​​​​​​的​​​​​​​ flapping。​​​​​​​
    要​​​​​​​防​​​​​​​止​​​​​​​这​​​​​​​种​​​​​​​情​​​​​​​况​​​​​​​发​​​​​​​生​​​​​​​,请​​​​​​​在​​​​​​​ /etc/xen/xend-config.sxp 中​​​​​​​使​​​​​​​用​​​​​​​以​​​​​​​下​​​​​​​行​​​​​​​替​​​​​​​换​​​​​​​标​​​​​​​准​​​​​​​ network-script 行​​​​​​​:
    			
    (network-script network-bridge-bonding netdev=bond0)
    
    这​​​​​​​样​​​​​​​做​​​​​​​将​​​​​​​禁​​​​​​​用​​​​​​​ netloop 设​​​​​​​备​​​​​​​,这​​​​​​​样​​​​​​​做​​​​​​​可​​​​​​​防​​​​​​​止​​​​​​​在​​​​​​​地​​​​​​​址​​​​​​​转​​​​​​​换​​​​​​​过​​​​​​​程​​​​​​​中​​​​​​​地​​​​​​​址​​​​​​​解​​​​​​​析​​​​​​​协​​​​​​​议​​​​​​​(ARP)监​​​​​​​控​​​​​​​失​​​​​​​败​​​​​​​。​​​​​​​
  • When running multiple guest domains, guest networking may temporarily stop working, resulting in the following error being reported in the dom0 logs:
    Memory squeeze in netback driver
    
    To work around this, raise the amount of memory available to the dom0 with the dom0_mem hypervisor command line option.

6.3.2. x86 Architectures

  • Migrating paravirtualized guests through xm migrate [domain] [dom0 IP address] does not work.
  • When installing Red Hat Enterprise Linux 5 on a fully virtualized SMP guest, the installation may freeze. This can occur when the host (dom0) is running Red Hat Enterprise Linux 5.2.
    要​​​​​​​防​​​​​​​止​​​​​​​此​​​​​​​现​​​​​​​象​​​​​​​的​​​​​​​发​​​​​​​生​​​​​​​,请​​​​​​​将​​​​​​​客​​​​​​​户​​​​​​​机​​​​​​​设​​​​​​​为​​​​​​​在​​​​​​​安​​​​​​​装​​​​​​​时​​​​​​​使​​​​​​​用​​​​​​​单​​​​​​​一​​​​​​​处​​​​​​​理​​​​​​​器​​​​​​​。​​​​​​​您​​​​​​​可​​​​​​​以​​​​​​​使​​​​​​​用​​​​​​​ virt-install 的​​​​​​​ --vcpus=1 选​​​​​​​项​​​​​​​达​​​​​​​到​​​​​​​此​​​​​​​目​​​​​​​的​​​​​​​。​​​​​​​安​​​​​​​装​​​​​​​完​​​​​​​成​​​​​​​后​​​​​​​,您​​​​​​​可​​​​​​​以​​​​​​​通​​​​​​​过​​​​​​​在​​​​​​​ virt-manager 中​​​​​​​修​​​​​​​改​​​​​​​分​​​​​​​配​​​​​​​的​​​​​​​ vcpus 来​​​​​​​将​​​​​​​客​​​​​​​户​​​​​​​机​​​​​​​设​​​​​​​定​​​​​​​为​​​​​​​ SMP。​​​​​​​

6.3.3. x86_64 Architectures

  • Migrating paravirtualized guests through xm migrate [domain] [dom0 IP address] does not work.
  • Installing the Virtualization feature may cause a time went backwards warning on HP systems with model numbers xw9300 and xw9400.
    对​​​​​​​于​​​​​​​ xw9400 机​​​​​​​器​​​​​​​,如​​​​​​​果​​​​​​​要​​​​​​​解​​​​​​​决​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,可​​​​​​​以​​​​​​​在​​​​​​​ BIOS 里​​​​​​​启​​​​​​​用​​​​​​​ HPET 计​​​​​​​时​​​​​​​器​​​​​​​。​​​​​​​注​​​​​​​意​​​​​​​,这​​​​​​​个​​​​​​​选​​​​​​​项​​​​​​​对​​​​​​​于​​​​​​​ xw9300 机​​​​​​​器​​​​​​​是​​​​​​​不​​​​​​​可​​​​​​​用​​​​​​​的​​​​​​​。​​​​​​​
  • Installing Red Hat Enterprise Linux 3.9 on a fully virtualized guest may be extremely slow. In addition, booting up the guest after installation may result in hda: lost interrupt errors.
    要​​​​​​​避​​​​​​​免​​​​​​​此​​​​​​​启​​​​​​​动​​​​​​​错​​​​​​​误​​​​​​​,请​​​​​​​将​​​​​​​客​​​​​​​户​​​​​​​机​​​​​​​配​​​​​​​置​​​​​​​为​​​​​​​使​​​​​​​用​​​​​​​ SMP 内​​​​​​​核​​​​​​​。​​​​​​​
  • Upgrading a host (dom0) system to Red Hat Enterprise Linux 5.2 may render existing Red Hat Enterprise Linux 4.5 SMP paravirtualized guests unbootable. This is more likely to occur when the host system has more than 4GB of RAM.
    要​​​​​​​解​​​​​​​决​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​,请​​​​​​​在​​​​​​​单​​​​​​​一​​​​​​​ CPU 模​​​​​​​式​​​​​​​下​​​​​​​引​​​​​​​导​​​​​​​每​​​​​​​个​​​​​​​红​​​​​​​帽​​​​​​​企​​​​​​​业​​​​​​​版​​​​​​​ Linux 4.5 客​​​​​​​户​​​​​​​端​​​​​​​,并​​​​​​​将​​​​​​​其​​​​​​​内​​​​​​​核​​​​​​​升​​​​​​​级​​​​​​​到​​​​​​​最​​​​​​​新​​​​​​​的​​​​​​​版​​​​​​​本​​​​​​​(为​​​​​​​红​​​​​​​帽​​​​​​​企​​​​​​​业​​​​​​​版​​​​​​​ Linux 4.5.z)。​​​​​​​

6.3.4. ia64 Architecture

  • Migrating paravirtualized guests through xm migrate [domain] [dom0 IP address] does not work.
  • On some Itanium systems configured for console output to VGA, the dom0 virtualized kernel may fail to boot. This is because the virtualized kernel failed to properly detect the default console device from the Extensible Firmware Interface (EFI) settings.
    出​​​​​​​现​​​​​​​这​​​​​​​个​​​​​​​问​​​​​​​题​​​​​​​时​​​​​​​,在​​​​​​​ /boot/efi/elilo.conf 文​​​​​​​件​​​​​​​的​​​​​​​内​​​​​​​核​​​​​​​引​​​​​​​导​​​​​​​选​​​​​​​项​​​​​​​中​​​​​​​添​​​​​​​加​​​​​​​ console=tty 引​​​​​​​导​​​​​​​参​​​​​​​数​​​​​​​即​​​​​​​可​​​​​​​解​​​​​​​决​​​​​​​。​​​​​​​
  • On some Itanium systems (such as the Hitachi Cold Fusion 3e), the serial port cannot be detected in dom0 when VGA is enabled by the EFI Maintenance Manager. As such, you need to supply the following serial port information to the dom0 kernel:
    • 速​​​​​​​度​​​​​​​(以​​​​​​​字​​​​​​​节​​​​​​​/秒​​​​​​​为​​​​​​​单​​​​​​​位​​​​​​​)
    • 数​​​​​​​据​​​​​​​位​​​​​​​数​​​​​​​
    • 奇​​​​​​​偶​​​​​​​性​​​​​​​
    • io_base 地​​​​​​​址​​​​​​​
    这​​​​​​​些​​​​​​​情​​​​​​​况​​​​​​​必​​​​​​​须​​​​​​​在​​​​​​​ /boot/efi/elilo.conf 内​​​​​​​核​​​​​​​的​​​​​​​ append= 行​​​​​​​中​​​​​​​指​​​​​​​定​​​​​​​。​​​​​​​例​​​​​​​如​​​​​​​:
    append="com1=19200,8n1,0x3f8 -- quiet rhgb console=tty0 console=ttyS0,19200n8"
    在​​​​​​​这​​​​​​​个​​​​​​​示​​​​​​​例​​​​​​​中​​​​​​​,com1 为​​​​​​​串​​​​​​​行​​​​​​​端​​​​​​​口​​​​​​​,19200 为​​​​​​​速​​​​​​​度​​​​​​​(字​​​​​​​节​​​​​​​/秒​​​​​​​),8n1 指​​​​​​​定​​​​​​​数​​​​​​​据​​​​​​​位​​​​​​​数​​​​​​​的​​​​​​​奇​​​​​​​偶​​​​​​​性​​​​​​​设​​​​​​​置​​​​​​​,0x3f8 是​​​​​​​ io_base 的​​​​​​​地​​​​​​​址​​​​​​​。​​​​​​​
  • Virtualization does not work on some architectures that use Non-Uniform Memory Access (NUMA). As such, installing the virtualized kernel on systems that use NUMA will result in a boot failure.
    有​​​​​​​些​​​​​​​安​​​​​​​装​​​​​​​号​​​​​​​可​​​​​​​默​​​​​​​认​​​​​​​安​​​​​​​装​​​​​​​虚​​​​​​​拟​​​​​​​的​​​​​​​内​​​​​​​核​​​​​​​。​​​​​​​如​​​​​​​果​​​​​​​您​​​​​​​有​​​​​​​这​​​​​​​样​​​​​​​一​​​​​​​个​​​​​​​安​​​​​​​装​​​​​​​号​​​​​​​,且​​​​​​​您​​​​​​​的​​​​​​​系​​​​​​​统​​​​​​​使​​​​​​​用​​​​​​​ NUMA,并​​​​​​​不​​​​​​​与​​​​​​​ knernel-xen 合​​​​​​​作​​​​​​​,请​​​​​​​在​​​​​​​安​​​​​​​装​​​​​​​过​​​​​​​程​​​​​​​中​​​​​​​不​​​​​​​要​​​​​​​选​​​​​​​择​​​​​​​虚​​​​​​​拟​​​​​​​化​​​​​​​选​​​​​​​项​​​​​​​。​​​​​​​
  • Currently, live migration of fully virtualized guests is not supported on this architecture. In addition, kexec and kdump are also not supported for virtualization on this architecture.
Red Hat logoGithubRedditYoutubeTwitter

学习

尝试、购买和销售

社区

关于红帽文档

通过我们的产品和服务,以及可以信赖的内容,帮助红帽用户创新并实现他们的目标。

让开源更具包容性

红帽致力于替换我们的代码、文档和 Web 属性中存在问题的语言。欲了解更多详情,请参阅红帽博客.

關於紅帽

我们提供强化的解决方案,使企业能够更轻松地跨平台和环境(从核心数据中心到网络边缘)工作。

© 2024 Red Hat, Inc.