mirror of
https://github.com/cloud-hypervisor/cloud-hypervisor.git
synced 2024-11-09 21:30:07 +00:00
7bf0cc1ed5
This fixes all typos found by the typos utility with respect to the config file. Signed-off-by: Philipp Schuster <philipp.schuster@cyberus-technology.de>
146 lines
4.9 KiB
Rust
146 lines
4.9 KiB
Rust
// Copyright © 2022 Intel Corporation
|
|
//
|
|
// SPDX-License-Identifier: Apache-2.0
|
|
|
|
#![no_main]
|
|
|
|
use libfuzzer_sys::fuzz_target;
|
|
use seccompiler::SeccompAction;
|
|
use std::os::unix::io::{AsRawFd, FromRawFd};
|
|
use std::sync::Arc;
|
|
use virtio_devices::{VirtioDevice, VirtioInterrupt, VirtioInterruptType};
|
|
use virtio_queue::{Queue, QueueT};
|
|
use vm_memory::{bitmap::AtomicBitmap, Bytes, GuestAddress, GuestMemoryAtomic};
|
|
use vmm_sys_util::eventfd::{EventFd, EFD_NONBLOCK};
|
|
|
|
type GuestMemoryMmap = vm_memory::GuestMemoryMmap<AtomicBitmap>;
|
|
|
|
macro_rules! align {
|
|
($n:expr, $align:expr) => {{
|
|
(($n + $align - 1) / $align) * $align
|
|
}};
|
|
}
|
|
|
|
const QUEUE_DATA_SIZE: usize = 4;
|
|
const MEM_SIZE: usize = 32 * 1024 * 1024;
|
|
// Reuse what's being done from DeviceManager::get_msi_iova_space()
|
|
const IOVA_SPACE_SIZE: usize = (0xfeef_ffff - 0xfee0_0000) + 1;
|
|
|
|
// Max entries in the queue.
|
|
const QUEUE_SIZE: u16 = 256;
|
|
// Descriptor table alignment
|
|
const DESC_TABLE_ALIGN_SIZE: u64 = 16;
|
|
// Available ring alignment
|
|
const AVAIL_RING_ALIGN_SIZE: u64 = 2;
|
|
// Used ring alignment
|
|
const USED_RING_ALIGN_SIZE: u64 = 4;
|
|
// Descriptor table size
|
|
const DESC_TABLE_SIZE: u64 = 16_u64 * QUEUE_SIZE as u64;
|
|
// Available ring size
|
|
const AVAIL_RING_SIZE: u64 = 6_u64 + 2 * QUEUE_SIZE as u64;
|
|
// Used ring size
|
|
const USED_RING_SIZE: u64 = 6_u64 + 8 * QUEUE_SIZE as u64;
|
|
|
|
// Guest memory gap
|
|
const GUEST_MEM_GAP: u64 = 1 * 1024 * 1024;
|
|
// Guest physical address for descriptor table.
|
|
const DESC_TABLE_ADDR: u64 = align!(MEM_SIZE as u64 + GUEST_MEM_GAP, DESC_TABLE_ALIGN_SIZE);
|
|
// Guest physical address for available ring
|
|
const AVAIL_RING_ADDR: u64 = align!(DESC_TABLE_ADDR + DESC_TABLE_SIZE, AVAIL_RING_ALIGN_SIZE);
|
|
// Guest physical address for used ring
|
|
const USED_RING_ADDR: u64 = align!(AVAIL_RING_ADDR + AVAIL_RING_SIZE, USED_RING_ALIGN_SIZE);
|
|
// Virtio-queue size in bytes
|
|
const QUEUE_BYTES_SIZE: usize = (USED_RING_ADDR + USED_RING_SIZE - DESC_TABLE_ADDR) as usize;
|
|
|
|
fuzz_target!(|bytes| {
|
|
if bytes.len() < (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE)
|
|
|| bytes.len() > (QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE + MEM_SIZE)
|
|
{
|
|
return;
|
|
}
|
|
|
|
let (mut iommu, _) = virtio_devices::Iommu::new(
|
|
"fuzzer_iommu".to_owned(),
|
|
SeccompAction::Allow,
|
|
EventFd::new(EFD_NONBLOCK).unwrap(),
|
|
((MEM_SIZE - IOVA_SPACE_SIZE) as u64, (MEM_SIZE - 1) as u64),
|
|
None,
|
|
)
|
|
.unwrap();
|
|
|
|
let queue_data = &bytes[..QUEUE_DATA_SIZE];
|
|
let queue_bytes = &bytes[QUEUE_DATA_SIZE..QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE];
|
|
let mem_bytes = &bytes[QUEUE_DATA_SIZE + QUEUE_BYTES_SIZE..];
|
|
|
|
// Setup the request queue with the input bytes
|
|
let request_queue = setup_virt_queue(queue_data.try_into().unwrap());
|
|
// Given the "event queue" events are not handled from the current
|
|
// implementation of virtio-iommu, we simply setup the 'event_queue'
|
|
// with exactly the same content as the 'request_queue'.
|
|
let _event_queue = setup_virt_queue(queue_data.try_into().unwrap());
|
|
|
|
// Setup the guest memory with the input bytes
|
|
let mem = GuestMemoryMmap::from_ranges(&[
|
|
(GuestAddress(0), MEM_SIZE),
|
|
(GuestAddress(DESC_TABLE_ADDR), QUEUE_BYTES_SIZE),
|
|
])
|
|
.unwrap();
|
|
if mem
|
|
.write_slice(queue_bytes, GuestAddress(DESC_TABLE_ADDR))
|
|
.is_err()
|
|
{
|
|
return;
|
|
}
|
|
if mem.write_slice(mem_bytes, GuestAddress(0 as u64)).is_err() {
|
|
return;
|
|
}
|
|
let guest_memory = GuestMemoryAtomic::new(mem);
|
|
|
|
let request_evt = EventFd::new(0).unwrap();
|
|
let request_queue_evt = unsafe { EventFd::from_raw_fd(libc::dup(request_evt.as_raw_fd())) };
|
|
let _event_evt = EventFd::new(0).unwrap();
|
|
|
|
// Kick the 'queue' event before activate the vIOMMU device
|
|
request_queue_evt.write(1).unwrap();
|
|
|
|
iommu
|
|
.activate(
|
|
guest_memory,
|
|
Arc::new(NoopVirtioInterrupt {}),
|
|
vec![
|
|
(0, request_queue, request_evt),
|
|
(0, _event_queue, _event_evt),
|
|
],
|
|
)
|
|
.ok();
|
|
|
|
// Wait for the events to finish and vIOMMU device worker thread to return
|
|
iommu.wait_for_epoll_threads();
|
|
});
|
|
|
|
pub struct NoopVirtioInterrupt {}
|
|
|
|
impl VirtioInterrupt for NoopVirtioInterrupt {
|
|
fn trigger(&self, _int_type: VirtioInterruptType) -> std::result::Result<(), std::io::Error> {
|
|
Ok(())
|
|
}
|
|
}
|
|
|
|
fn setup_virt_queue(bytes: &[u8; QUEUE_DATA_SIZE]) -> Queue {
|
|
let mut q = Queue::new(QUEUE_SIZE).unwrap();
|
|
q.set_next_avail(bytes[0] as u16); // 'u8' is enough given the 'QUEUE_SIZE' is small
|
|
q.set_next_used(bytes[1] as u16);
|
|
q.set_event_idx(bytes[2] % 2 != 0);
|
|
q.set_size(bytes[3] as u16 % QUEUE_SIZE);
|
|
|
|
q.try_set_desc_table_address(GuestAddress(DESC_TABLE_ADDR))
|
|
.unwrap();
|
|
q.try_set_avail_ring_address(GuestAddress(AVAIL_RING_ADDR))
|
|
.unwrap();
|
|
q.try_set_used_ring_address(GuestAddress(USED_RING_ADDR))
|
|
.unwrap();
|
|
q.set_ready(true);
|
|
|
|
q
|
|
}
|