Deep dive into Kernel

To enable SR-IOV, need kernel parameter for iommu edit /etc/default/grub GRUB_CMDLINE_LINUX_DEFAULT="pci=assign-busses iommu=pt intel_iommu=on" Next, execute update-grub $ update-grub On CentOS grub2-mkconfig -o /boot/grub2/grub.cfgIn a Intel machine, check dmesgIntel Machine$dmesg | grep -e DMAR -e IOMMU ... ACPI: DMAR 00000000ac7ebf18 00094 (v01 A M I OEMDMAR 00000001 INTL 00000001) Intel-IOMM..

KVM Interrupt related structures KVM Interrupt Injection to Guest IOAPIC based interrupt delivery MSI interrupt delivery kvm_irq_delivery_to_apic(kvm, NULL, &irq, NULL);- pass no kvm_lapic *src- pass no dest_mapformat : int kvm_irq_delivery_to_apic(struct kvm *kvm, struct kvm_lapic *src, struct kvm_lapic_irq *irq, unsigned long *dest_map) Jul 7 23:10:39 gaus02-m kernel: [49646.052828] # SON MSI ..

Interrupt Delivery analysis in KVM with SR-IOV MSI-X 에서 Multiple Queue 로 interrupt 를 전달하는 방법http://www.alexonlinux.com/msi-x-the-right-way-to-spread-interrupt-load

1. Source download root@gaus01-m:~# git clone git://libvirt.org/libvirt.git Cloning into 'libvirt'... remote: Counting objects: 146441, done. remote: Compressing objects: 100% (22449/22449), done. remote: Total 146441 (delta 123724), reused 146078 (delta 123482) Receiving objects: 100% (146441/146441), 146.43 MiB | 7.53 MiB/s, done. Resolving deltas: 100% (123724/123724), done. Checking connecti..

@include/block/block_int.h struct BlockDriverState /* * Note: the function bdrv_append() copies and swaps contents of * BlockDriverStates, so if you add new fields to this struct, please * inspect bdrv_append() to determine if the new fields need to be * copied as well. */ struct BlockDriverState { int64_t total_sectors; /* if we are reading a disk image, give its size in sectors */ int read_onl..

file: qemu-img.c ㅇ flags o 주요 변수들BlockDriverState *bs, *bs_old_backing = NULL, *bs_new_backing = NULL;BlockDriver *old_backing_drv, *new_backing_drv;int flags; o Open the image/* * Open the images. * * Ignore the old backing file for unsafe rebase in case we want to correct * the reference to a renamed or moved backing file. */ bs = bdrv_new_open("image", filename, fmt, flags, true, quiet); if (..

1. Write the command's and type(s) specification in the QAPI schema file (qapi-schema.json in the root source directory) { 'command': 'block_rebase', 'data': { 'device': 'str', 'file': 'str' } }Currently, I only support rebase from qcow2 to qcow2. edit qmp-commands.hx{ .name = "block_rebase", .args_type = "device:s,file:s", .mhandler.cmd_new = qmp_marshal_input_block_rebase, }, SQMPblock_rebase-..

QCOW2(QEMU Copy On Write) format disk 1. QCOW2 Creation 1. Create Base Image with 10G size diskqemu-img create -f qcow2 base.img 10G 2. Create Root Image which using base.imgqemu-img create -b base.img -f qcow2 root.img qemu-img info shows the chaining information of root.img sunshout@ubuntu-vm:/storage$ qemu-img info root.imgimage: root.imgfile format: qcow2virtual size: 10G (10737418240 bytes)..

Target model:- Guest VM with virtio NIC(eth0) and sr-iov VF NIC(eth1). 1. Create VF NIC in Host 1.1 Detect current SR-IOV network card information for VF creation.- lspci show PCI device information such as PCI Bus:Device:Function.- Following example shows that p4p1 is located at 04:00.0 (Bus:Device:Function). 1.2 Create VF using sysfs.- recently, creating VF is done by sysfs 2. Create Bridge In..

APICv 는 Guest OS가 APIC 을 Read/Write 할 때 VM Exit 가 발생하는데, 이를 해소하기 위한 H/W Feature 임Intel Xeon E2600 v2 Family 이상부터 지원하는 기능 http://www.linux-kvm.org/wiki/images/7/70/2012-forum-nakajima_apicv.pdf * Reviewing Unused and New Features for Interrupt/APIC Virtualizationhttp://www.linuxplumbersconf.org/2012/wp-content/uploads/2012/09/2012-lpc-virt-intel-vt-feat-nakajima.pdf * APIC Virtualization Performa..