Arquitetura e infraestrutura
Publicado por Marcelo Soares Souza
licenciado sobre Creative Commons 3.0 Brasil
Autor: Thorsten Leemhuis / Copyright (c) 2012 Heise Media UK Ltd.
Traduzido: Marcelo Soares Souza
Originalmente: http://h-online.com/-1463482
Traduzido: Marcelo Soares Souza
Originalmente: http://h-online.com/-1463482
Uma grande quantidade de mudanças no código do gerenciador de memória irá corrigir diversos problemas e melhorar a performance em algumas áreas. O kernel agora suporte Large Physical Address Extension (LPAE) e realiza boot em sistemas x86 EFI mesmo sem o boot loader.
Um grande número de mudanças no Linux 3.3 tem como objetivo resolver vários problemas que podem, em certas situações, causar lentidão ou paradas temporárias quando o kernel esta escrevendo grandes quantidades de dados em discos lentos (tais como drivers USB flash). O Linux 3.2 já corrigiu alguns problemas deste tipo[1] onde o código de writeback era o culpado - agora as mudanças serão no código de compactação de memória e na correção de problemas relacionados ao transparent hugepages (THP).
Em novembro, LWN.net detalhou algumas histórias no artigo "Huge pages, slow drives, and long delays[2]"; os desenvolvedores trabalharam para a desativação do "synchronous compaction" mencionado no artigo com uma abordagem desenvolvida após a publicação deste. Detalhes e resultados de testes relacionados aos efeitos destas mudanças foram disponibilizados em quase quatrocentos linhas de comentário em um commit[3] que apenas remove três linhas de código mas que é apenas uma das muitas mudanças no código de compactação (1[4], 2[5], 3[6], 4[7], 5[8], 6[9], e outros).
Outra mudança[10] para o código de gerenciamento de memória irá ajudar a prevenir que outras fontes de problemas surjam enquanto se escreve em volumes de dados lentos e que esta relacionado com a infraestrutura de writeback. Com o drive fuse NTFS-3G, também é possível melhorar a performance de escrita em até cinqüenta porcentos em algumas circunstâncias especiais, como demostrado nos resultados de testes relatados na mensagem do commit desta solução.
Naturalização
Os desenvolvedores do Kernel incluíram algumas mudanças no Linux 3.3 categorizada como "memory control group naturalisation patches" que devem reduzir significativamente o gerenciamento para o controlador cgroup (1[12], 2[13], 3[14], e outros). Isto irá reduzir a sobrecarga do controlador sobre o uso da RAM; algumas distribuições já habilitaram isto por padrão. Em maio de 2011, a LWN.net disponibilizou algumas informações sobre o inicio do desenvolvimento desta solução no artigo "Integrating memory control groups[15]".
No Linux 3.3, o controlador de memória do cgroup oferece funções básicas para limitar a quantidade de RAM que o kernel pode alocar para si para gerenciar os processos em execução (1[16], 2[17]); o controlador do tamanho do buffer TCP mencionado na primeira parte [18] da série de artigos "Em Breve no Kernel 3.3" utiliza estas funções para limitar a quantidade de memória que os buffers, usados para a comunicação TCP, tem permissão para consumir.
Outra mudança[19] corrige um problema de performance no código readahead que surge quando um programa solicita uma grande quantidade de dados de um volume de dados rápido (tal como um SSD). Isto é causado por um efeito colateral indesejado de mudanças[20] no Linux 2.6.39; LWN.net fornece mais informações no artigo "What happened to disk performance in 2.6.39[21]".
Virtualização
Entre outras mudanças[22], o código do KVM agora pode fornecer uma unidade de monitoramento de performance virtual possibilitando que sistemas convidados realizem analises de performance e outras tarefas de rastreamento (1[23], 2[24], e outros).
O Suporte a descarte no código do Xen inclui agora uma funcionalidade opcional[25] para permanentemente remover dados em áreas liberadas dos dispositivos de armazenamento ("suporte a descarte com segurança na remoção"); uma requisição git-pull [26] de Konrad Rzeszutek Wilk, que trabalha na Oracle, menciona algumas mudanças especificas para o Xen. Os drivers balloon[27], blk[28], console[29] e net[30] Virtio agora inclui tudo o que é necessário necessários para ACPI S4, permitindo ao sistema convidado hibernar usando estas interfaces de troca de dados.
Arquitetura
Começando com o Linux 3.3[31], habilitando a opção de configuração "EFI_STUB" irá permitir que os kernels x86 sejam lançados diretos do EFI[32] sem a necessidade de um boot loader. É padrão agora [33] em sistemas x86, para o kernel, a emulação do processamento de funcionalidades vsyscall, uma interface herdada do vDSO que os programas podem usar para fazer syscalls; isto deve aprimorar a segurança[34] e foi planejado para uma versão anterior do kernel porém teve que ser atrasado[35] por conta de alguns problemas. Outra nova funcionalidade no Linux 3.3 é o suporte básico para ACPI 5.0, que apenas recentemente se tornou uma especificação oficial e ainda não foi visto em nenhum outro sistema (1[36], 2[37], 3[38], e outros).
O código ARM agora utiliza o large physical address extension (LPAE) (1[39], 2[40], 3[41], e outros), com o qual o sistema operacional 32-bit ARM v7 com o LPAE podem acessar mais do que 4 GB de RAM. O código ARM agora suporta[42] o subsistema de auditoria e provê suporte básico ao NVIDIA's Tegra 3 SoC (system on a chip) (1[43], 2[44], e outros).
O código para a arquitetura S390 agora pode acessar até 64 TB de RAM[45]; anteriormente, o limite era 3.8 TB. A arquitetura M68knommu agora suporta MMU (memory management unit) inclusa na quarta geração de processadores Coldfire. A versão 3.3 marca a primeira vez que o kernel suporta a arquitetura C6X.
Segurança
O kernel ganhou uma biblioteca baseado no código do GnuPG para cálculo com o MPI (multi-precision integers) (1[50], 2[51], 3[52], 4[53]). As tecnologias IMA[54] (integrity measurement architecture) e EVM[55] (extended verification module) para proteção contra ataques podem usar esta biblioteca para revisar a assinatura RSA e teste de arquivos (1[56], 2[57], 3[58], 4[59], 5[60]). Enquanto o Kernel 3.3 estava sendo desenvolvido, patches foram discutidos[61] para que este utilize esta infraestrutura para assinar módulos do kernel, mas os desenvolvedores ainda não chegaram a um acordo sobre o uso do mesmo.
Entre as mudanças[62] no subsistema crypto, alguns algoritmos de encriptação e assinatura foram aprimorados:
serpent - add 8-way parallel x86_64/SSE2 assembler implementation[63] serpent - add 4-way parallel i586/SSE2 assembler implementation[64] serpent-sse2 - add lrw support[65] serpent-sse2 - add xts support[66] twofish-x86_64-3way - add lrw support[67] twofish-x86_64-3way - add xts support[68]Diversos
Graças a algumas mudanças no código do ASPM (active state power management), o Linux 3.3 agora utiliza a tecnologia de conservação de energia do PCIe mesmo se [69] exista algumas inconsistências entre como o BIOS configura o hardware e o que o BIOS reporta para o kernel sobre o suporte para o ASPM. Algumas semanas após, este comportamento foi alteradoo no Linux 3.3, e os desenvolvedores do kernel também adicionaram este recurso nos Kernels 3.0.20 e 3.2.5 lançado em Fevereiro[70].
Melhorias para a infraestrutura IOMMU[71] inclui suporte[72] para a segunda versão da implementação da AMD e adiciona a habilidade para gerenciar páginas de memória de diferentes tamanhos. O código do IOMMU também ganhou algumas funções básicas de agrupamento e de isolamento seguro de dispositivos, funcionalidades particularmente interessantes para o VFIO[73] (Virtual Function I/O), que ainda esta em desenvolvimento.
De acordo com a documentação[74], o VFIO irá não apenas substituir as funções de troca entre componentes de hardware para convidados KVM mas também irá herdar o framework de drivers UIO, desde que o VFIO oferece mais possibilidades para tais drivers; um artigo[75] no LWN.net provê mais informações sobre o VFIO.
John Stultz contribuiu[76] com um script para combinar arquivos com a configuração do kernel dentro de um ".config". Muitas distribuições usam este tipo de scripts para constuir os seus kernel. Porém, o desenvolvedor do kernel Josh Boyer apontou que [77] este script incluso no Linux 3.3 precisa mais de 25 minutos em um cenário de teste, enquanto o script usado no Fedora precisa apenas de 47 segundos; mesmo após mais otimizações, o script ainda precisa mais de três minutos.
Outras melhorias
Muitas pequenas melhorias nesta versão do Kernel, mas de forma alguma insignificantes, as mudanças podem ser encontrados na lista abaixo. Esta lista contém os cabeçalhos dos commits referindo as respectivas mudanças.
ACPi, PCI and Co.
ACPI 5.0: Add new predefined names[83]
ACPI 5.0: Implement Connection() and AccessAs() changes[84]
ACPI 5.0: New interfaces to allow driver access to AML mutex objects[85]
ACPI, Add RAM mapping support to ACPI atomic IO support[86]
acpi/apei/einj: Add extensions to EINJ from rev 5.0 of acpi spec[87]
ACPI, APEI, GHES: Add PCIe AER recovery support[88]
ACPICA: Add support for region address conflict checking[89]
ACPI: kernel-parameters.txt : Add intel_idle.max_cstate[90]
ACPI / PM: Add Sony Vaio VPCCW29FX to nonvs blacklist.[91]
iommu: Add iommu_device_group callback and iommu_group sysfs entry[92]
iommu: Add option to group multi-function devices[93]
iommu/amd: Add driver stub for AMD IOMMUv2 support[94]
iommu/amd: Add function to get IOMMUv2 domain for pdev[95]
iommu/amd: Add iommuv2 flag to struct amd_iommu[96]
iommu/amd: Add stat counter for IOMMUv2 events[97]
iommu/amd: Add support for IOMMUv2 domain mode[98]
iommu/amd: Implement device aquisition code for IOMMUv2[99]
iommu/amd: Implement iommu_device_group[100]
iommu/amd: Implement IOMMUv2 TLB flushing routines[101]
iommu/amd: Implement notifiers for IOMMUv2[102]
iommu/amd: Put IOMMUv2 capable devices in pt_domain[103]
iommu/core: split mapping to page sizes as supported by the hardware[104]
iommu/intel: Implement iommu_device_group[105]
PCI: deprecate pci_scan_bus_parented()[106]
PCI: Introduce INTx check & mask API[107]
PCI/sysfs: add per pci device msi× irq listing (v5)[108]
PNP: work around Dell 1536/1546 BIOS MMCONFIG bug that breaks USB[109]
spi-topcliff-pch: Support new device LAPIS Semiconductor ML7831 IOH[110]
Crypto and Security
crypto: caam - add support for MD5 algorithm variants[111]
crypto: lrw - add interface for parallelized cipher implementions[112]
crypto: lrw - remove dependency on EXPERIMENTAL[113]
crypto: serpent - export common functions for x86_64/i386-sse2 assembler implementations[114]
crypto: talitos - add hmac algorithms[115]
crypto: talitos - support for channel remap and 2nd IRQ[116]
crypto: tcrypt - add lrw(serpent) tests[117]
crypto: tcrypt - add lrw(twofish) tests[118]
crypto: tcrypt - add serpent speed tests[119]
crypto: tcrypt - add test_acipher_speed[120]
crypto: tcrypt - add xts(serpent) tests[121]
crypto: tcrypt - add xts(twofish) tests[122]
crypto: testmgr - add lrw(serpent) test vectors[123]
crypto: testmgr - add lrw(twofish) test vectors[124]
crypto: testmgr - add new serpent test vectors[125]
crypto: testmgr - add xts(serpent) test vectors[126]
crypto: testmgr - add xts(twofish) test vectors[127]
crypto: xts: add interface for parallelized cipher implementations[128]
crypto: xts - remove dependency on EXPERIMENTAL[129]
Documentation: clarify the purpose of LSMs[130]
MM
kmemleak: Add support for memory hotplug[131]
kmemleak: Handle percpu memory allocation[132]
memblock: Kill early_node_map[][133]
mm: add free_hot_cold_page_list() helper[134]
mm: more intensive memory corruption debugging[135]
mm: simplify find_vma_prev()[136]
mm,slub,x86: decouple size of struct page from CONFIG_CMPXCHG_LOCAL[137]
mm-tracepoint: fix documentation and examples[138]
mm-tracepoint: rename page-free events[139]
mm,x86,um: move CMPXCHG_DOUBLE config option[140]
mm,x86,um: move CMPXCHG_LOCAL config option[141]
slab: add taint flag outputting to debug paths.[142]
slab: introduce slab_max_order kernel parameter[143]
slub: add taint flag outputting to debug paths[144]
slub: document setting min order with debug_guardpage_minorder > 0[145]
slub: fix slub_max_order Documentation[146]
tracepoint: add tracepoints for debugging oom_score_adj[147]
Power Management
[CPUFREQ] EXYNOS: Make EXYNOS common cpufreq driver[148]
freezer: don't unnecessarily set PF_NOFREEZE explicitly[149]
freezer: unexport refrigerator() and update try_to_freeze() slightly[150]
PM/Devfreq: Add Exynos4-bus device DVFS driver for Exynos4210/4212/4412.[151]
PM / Domains: Add device stop governor function (v4)[152]
PM / Hibernate: Remove deprecated hibernation snapshot ioctls[153]
PM / Hibernate: Remove deprecated hibernation test modes[154]
PM: Run the driver callback directly if the subsystem one is not there[155]
PM / Sleep: Recommend [un]lock_system_sleep() over using pm_mutex directly[156]
Tracing
events, sched: Add tracepoint for accounting blocked time[157]
ftrace: Do not function trace inlined functions[158]
ktest: Allow bisect test to restart where it left off[159]
ktest: Allow overriding bisect test results[160]
ktest: Allow success logs to be stored[161]
ktest: Save test output[162]
ktest: When creating a new config, ask for BUILD_OPTIONS[163]
oprofile, s390: Add event interface to the System z hardware sampling module[164]
oprofile, x86: Add kernel parameter oprofile.cpu_type=timer[165]
oprofile, x86: Reimplement nmi timer mode using perf event[166]
perf: Add support for PERF_HW_COUNT_REF_CPU_CYCLES[167]
perf, core: Rate limit perf_sched_events jump_label patching[168]
perf events: Add Intel x86 mapping for PERF_COUNT_HW_REF_CPU_CYCLES[169]
perf events: Add PERF_COUNT_HW_REF_CPU_CYCLES generic PMU event[170]
perf script: Add comm filtering option[171]
perf script: Add generic perl handler to process events[172]
perf script: Implement option for system-wide profiling[173]
perf test: Add more automated tests for event parsing[174]
perf tools: Add ability to synthesize event according to a sample[175]
perf tools: Look up thread names for system wide profiling[176]
perf tools: Save some loops using perf_evlist__id2evsel[177]
perf, x86: Expose perf capability to other modules[178]
selftests: new x86 breakpoints selftest[179]
tracing: Add entries in buffer and total entries to default output header[180]
tracing: Add irq, preempt-count and need resched info to default trace output[181]
tracing: Have stack tracing set filtered functions at boot[182]
tracing: update Documentation on max preds limit[183]
x86, perf: Add a build-time sanity test to the x86 decoder[184]
Virtualisation
KVM: Document KVM_NMI[185]
KVM: expose latest Intel cpu new features (BMI1/BMI2/FMA/AVX2) to guest[186]
KVM guest: remove KVM guest pv mmu support[187]
KVM: MMU: remove KVM host pv mmu support[188]
KVM: MMU: remove oos_shadow parameter[189]
vhost-net: add module alias (v2.1)[190]
virtio: harsher barriers for rpmsg.[191]
xen: Add privcmd device driver[192]
xen: Add xenbus_backend device[193]
xen: Add xenbus device driver[194]
xenbus: Support HVM backends[195]
xen: document backend sysfs files[196]
xen: document balloon driver sysfs files[197]
xen/granttable: Introducing grant table V2 stucture[198]
Various
cgroup: introduce cgroup_taskset and use it in subsys->can_attach(), cancel_attach() and attach()[199]
cgroup: kill subsys->can_attach_task(), pre_attach() and attach_task()[200]
coccinelle.txt: update documentation to include M= option[201]
cpu: convert 'cpu' and 'machinecheck' sysdev_class to a regular subsystem[202]
c/r: procfs: add start_data, end_data, start_brk members to /proc/$pid/stat v4[203]
debugfs: add tools to printk 32-bit registers[204]
debugobjects: Extend to assert that an object is initialized[205]
doc: Add load/store guarantees to Documentation/atomic-ops.txt[206]
Documentation: add missing tainted bits to Documentation/sysctl/kernel.txt[207]
dt/i2c: Enumerate some of the known trivial i2c devices[208]
i2c-designware: add OF binding support[209]
kbuild: Add support for an "archheaders" target[210]
lockdep: Update documentation for lock-class leak detection[211]
MAINTAINERS: stable: Update address[212]
module_param: avoid bool abuse, add bint for special cases.[213]
module_param: check that bool parameters really are bool.[214]
module_param: make bool parameters really bool (arch)[215]
module_param: make bool parameters really bool (core code)[216]
module_param: make bool parameters really bool (drivers & misc)[217]
module_param: make bool parameters really bool (drivers/video/i810)[218]
module_param: make bool parameters really bool (net & drivers/net)[219]
modules: sysfs - export: taint, coresize, initsize[220]
NFSD: Added fault injection script[221]
nohz: Allow rcu extended quiescent state handling seperately from tick stop[222]
nohz: Remove tick_nohz_idle_enter_norcu() / tick_nohz_idle_exit_norcu()[223]
nohz: Separate out irq exit and idle loop dyntick logic[224]
of/address: Add reg-names property to name an iomem resource[225]
of/irq: Add interrupts-names property to name an irq resource[226]
pinctrl: add a pin_base for sparse gpio-ranges[227]
pinctrl: add a pin config interface[228]
pinctrl: documentation update[229]
pinctrl: don't create a device for each pin controller[230]
pinctrl: GPIO direction support for muxing[231]
pinctrl: make a copy of pinmux map[232]
pinctrl: pass name instead of device to pin_config_*[233]
pinctrl: some typo fixes[234]
rcu: Add documentation for raw SRCU read-side primitives[235]
rcu: Add rcutorture CPU-hotplug capability[236]
rcu: Add rcutorture system-shutdown capability[237]
rcu: Control rcutorture startup from kernel boot parameters[238]
rcu: Document failing tick as cause of RCU CPU stall warning[239]
rcu: Permit RCU_FAST_NO_HZ to be used by TREE_PREEMPT_RCU[240]
rcu: Track idleness independent of idle tasks[241]
sched, nohz: Implement sched group, domain aware nohz idle load balancing[242]
sched: Only queue remote wakeups when crossing cache boundaries[243]
sched/tracing: Add a new tracepoint for sleeptime[244]
scripts/coccinelle: improve the coverage of some semantic patches[245]
selftests: new very basic kernel selftests directory[246]
stable: update documentation to ask for kernel version[247]
sysctl: add the kernel.ns_last_pid control[248]
x86
Next: Minor Gems - Architecturepci, x86/io-apic: Allow PCI_IOAPIC to be user configurable on x86[249]
x86_64, asm: Optimise fls(), ffs() and fls64()[250]
x86: Add NumaChip support[251]
x86: Add workaround to NMI iret woes[252]
x86/amd: Add missing feature flag for fam15h models 10h-1fh processors[253]
x86, centaur: Enable cx8 for VIA Eden too[254]
x86/config: Revamp configuration for MID devices[255]
x86/div64: Add a micro-optimization shortcut if base is power of two[256]
[N6] x86: Generate system call tables and unistd_*.h from tables[257]
x86/intel config: Revamp configuration to allow for Moorestown and Medfield[258]
x86: Machine-readable syscall tables and scripts to process them[259]
x86, mce, therm_throt: Don't report power limit and package level thermal throttle events in mcelog[260]
x86/microcode_amd: Add support for CPU family specific container files[261]
x86, NMI: Add knob to disable using NMI IPIs to stop cpus[262]
x86, NMI: Add NMI IPI selftest[263]
x86: Panic on detection of stack overflow[264]
x86: Simplify code by removing a !SMP #ifdefs from 'struct cpuinfo_x86'[265]
x86/tools: Add decoded instruction dump mode[266]
x86: Update instruction decoder to support new AVX formats[267]
ARM
ARM: 7169/1: topdown mmap support[268]
ARM: 7201/1: add EDAC atomic_scrub function[269]
ARM: 7202/1: Add Cortex-A7 proc info[270]
ARM: 7206/1: Add generic ARM instruction set condition code checks.[271]
ARM: 7240/1: Make ARCH_NR_GPIO a Kconfig variable[272]
ARM: Add arm_memblock_steal() to allocate memory away from the kernel[273]
ARM: Allow Kconfig to control the definition of NR_BANKS[274]
ARM: at91/net: add macb ethernet controller in 9g45/9g20 DT[275]
ARM: at91: removal of CAP9 SoC family[276]
arm/dt: add basic usb nodes to tegra device trees[277]
ARM: dts: Add intial dts file for EXYNOS4210 SoC, SMDKV310 and ORIGEN[278]
arm/dts: Add minimal device tree support for omap2420 and omap2430[279]
ARM: EXYNOS: Add Exynos4 device tree enabled board file[280]
ARM: EXYNOS: Add USB OHCI device[281]
ARM: EXYNOS: Add USB OHCI support to ORIGEN board[282]
ARM: EXYNOS: Add USB OHCI support to SMDKV310 board[283]
ARM: gic: allow GIC to support non-banked setups[284]
arm/imx6: add imx6q sabrelite board support[285]
arm/imx6q: Rename Sabreauto to Armadillo2[286]
ARM i.MX: Merge i.MX5 support into mach-imx[287]
arm: introduce cross trigger interface helpers[288]
ARM: LPAE: Add context switching support[289]
ARM: LPAE: Add fault handling support[290]
ARM: LPAE: Add identity mapping support for the 3-level page table format[291]
ARM: LPAE: add ISBs around MMU enabling code[292]
ARM: LPAE: Factor out classic-MMU specific code into proc-v7-2level.S[293]
ARM: LPAE: Invalidate the TLB before freeing the PMD[294]
ARM: LPAE: mark memory banks with start > ULONG_MAX as highmem[295]
ARM: LPAE: Move page table maintenance macros to pgtable-2level.h[296]
ARM: LPAE: Move the FSR definitions to separate files[297]
ARM: LPAE: Page table maintenance for the 3-level format[298]
ARM: mach-shmobile: Add kota2 defconfig.[299]
ARM: mach-shmobile: Add marzen defconfig.[300]
ARM: mach-shmobile: bonito: Add dip-switch selector[301]
ARM: mach-shmobile: bonito: add FPGA irq demux[302]
ARM: mach-shmobile: bonito: Add LCDC0 support[303]
ARM: mach-shmobile: bonito: add PMIC settings[304]
ARM: mach-shmobile: bonito: add SMSC9221 support[305]
ARM: mach-shmobile: r8a7740: add PFC support.[306]
ARM: mach-shmobile: r8a7779 and Marzen base support V2[307]
ARM: mach-shmobile: r8a7779 PFC function support[308]
ARM: mach-shmobile: r8a7779 PFC GPIO-only support V2[309]
ARM: mach-shmobile: r8a7779 SMP support V3[310]
ARM: move iotable mappings within the vmalloc region[311]
ARM: mxs: add saif clkmux functions[312]
ARM: OMAP1: Move dpll1 rates selection from config to runtime[313]
ARM: OMAP4: Fix errata i688 with MPU interconnect barriers.[314]
ARM: OMAP4: PM: Add CPUidle support[315]
ARM: OMAP4: PM: Add CPUX OFF mode support[316]
ARM: OMAP4: PM: Add SAR RAM support[317]
ARM: OMAP: Add Secure HAL and monitor mode API infrastructure.[318]
ARM: OMAP: am33xx: Update common OMAP machine specific sources[319]
ARM: OMAP: am33xx: Update common omap platform files[320]
ARM: OMAP: PM: Add support to allocate the memory for secure RAM[321]
ARM: OMAP: PRCM: add support for chain interrupt handler[322]
ARM: OMAP: TI814X: Create board support and enable build for TI8148 EVM[323]
ARM: OMAP: TI81XX: Prepare for addition of TI814X support[324]
ARM: perf: add support for stalled cycle ABI events[325]
arm: Remove mach-tcc8k directory[326]
arm: Remove plat-tcc directory[327]
ARM: S3C64XX: Implement basic power domain support[328]
ARM: S5P64X0: Add HSMMC setup for host Controller[329]
ARM: SAMSUNG: Consolidation of SPI platform devices to plat-samsung[330]
ARM: stop: execute platform callback from cpu_stop code[331]
arm/tegra: Add device-tree support for TrimSlice board[332]
arm/tegra: add support for tegra30 based board cardhu[333]
arm/tegra: pinmux tables and definitions for tegra30[334]
ARM: Update mach-types[335]
ARM: vic: device tree binding[336]
ARM: w90x900: add common nuc9xx.h[337]
cpufreq: OMAP: Add SMP support for OMAP4+[338]
MX1:apf9328: Add i2c support[339]
of: Add device tree selftests[340]
OMAP: DSS2: Support for UMSH-8173MD TFT panel[341]
omap-serial: Add minimal device tree support[342]
PPC
fsl-rio: Add two ports and rapidio message units support[343]
fsl-rio: Split rio driver into two parts, RapidIO endpoint and message unit[344]
powerpc/40x: Add APM8018X SOC support[345]
powerpc/44x: Enable CONFIG_RELOCATABLE for PPC44x[346]
powerpc/44x: Enable CRASH_DUMP for 440x[347]
powerpc/44x: Enable DYNAMIC_MEMSTART for 440x[348]
powerpc/476fpe: Add 476fpe SoC code[349]
powerpc/47x: Add support for the new IBM currituck platform[350]
powerpc/85xx: Add a P1010RDB 36-bit address map device tree[351]
powerpc/85xx: Add crypto engine to P1010 SoC device tree[352]
powerpc/85xx: Add P1020RDB 36-bit address map device tree[353]
powerpc/85xx: create dts components to build up an SoC[354]
powerpc/85xx: Rework MPC8536DS device trees[355]
powerpc/85xx: Rework MPC8544DS device tree[356]
powerpc/85xx: Rework MPC8548CDS device trees[357]
powerpc/85xx: Rework MPC8568MDS device tree[358]
powerpc/85xx: Rework MPC8569MDS device tree[359]
powerpc/85xx: Rework MPC8572DS device tree[360]
powerpc/85xx: Rework P1010RDB and P1010 device tree[361]
powerpc/85xx: Rework P1020RDB device tree[362]
powerpc/85xx: Rework P1021MDS device tree[363]
powerpc/85xx: Rework P1022DS device tree[364]
powerpc/85xx: Rework P1023RDS device tree[365]
powerpc/85xx: Rework P2020DS device tree[366]
powerpc/85xx: Rework P2020RDB device tree[367]
powerpc/85xx: Rework P2041RDB device tree[368]
powerpc/85xx: Rework P3041DS device tree[369]
powerpc/85xx: Rework P3060QDS device tree[370]
powerpc/85xx: Rework P4080DS device trees[371]
powerpc/85xx: Rework P5020DS device tree[372]
powerpc/85xx: Rework PCI nodes on P1020RDB[373]
powerpc/85xx: Update SRIO device tree nodes[374]
powerpc: Add support for OpenBlockS 600[375]
powerpc/book3e: Add Chroma as a new WSP/PowerEN platform.[376]
powerpc/book3e: Add ICSWX/ACOP support to Book3e cores like A2[377]
powerpc/boot: Add mfdcrx[378]
powerpc/cpuidle: Add cpu_idle_wait() to allow switching of idle routines[379]
powerpc/cpuidle: cpuidle driver for pSeries[380]
powerpc/cpuidle: Enable cpuidle and directly call cpuidle_idle_call() for pSeries[381]
powerpc: Enable Hugetlb by default for 32-bit 85xx/corenet[382]
powerpc: Enable hugetlb by default for corenet64 platforms[383]
powerpc/fsl: Add support for Integrated Flash Controller[384]
powerpc/fsl: Document rapidio node binding-information[385]
powerpc: Kconfig updates for FSL BookE HUGETLB 64-bit[386]
powerpc: POWER7 optimised copy_to_user/copy_from_user using VMX[387]
powerpc/powernv: PCI support for p7IOC under OPAL v2[388]
powerpc: Process dynamic relocations for kernel[389]
powerpc: Rename mapping based RELOCATABLE to DYNAMIC_MEMSTART for BookE[390]
powerpc: Split ICSWX ACOP and PID processing[391]
Various
blackfin: bf537: add capture support[392]
blackfin: bf548: add capture support[393]
blackfin: bf561: add adv7183 capture support[394]
C6X: devicetree support[395]
fs: binfmt_elf: create Kconfig variable for PIE randomization[396]
Initial support for the Ubiquiti Networks XM board (rev 1.0).[397]
m68k: add cache support for V4e ColdFire cores running with MMU enabled[398]
m68k: add code to setup a ColdFire 54xx platform when MMU enabled[399]
m68k: add ColdFire 54xx CPU MMU memory init code[400]
m68k: add ColdFire FPU support for the V4e ColdFire CPUs[401]
m68k: add ColdFire paging exception handling code[402]
m68k: add ColdFire with MMU enabled support to the m68k mem init code[403]
m68k: add machine and CPU definitions for ColdFire cores[404]
m68k: add page table size definitions for ColdFire V4e MMU[405]
m68k: add TASK definitions for ColdFires running with MMU[406]
m68k: add TLB flush support for the ColdFire V4e MMU hardware[407]
m68k: allow ColdFire 547x and 548x CPUs to be built with MMU enabled[408]
m68k: ColdFire V4e MMU paging init code and miss handler[409]
m68k: ColdFire with MMU enabled uses same clocking code as non-MMU[410]
m68k: compile appropriate mm arch files for ColdFire MMU support[411]
m68k: create ColdFire MMU pgalloc code[412]
m68k: definitions for the ColdFire V4e MMU hardware[413]
m68k/Kconfig: Separate classic m68k and coldfire early[414]
m68k/mac: early console[415]
m68k: modify user space access functions to support ColdFire CPUs[416]
m68k/mvme16x: Add support for EARLY_PRINTK[417]
m68k: page table support definitions and code for ColdFire MMU[418]
m68k: use addr_limit checking for m68k CPUs that do no support address spaces[419]
microblaze: generic atomic64 support[420]
MIPS: Add fast get_user_pages[421]
MIPS: Alchemy: Au1300 SoC support[422]
MIPS: Alchemy: Basic support for the DB1300 board.[423]
MIPS: Alchemy: Drop MIRAGE/BOSPORUS board support[424]
MIPS: Alchemy: Improved DB1550 support, with audio and serial busses.[425]
MIPS: Alchemy: merge Au1000 and Au1300-style IRQ controller code.[426]
MIPS: Alchemy: Merge PB1200 support into DB1200 code.[427]
MIPS: Alchemy: one kernel for DB1000/DB1500/DB1100[428]
MIPS: Alchemy: remove PB1000 support[429]
MIPS: ath79: add common USB Host Controller device[430]
MIPS: ath79: Add early printk support for the AR933X SoCs[431]
MIPS: ath79: Add initial support for the Atheros AP121 reference board[432]
MIPS: ath79: add revision id for the AR933X SoCs[433]
MIPS: BCM63XX: Add support for bcm6368 CPU.[434]
MIPS: BMIPS: Add SMP support code for BMIPS43xx/BMIPS5000[435]
MIPS: GIO bus support for SGI IP22/28[436]
MIPS: Initial PCI support for Atheros 724x SoCs.[437]
MIPS: Netlogic: Add basic MSI support for XLR/XLS[438]
MIPS: Netlogic: Add default XLP config.[439]
MIPS: Netlogic: Add support for XLP 3XX cores[440]
MIPS: Netlogic: Add XLP makefiles and config[441]
MIPS: Netlogic: Add XLP platform files for XLP SoC[442]
MIPS: Oceton: Update model detection code for new chips.[443]
MIPS: Octeon: Add support for OCTEON II PCIe[444]
MIPS: Octeon: Move some Ethernet support files out of staging.[445]
MIPS: Octeon: Update SOC PCI related register definitions for new chips.[446]
MIPS: randomize PIE load address[447]
S390: Add s390x description to Documentation/kdump/kdump.txt[448]
S390: entry[64].S improvements[449]
S390: Remove Kerntypes leftovers[450]
sh: Kill off remaining private runtime PM bits.[451]
sh: sh7723: use runtime PM implementation, common with arm/mach-shmobile[452]
sh: sh7724: use runtime PM implementation, common with arm/mach-shmobile[453]...
Caso tenha gostado do post, por favor, prestigie a fonte para este e outros assuntos afins
o 
agradece sua preferência, volte sempre! 
ORIGEM CRIACIONAL: Marcelo Junta DadosVivamos a LIBERDADE com total DIGNIDADE!
