Bayan watanni takwas na ci gaba, an saki hypervisor Xen 4.16 kyauta. Kamfanoni irin su Amazon, Arm, Bitdefender, Citrix da EPAM Systems sun shiga cikin haɓaka sabon sakin. Sakin sabuntawa na reshen Xen 4.16 zai šauki har zuwa Yuni 2, 2023, da kuma buga gyare-gyaren raunin har zuwa Disamba 2, 2024.
Canje-canje masu mahimmanci a cikin Xen 4.16:
- Manajan TPM, wanda ke tabbatar da aikin kwakwalwan kwamfuta masu kama-da-wane don adana maɓallan cryptographic (vTPM), wanda aka aiwatar bisa tushen TPM na zahiri (Trusted Platform Module), an gyara don aiwatar da goyan bayan ƙayyadaddun TPM 2.0.
- Ingantacciyar dogaro akan layin PV Shim da aka yi amfani da shi don gudanar da baƙi mara kyau (PV) a cikin mahallin PVH da HVM. Ci gaba, yin amfani da 32-bit mara kyau baƙi zai yiwu ne kawai a cikin yanayin PV Shim, wanda zai rage adadin wurare a cikin hypervisor wanda zai iya haifar da lahani.
- Ƙara ikon yin taya akan na'urorin Intel ba tare da mai ƙidayar lokaci ba (PIT, Timer Interval Timer).
- An tsabtace abubuwan da ba a taɓa amfani da su ba, dakatar da gina tsohuwar lambar "qemu-xen-gargajiya" da PV-Grub (buƙatar waɗannan ƙayyadaddun cokali na Xen sun ɓace bayan an canza canje-canje tare da tallafin Xen zuwa babban tsarin QEMU da Grub).
- Ga baƙi tare da gine-ginen ARM, an aiwatar da tallafin farko don ƙididdiga masu lura da ayyuka.
- Ingantattun tallafi don yanayin dom0less, wanda ke ba ku damar guje wa tura yanayin dom0 lokacin fara injunan kama-da-wane a farkon matakin boot ɗin uwar garken. Canje-canjen sun ba da damar aiwatar da tallafi don tsarin 64-bit ARM tare da firmware EFI.
- Ingantaccen tallafi don tsarin 64-bit ARM daban-daban dangane da babban gine-gine.LITTLE, wanda ya haɗu da maɗaukaki masu ƙarfi amma masu fama da yunwa da ƙananan aiki amma mafi yawan ƙarfin wutar lantarki a cikin guntu guda ɗaya.
A lokaci guda, Intel ya buga sakin Cloud Hypervisor 20.0 hypervisor, wanda aka gina bisa tushen tsarin haɗin gwiwa na Rust-VMM, wanda, ban da Intel, Alibaba, Amazon, Google da Red Hat suma suna shiga. Rust-VMM an rubuta shi a cikin yaren Rust kuma yana ba ku damar ƙirƙira takamaiman hypervisors na ɗawainiya. Cloud Hypervisor shine irin wannan hypervisor wanda ke ba da babban matakin injin saka idanu (VMM) yana gudana akan KVM kuma an inganta shi don ayyuka na asali na girgije. Ana samun lambar aikin a ƙarƙashin lasisin Apache 2.0.
Cloud Hypervisor yana mai da hankali kan gudanar da rarrabawar Linux ta zamani ta amfani da na'urorin da ba su dace ba. Daga cikin mahimman manufofin da aka ambata sune: babban amsawa, ƙarancin amfani da ƙwaƙwalwar ajiya, babban aiki, sauƙaƙan daidaitawa da rage yiwuwar kai hari. Taimakon kwaikwayi ana kiyaye shi zuwa mafi ƙanƙanta kuma an mai da hankali kan ƙayyadaddun abubuwa. A halin yanzu tsarin x86_64 kawai ake tallafawa, amma ana shirin tallafin AArch64. Don tsarin baƙo, ginanniyar 64-bit na Linux a halin yanzu ana tallafawa. An saita CPU, ƙwaƙwalwar ajiya, PCI da NVDIMM a matakin taro. Yana yiwuwa a yi ƙaura na'urori masu kama da juna tsakanin sabobin.
A cikin sabon sigar:
- Don gine-ginen x86_64 da aarch64, har zuwa sassan PCI 16 yanzu an yarda, wanda ke ƙara adadin adadin na'urorin PCI da aka yarda daga 31 zuwa 496.
- An aiwatar da goyan baya don ɗaure CPUs na kama-da-wane zuwa ainihin abubuwan CPU na zahiri (pinning CPU). Ga kowane vCPU, yanzu yana yiwuwa a ayyana ƙayyadaddun ƙayyadaddun ƙayyadaddun tsarin CPUs wanda aka ba da izinin aiwatarwa, wanda zai iya zama da amfani lokacin yin taswira kai tsaye (1: 1) mai masaukin baki da albarkatun baƙi ko lokacin gudanar da injin kama-da-wane akan takamaiman kumburin NUMA.
- Ingantattun goyan bayan I/O na gani. Kowane yanki na VFIO yanzu ana iya tsara taswira zuwa ƙwaƙwalwar ajiya, wanda ke rage adadin fitowar injin kama-da-wane kuma yana haɓaka aikin isar da na'urar zuwa injin kama-da-wane.
- A cikin lambar tsatsa, an yi aiki don maye gurbin ɓangarori marasa aminci tare da madadin aiwatar da aiwatarwa a cikin yanayin aminci. Ga sauran sassan marasa aminci, an ƙara cikakkun bayanai da ke bayanin dalilin da yasa za a iya ɗaukar ragowar lambar mara aminci.
source: budenet.ru