BACKGROUND
The embodiments relate to guest operating system integrity validation, and more particularly to guest operating system integrity validation through a virtual machine monitor (VMM) hypervisor trap.
Today's computers make use of various memory partitioning and the use of virtual machines (VM) to increase operability and performance. More specifically, some of the reasons for using VMs are as follows. VMs can be used to consolidate the workload of underused computers (e.g., servers) within a single host machine, thus saving hardware, maintenance and administration costs. VMs can run legacy software applications where the legacy software may not be capable of running on the new host machine or with a new operating system (OS). This makes the choice of upgrading OS software an easier for system administration personnel since a VM can run the legacy OS applications. VMs can be used for running isolated programs for security purposes. VMs can be used to run multiple OSs or different versions of the same OS on a single host computer. The illusion of a hardware configuration not included on the host machine can be served by a VM. VMs provide isolation for applications and can provide tools, such as debugging, performance monitoring, fault injection, etc.
Traditionally, a layer of software is used to provide a virtual "real" machine to multiple instances of VMs. This software layer is known as the virtual machine monitor (VMM) or hypervisor. The VMM creates a series of virtual environments that run on top of the hardware. Therefore, each VM appears as a separate, unshared collection of resources. A VMM can be used in different scenarios to provide virtualization or logical partitioning. A VMM can run by itself directly on the real hardware without the need of a host OS. A VMM can be hosted and run as an application on top of a host OS. In this scenario, the VMM uses the host OS application program interface (API). And, if the VM architecture is different from the host machine, instruction set emulation would be required.
There are two types of VMM architectures typically used. The first type runs directly on the host machines hardware and typically has good performance. FIG. 1 illustrates this type of VMM architecture. The second type of VMM architecture uses the existing host OS abstractions to implement services, and generally has poorer performance. FIG. 2 illustrates this type of VMM architecture.
A VMM can be used for manageability and can read a host's file system prior to the system being booted. The problem with this is that systems today have continuous states of operation (e.g., suspend, hibernate, etc.) as compared to a power off.
DETAILED DESCRIPTION
FIG. 3 illustrates a block diagram of a process of an embodiment where a guest virtual machine (VM) can safely mount a read-only file system for a short period of time on another active guest VM live file system. Process 300 refers to the execution of a VM entity. In this embodiment, a first guest process (e.g., Guest A) and a second guest process (e.g., Guest B) are executing on a host machine (e.g., a host computer) that is partitioned, e.g., a capability partition, a management partition, a hypervisor running on top of the host hardware, etc. As illustrated in FIG. 3, process 300 begins with block 305 where a first process and a second process executing (i.e., running) on a host machine (e.g., host computer). In block 310, the second process traps to a process exit (i.e., VM exit) into a monitor process. In one embodiment, the monitor process is a virtual machine monitor (VMM). The trapping of the process exit for the second process can occur for any event that causes an exit to occur. Once the VM exit is trapped the second process is placed in a suspended state.
In block 310 a hypervisor (e.g., VMM) determines whether a predetermined amount of time has elapsed (e.g., one hour, eight hours, one day, etc.). In one embodiment, a clock is used to determine elapsed time and a process computes elapsed time from the last trap according to the present clock. In another embodiment, a counter s used to count up to a predetermined number (e.g., based on time). If the predetermined number has not been reached when a VM exit trap occurs for the second process, the counter increments. When the predetermined number has been met on a VM exit trap for the second process, the counter resets. In block 320, if the predetermined amount of time or number has not been met, process 300 continues with block310, otherwise process 300 continues with block 330.
In block 330, the hypervisor determines whether it is safe to map the first processes file system. In one embodiment, the hypervisor determines if the second process has masked interrupt signals. If the second process does have masked interrupt signals, these can potentially be in the top half of the processes device driver and it would be unsafe for the second process's file system to be redirected. In another embodiment, it is safe to map the first processes file system if a write is not in process. In this embodiment, an operating system on the VM manages a memory address that indicates status of a write in progress or not. Therefore, in the case of finding masked interrupt signals, process 300 continues with block 370 where the first and second processes can be scheduled by the hypervisor for execution on the host machine, that is, safe to map the first process file system on the second processes file system.
If the hypervisor determines that the second process does not have masked interrupt signals, in block 340 the hypervisor remaps the first processes resource. In one embodiment, the hypervisor remaps the integrated device electronics (IDE) direct memory access (DMA) to the first process. In another embodiment the hypervisor unmounts the first process's file system and mounts it on the second process. In yet another embodiment, the hypervisor does not unmount the first process's file system if DMA is disabled or a DMA transfer is detected as not in the process of transferring (e.g., reading a Bus Master IDE active bit). Note that the resource can be system resources, such as input/output (I/O) access, memory access, etc.
In one embodiment the hypervisor will not schedule the second process for execution until the first process has an unmounted file system. In another embodiment, the hypervisor will schedule the second process for execution if the first process is mounted on the second process and if DMA is disabled or a DMA transfer is detected as not in the process of transferring. Therefore, block 350 determines whether it is necessary to unmount the first process file system. If the first processes file system is necessary to be unmounted, process 300 continues at block 355 where the first processes file system is unmounted. If it is determined that it is not necessary to unmount the first processes file system, process 300 continues with block 360.
When the first process has its file system unmounted, it is safe to have the hypervisor perform a read-only analysis of the file system for configuration information to perform an integrity validation. Therefore, process 300 allows for a more frequent validation than systems that do not perform file-system analysis unless the host is rebooted. Thus, allowing integrity validation/accommodation independently of a system reboot. This is important in case the file system configuration information has been changed before the host is rebooted. In this case, restrictions or hardware behavior modification can be applied for actions, such as connecting to an isolated virtual local area network (VLAN).
In block 360 the resource (e.g., the IDE-DMA) is remapped to the second process (i.e., after allowing for a validation based on file system content). After block 360, in block 370 the hypervisor schedules the first and second processes execution on the host machine and process 300 continues to check for the next VM exit trap of the second process at block 310.
FIG. 4 illustrates a block diagram of a process that performs a file system integrity validation independently of a reboot on a host machine having a hypervisor architecture and modifies restrictions and/or hardware behavior if necessary. Similar to process 300, process 400 includes the addition of blocks 410 and 420.
After process 400 completes block 350, process 400 performs a read-only analysis of the host's file system for configuration information in block 410when it is safe to perform the read-only analysis in case any modifications have been made. This analysis occurs independently of the host rebooting. Therefore, the hypervisor uses the VM exit trap to start process 400, instead of waiting for a system reboot.
After block 410 is completed block 420 modifies host and VM restrictions and/or hardware behavior if necessary based on the analysis. Therefore, network connectivity access, privileges, etc. can be modified based on the analysis. Process 400 then continues as illustrated with process 300.
FIG. 5 illustrates a block diagram of a hypervisor architecture used in the above embodiments. As illustrated, reference numeral 510 is a first guest process (i.e., VM) running on host hardware 550. In one embodiment, guest process 510 is partitioned as a capability partition (i.e., containing system capability information). In one embodiment second guest process 520 is partitioned as a management partition. Memory 555 access and I/O of host hardware 550 is virtually mapped to the guest processes running on host hardware 550. Virtual file systems for the guest processes use the host memory file system on memory 555. The guest processes and host can communicate externally to other networks and/or machines through physical or wireless LAN 560.
The VMs running on host hardware 550 can communicate with each other via virtual network 530. Hypervisor (VMM) 545 keeps track of guest processes VM states, such as virtual central processing unit (CPU) state, register states, etc. As asserted above, when a VM exit is trapped for a first guest process into the hypervisor (VMM) the hypervisor will begin a process (i.e., process 300/400).
FIG. 6 illustrates an embodiment of a system. System 600 includes processor 610 connected to memory 620. In one embodiment memory 620 is a main memory, such as random-access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), read-only memory (ROM), etc. In another embodiment, memory 620 is a cache memory. System 600 includes process 601 is in the form of an executable process running in processor 610 and communicating with memory 620. In this embodiment process 601 is process 300 as described above. In another embodiment, process 601 is process 400 as described above. In system 600, process 602 and 603 are guest processes running on system 600. System 600 can be combined with other known elements depending on the implementation. For example, if system 600 is used in a multiprocessor system, other known elements typical of multiprocessor systems would be coupled to system 600. System 600 can be used in a variety of implementations, such as personal computers (PCs), personal desk assistants (PDAs), notebook computers, servers, MRTEs, dynamic binary optimizers, dynamic binary translators, etc.
The above embodiments can also be stored on a device or machine-readable medium and be read by a machine to perform instructions. The machine-readable medium includes any mechanism that provides (i.e., stores and/or transmits) information in a form readable by a machine (e.g., a computer). For example, a machine-readable medium includes read-only memory (ROM); random-access memory (RAM); magnetic disk storage media; optical storage media; flash memory devices; biological electrical, mechanical systems; electrical, optical, acoustical or other form of propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.). The device or machine-readable medium may include a micro-electromechanical system (MEMS), nanotechnology devices, organic, holographic, solid-state memory device and/or a rotating magnetic or optical disk. The device or machine-readable medium may be distributed when partitions of instructions have been separated into different machines, such as across an interconnection of computers or as different VMs.