README

@@ -6,17 +6,24 @@ Compile test & install kernel image for vz5's VM.
 Dependencies :
 ==============
 
-- virsh
 - expect
+- qemu/kvm
 - `apt build-dep linux-image-amd64`
 - `apt install linux-source`
 
+Optionnal dependencies :
+------------------------
+- virsh
+- debvm
+
 Install & config :
 ==================
 
 Copy sample config `cp config.mk.inc config.mk`
 Edit configuration file to meet your needs `editor config.mk`
 
+Then you will need kernel sources & kernel configuration.
+
 Kernel sources  :
 -----------------
 
@@ -45,10 +52,32 @@ git -C kernel-config commit -m 'Initial kernel config'
 Usage :
 =======
 
+Get current status with `make`
+
 Compile a kernel `make kernel`
 
 Test a kernel on virsh vm `make test`
+	If make test is run with root, we will try to start TEST_VM_NAME with virsh
+	If make test is run without privileges, a disk image will be built using debvm
+	and the tests will be run using qemu.
+	NOTE : the user needs access to /dev/kvm (adduser username kvm)
 
 Install kernel `make install`
 
 Commit tested kernel config `make commit-config`
+
+Notes :
+=======
+
+All scripts & tests are designed for monolitic kernel without initrd.
+
+Kernel config should contains :
+```
+# CONFIG_BLK_DEV_INITRD=n
+# CONFIG_MODULES=n
+```
+
+KVM has a documentation about kernel tuning for guest that can be found here :
+https://www.linux-kvm.org/page/Tuning_Kernel#Kernel_for_guest_with_paravirtualization
+
+The script `scripts/checkconfig.sh` check that kernel config fits basic needs.

scripts/checkconfig.sh

@@ -0,0 +1,48 @@
+#!/bin/sh
+set -e
+
+if [ "$#" -lt 1 ]
+then
+	echo "Usage : $0 configfile" 2>/dev/null
+	exit 1
+fi
+config=$1
+err=0
+
+while read rule
+do
+	opt="$(echo $rule | cut -d ":" -f 1)"
+	comment="$(echo $rule | cut -d ":" -f 2-)"
+	#echo "opt='$opt'"
+	#echo "comment='$comment'"
+	printf "%-40s" "$opt"
+	if grep "$opt" "$config" >/dev/null
+	then
+		echo " [$(tput setaf 2)OK$(tput sgr0)]"
+	else
+		err=1
+		echo "[$(tput setaf 1)FAIL$(tput sgr0)] $comment"
+	fi
+done << __EOF__
+# CONFIG_BLK_DEV_INITRD is not set:Initrd should be disabled
+# CONFIG_MODULES is not set:Kernel should be monolitic (without module support)
+CONFIG_VIRTIO=y:Needed bu VIRTIO_PCI and VIRTIO_BALOON
+CONFIG_VIRTIO_MENU=y:The virtio drivers
+CONFIG_VIRTIO_NET=y:The virtio network driver
+CONFIG_VIRTIO_BLK=y:The virtio block device driver
+CONFIG_VIRTIO_PCI=y:Support for virtio PCI devices
+CONFIG_VIRTIO_MMIO:Support for memory mapped virtio driver
+CONFIG_VIRTIO_MEM=y:The virtio memory driver
+CONFIG_VIRTIO_BALLOON=y:Support for hot memory amount change for KVM guest
+CONFIG_VIRTIO_INPUT=y:The virtio input (kbd, mice) driver
+CONFIG_VIRTIO_CONSOLE=y:Support for virtio serial console
+CONFIG_VIRTIO_VDPA=y:Virtio data path acceleration support
+CONFIG_VP_VDPA=y:Bridges virtio PCI to vDPA
+CONFIG_PCI_MSI=y:Allows driver to enable Message Signaled Interrupts
+CONFIG_KVM_GUEST=y:Optimization for KVM guest
+CONFIG_PARAVIRT=y:Optimization for linux guest
+CONFIG_MEMORY_HOTPLUG=y:Needed by VIRTIO_BALLOON
+CONFIG_MEMORY_HOTREMOVE=y:Needed by VIRTIO_BALLOON
+__EOF__
+
+exit $err