msix: Handle MSI-X device masking

As mentioned in the PCI specification, the Function Mask from the
Message Control Register can be set to prevent a device from injecting
MSI-X messages. This supersedes the vector masking as it interacts at
the device level.

Here quoted from the specification:
For MSI and MSI-X, while a vector is masked, the function is prohibited
from sending the associated message, and the function must set the
associated Pending bit whenever the function would otherwise send the
message. When software unmasks a vector whose associated Pending bit is
set, the function must schedule sending the associated message, and
clear the Pending bit as soon as the message has been sent. Note that
clearing the MSI-X Function Mask bit may result in many messages
needing to be sent.

This commit implements the behavior described above by reorganizing
the way the PCI configuration space is being written. It is indeed
important to be able to catch a change in the Message Control
Register without having to implement it for every PciDevice
implementation. Instead, the PciConfiguration has been modified to
take care of handling any update made to this register.

Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
This commit is contained in:
Sebastien Boeuf
2019-06-06 18:46:11 -07:00
committed by Rob Bradford
parent d810c7712d
commit 4d98dcb077
5 changed files with 98 additions and 56 deletions

View File

@@ -12,7 +12,6 @@ extern crate vm_allocator;
extern crate vm_memory;
extern crate vmm_sys_util;
use byteorder::{ByteOrder, LittleEndian};
use libc::EFD_NONBLOCK;
use std::sync::atomic::{AtomicUsize, Ordering};
use std::sync::Arc;
@@ -211,6 +210,8 @@ impl VirtioPciDevice {
let pci_device_id = VIRTIO_PCI_DEVICE_ID_BASE + device.device_type() as u16;
let msix_config = Arc::new(Mutex::new(MsixConfig::new(msix_num)));
let configuration = PciConfiguration::new(
VIRTIO_PCI_VENDOR_ID,
pci_device_id,
@@ -220,6 +221,7 @@ impl VirtioPciDevice {
PciHeaderType::Device,
VIRTIO_PCI_VENDOR_ID,
pci_device_id,
Some(msix_config.clone()),
);
Ok(VirtioPciDevice {
@@ -232,7 +234,7 @@ impl VirtioPciDevice {
queue_select: 0,
msix_config: 0,
},
msix_config: Arc::new(Mutex::new(MsixConfig::new(msix_num))),
msix_config,
msix_num,
device,
device_activated: false,
@@ -366,15 +368,16 @@ impl PciDevice for VirtioPciDevice {
let msix_config = self.msix_config.clone();
let cb = Arc::new(Box::new(move |queue: &Queue| {
let entry = &msix_config.lock().unwrap().table_entries[queue.vector as usize];
let config = &mut msix_config.lock().unwrap();
let entry = &config.table_entries[queue.vector as usize];
// In case the vector control register associated with the entry
// has its first bit set, this means the vector is masked and the
// device should not inject the interrupt.
// Instead, the Pending Bit Array table is updated to reflect there
// is a pending interrupt for this specific vector.
if entry.vector_ctl == 0x0000_0001u32 {
msix_config.lock().unwrap().set_pba_bit(queue.vector, false);
if config.is_masked() || entry.is_masked() {
config.set_pba_bit(queue.vector, false);
return Ok(());
}
@@ -384,14 +387,6 @@ impl PciDevice for VirtioPciDevice {
self.interrupt_cb = Some(cb);
}
fn config_registers(&self) -> &PciConfiguration {
&self.configuration
}
fn config_registers_mut(&mut self) -> &mut PciConfiguration {
&mut self.configuration
}
fn ioeventfds(&self) -> Vec<(&EventFd, u64, u64)> {
let bar0 = self
.configuration
@@ -578,24 +573,11 @@ impl BusDevice for VirtioPciDevice {
}
fn write_config_register(&mut self, reg_idx: usize, offset: u64, data: &[u8]) {
if offset as usize + data.len() > 4 {
return;
}
let regs = self.config_registers_mut();
match data.len() {
1 => regs.write_byte(reg_idx * 4 + offset as usize, data[0]),
2 => regs.write_word(
reg_idx * 4 + offset as usize,
u16::from(data[0]) | (u16::from(data[1]) << 8),
),
4 => regs.write_reg(reg_idx, LittleEndian::read_u32(data)),
_ => (),
}
self.configuration
.write_config_register(reg_idx, offset, data);
}
fn read_config_register(&self, reg_idx: usize) -> u32 {
self.config_registers().read_reg(reg_idx)
self.configuration.read_config_register(reg_idx)
}
}