4. The MSI Driver Guide HOWTO

Authors:Tom L Nguyen; Martine Silbermann; Matthew Wilcox
Copyright:2003, 2008 Intel Corporation

4.1. About this guide

This guide describes the basics of Message Signaled Interrupts (MSIs),the advantages of using MSI over traditional interrupt mechanisms, howto change your driver to use MSI or MSI-X and some basic diagnostics totry if a device doesn’t support MSIs.

4.2. What are MSIs?

A Message Signaled Interrupt is a write from the device to a specialaddress which causes an interrupt to be received by the CPU.

The MSI capability was first specified in PCI 2.2 and was later enhancedin PCI 3.0 to allow each interrupt to be masked individually. The MSI-Xcapability was also introduced with PCI 3.0. It supports more interruptsper device than MSI and allows interrupts to be independently configured.

Devices may support both MSI and MSI-X, but only one can be enabled ata time.

4.3. Why use MSIs?

There are three reasons why using MSIs can give an advantage overtraditional pin-based interrupts.

Pin-based PCI interrupts are often shared amongst several devices.To support this, the kernel must call each interrupt handler associatedwith an interrupt, which leads to reduced performance for the system asa whole. MSIs are never shared, so this problem cannot arise.

When a device writes data to memory, then raises a pin-based interrupt,it is possible that the interrupt may arrive before all the data hasarrived in memory (this becomes more likely with devices behind PCI-PCIbridges). In order to ensure that all the data has arrived in memory,the interrupt handler must read a register on the device which raisedthe interrupt. PCI transaction ordering rules require that all the dataarrive in memory before the value may be returned from the register.Using MSIs avoids this problem as the interrupt-generating write cannotpass the data writes, so by the time the interrupt is raised, the driverknows that all the data has arrived in memory.

PCI devices can only support a single pin-based interrupt per function.Often drivers have to query the device to find out what event hasoccurred, slowing down interrupt handling for the common case. WithMSIs, a device can support more interrupts, allowing each interruptto be specialised to a different purpose. One possible design givesinfrequent conditions (such as errors) their own interrupt which allowsthe driver to handle the normal interrupt handling path more efficiently.Other possible designs include giving one interrupt to each packet queuein a network card or each port in a storage controller.

4.4. How to use MSIs

PCI devices are initialised to use pin-based interrupts. The devicedriver has to set up the device to use MSI or MSI-X. Not all machinessupport MSIs correctly, and for those machines, the APIs described belowwill simply fail and the device will continue to use pin-based interrupts.

4.4.1. Include kernel support for MSIs

To support MSI or MSI-X, the kernel must be built with the CONFIG_PCI_MSIoption enabled. This option is only available on some architectures,and it may depend on some other options also being set. For example,on x86, you must also enable X86_UP_APIC or SMP in order to see theCONFIG_PCI_MSI option.

4.4.2. Using MSI

Most of the hard work is done for the driver in the PCI layer. The driversimply has to request that the PCI layer set up the MSI capability for thisdevice.

To automatically use MSI or MSI-X interrupt vectors, use the followingfunction:

int pci_alloc_irq_vectors(struct pci_dev *dev, unsigned int min_vecs,              unsigned int max_vecs, unsigned int flags);

which allocates up to max_vecs interrupt vectors for a PCI device. Itreturns the number of vectors allocated or a negative error. If the devicehas a requirements for a minimum number of vectors the driver can pass amin_vecs argument set to this limit, and the PCI core will return -ENOSPCif it can’t meet the minimum number of vectors.

The flags argument is used to specify which type of interrupt can be usedby the device and the driver (PCI_IRQ_LEGACY, PCI_IRQ_MSI, PCI_IRQ_MSIX).A convenient short-hand (PCI_IRQ_ALL_TYPES) is also available to ask forany possible kind of interrupt. If the PCI_IRQ_AFFINITY flag is set,pci_alloc_irq_vectors() will spread the interrupts around the available CPUs.

To get the Linux IRQ numbers passed torequest_irq() andfree_irq() and thevectors, use the following function:

int pci_irq_vector(struct pci_dev *dev, unsigned int nr);

Any allocated resources should be freed before removing the device usingthe following function:

void pci_free_irq_vectors(struct pci_dev *dev);

If a device supports both MSI-X and MSI capabilities, this API will use theMSI-X facilities in preference to the MSI facilities. MSI-X supports anynumber of interrupts between 1 and 2048. In contrast, MSI is restricted toa maximum of 32 interrupts (and must be a power of two). In addition, theMSI interrupt vectors must be allocated consecutively, so the system mightnot be able to allocate as many vectors for MSI as it could for MSI-X. Onsome platforms, MSI interrupts must all be targeted at the same set of CPUswhereas MSI-X interrupts can all be targeted at different CPUs.

If a device supports neither MSI-X or MSI it will fall back to a singlelegacy IRQ vector.

The typical usage of MSI or MSI-X interrupts is to allocate as many vectorsas possible, likely up to the limit supported by the device. If nvec islarger than the number supported by the device it will automatically becapped to the supported limit, so there is no need to query the number ofvectors supported beforehand:

nvec = pci_alloc_irq_vectors(pdev, 1, nvec, PCI_IRQ_ALL_TYPES)if (nvec < 0)        goto out_err;

If a driver is unable or unwilling to deal with a variable number of MSIinterrupts it can request a particular number of interrupts by passing thatnumber to pci_alloc_irq_vectors() function as both ‘min_vecs’ and‘max_vecs’ parameters:

ret = pci_alloc_irq_vectors(pdev, nvec, nvec, PCI_IRQ_ALL_TYPES);if (ret < 0)        goto out_err;

The most notorious example of the request type described above is enablingthe single MSI mode for a device. It could be done by passing two 1s as‘min_vecs’ and ‘max_vecs’:

ret = pci_alloc_irq_vectors(pdev, 1, 1, PCI_IRQ_ALL_TYPES);if (ret < 0)        goto out_err;

Some devices might not support using legacy line interrupts, in which casethe driver can specify that only MSI or MSI-X is acceptable:

nvec = pci_alloc_irq_vectors(pdev, 1, nvec, PCI_IRQ_MSI | PCI_IRQ_MSIX);if (nvec < 0)        goto out_err;

4.4.3. Legacy APIs

The following old APIs to enable and disable MSI or MSI-X interrupts shouldnot be used in new code:

pci_enable_msi()              /* deprecated */pci_disable_msi()             /* deprecated */pci_enable_msix_range()       /* deprecated */pci_enable_msix_exact()       /* deprecated */pci_disable_msix()            /* deprecated */

Additionally there are APIs to provide the number of supported MSI or MSI-Xvectors:pci_msi_vec_count() andpci_msix_vec_count(). In general theseshould be avoided in favor of letting pci_alloc_irq_vectors() cap thenumber of vectors. If you have a legitimate special use case for the countof vectors we might have to revisit that decision and add apci_nr_irq_vectors() helper that handles MSI and MSI-X transparently.

4.4.4. Considerations when using MSIs

4.4.4.1. Spinlocks

Most device drivers have a per-device spinlock which is taken in theinterrupt handler. With pin-based interrupts or a single MSI, it is notnecessary to disable interrupts (Linux guarantees the same interrupt willnot be re-entered). If a device uses multiple interrupts, the drivermust disable interrupts while the lock is held. If the device sendsa different interrupt, the driver will deadlock trying to recursivelyacquire the spinlock. Such deadlocks can be avoided by usingspin_lock_irqsave() or spin_lock_irq() which disable local interruptsand acquire the lock (see Documentation/kernel-hacking/locking.rst).

4.4.5. How to tell whether MSI/MSI-X is enabled on a device

Using ‘lspci -v’ (as root) may show some devices with “MSI”, “MessageSignalled Interrupts” or “MSI-X” capabilities. Each of these capabilitieshas an ‘Enable’ flag which is followed with either “+” (enabled)or “-” (disabled).

4.5. MSI quirks

Several PCI chipsets or devices are known not to support MSIs.The PCI stack provides three ways to disable MSIs:

  1. globally
  2. on all devices behind a specific bridge
  3. on a single device

4.5.1. Disabling MSIs globally

Some host chipsets simply don’t support MSIs properly. If we’relucky, the manufacturer knows this and has indicated it in the ACPIFADT table. In this case, Linux automatically disables MSIs.Some boards don’t include this information in the table and so we haveto detect them ourselves. The complete list of these is found near thequirk_disable_all_msi() function in drivers/pci/quirks.c.

If you have a board which has problems with MSIs, you can pass pci=nomsion the kernel command line to disable MSIs on all devices. It would bein your best interests to report the problem tolinux-pci@vger.kernel.orgincluding a full ‘lspci -v’ so we can add the quirks to the kernel.

4.5.2. Disabling MSIs below a bridge

Some PCI bridges are not able to route MSIs between busses properly.In this case, MSIs must be disabled on all devices behind the bridge.

Some bridges allow you to enable MSIs by changing some bits in theirPCI configuration space (especially the Hypertransport chipsets suchas the nVidia nForce and Serverworks HT2000). As with host chipsets,Linux mostly knows about them and automatically enables MSIs if it can.If you have a bridge unknown to Linux, you can enableMSIs in configuration space using whatever method you know works, thenenable MSIs on that bridge by doing:

echo 1 > /sys/bus/pci/devices/$bridge/msi_bus

where $bridge is the PCI address of the bridge you’ve enabled (eg0000:00:0e.0).

To disable MSIs, echo 0 instead of 1. Changing this value should bedone with caution as it could break interrupt handling for all devicesbelow this bridge.

Again, please notifylinux-pci@vger.kernel.org of any bridges that needspecial handling.

4.5.3. Disabling MSIs on a single device

Some devices are known to have faulty MSI implementations. Usually thisis handled in the individual device driver, but occasionally it’s necessaryto handle this with a quirk. Some drivers have an option to disable useof MSI. While this is a convenient workaround for the driver author,it is not good practice, and should not be emulated.

4.5.4. Finding why MSIs are disabled on a device

From the above three sections, you can see that there are many reasonswhy MSIs may not be enabled for a given device. Your first step shouldbe to examine your dmesg carefully to determine whether MSIs are enabledfor your machine. You should also check your .config to be sure youhave enabled CONFIG_PCI_MSI.

Then, ‘lspci -t’ gives the list of bridges above a device. Reading/sys/bus/pci/devices/*/msi_bus will tell you whether MSIs are enabled (1)or disabled (0). If 0 is found in any of the msi_bus files belongingto bridges between the PCI root and the device, MSIs are disabled.

It is also worth checking the device driver to see whether it supports MSIs.For example, it may contain calls to pci_alloc_irq_vectors() with thePCI_IRQ_MSI or PCI_IRQ_MSIX flags.