Linux 7.3 Adds Support for Non-Fatal PCI Errors and New Intel Chips

Linux 7.3 will incorporate changes in the PCI/PCIe subsystem to report certain non-fatal errors, add Intel hardware, and improve power and communication functions between devices.

• The kernel will add support for non-fatal advisory errors as outlined in PCI Express 7.0.
• The VMD driver will receive new identifiers for Intel Dunlow and Nova Lake.
• NVIDIA Vera Rubin, Arrow Lake HX systems, and certain PCIe ports will see improvements in compatibility and power management.

The development of Linux 7.3 is progressing with a new set of changes for the PCI and PCI Express subsystem, a fundamental part of the kernel that allows the operating system to communicate with cards, processors, storage, and other connected devices. The updates include support for errors that have remained hidden until now, new hardware identifiers, and adjustments aimed at improving compatibility in servers, workstations, and machines of various architectures.

This information was reported by Phoronix, which reviewed the integration request for the changes in the PCI/PCIe subsystem for the upcoming kernel version. The set includes fixes, specific features for certain devices, and improvements in power management, although it still corresponds to a development version and not a final release of Linux 7.3.

PCIe Errors That Will No Longer Go Unnoticed

One of the most relevant changes is the incorporation of support for so-called non-fatal advisory errors, known in technical documentation as Advisory Non-Fatal Errors. The PCI Express 7.0 specification allows for the indication of certain non-fatal conditions through this category, but Linux had masked them by default and had not reported them to the system until now.

The adjustment does not mean that every advisory will interrupt the operation of the equipment or that an alert necessarily implies a critical failure. Its importance lies in providing greater visibility into events that can help diagnose communication or operational problems in the PCIe interconnection, especially on platforms where preventive monitoring is important.

The difference between a fatal error and a non-fatal one also helps to put the change in context: the former can compromise the continuity of an operation, while the latter allows the system to continue functioning under certain conditions. By reporting the advisories outlined by PCI Express 7.0, the kernel can provide more information to administrators and diagnostic tools without automatically turning every event into a crash.

The modification, therefore, aims to improve hardware observability and does not promise that PCIe failures will disappear. Its utility will depend on how each platform implements the specification, the quality of its drivers, and how users interpret the messages produced by the system.

New Identifiers for Intel Dunlow and Nova Lake

The code for Linux 7.3 also adds new device identifiers for Intel Nova Lake and Dunlow within the VMD driver. According to the available information, the changes are limited to recognizing these identifiers, without requiring additional modifications in the driver logic to enable basic support.

The name Dunlow appeared during the summer and is considered associated with entry-level Intel Xeon processors, aimed at server and workstation platforms. The source information describes it as a platform derived from Nova Lake, although that characterization is presented as an estimate and not as a fully confirmed technical specification in the kernel announcement.

VMD, short for Volume Management Device, is a technology used by Intel to manage certain PCIe storage devices through the processor and platform firmware. The inclusion of identifiers allows the kernel to correctly recognize new hardware designs when they arrive at compatible systems, a small step in the code but necessary to prevent recent components from being left without proper identification.

The arrival of these identifiers does not equate to a complete performance evaluation nor does it confirm all commercial features of Dunlow or Nova Lake. It does show that the kernel's development tree already considers future hardware and that Linux compatibility is being prepared before these platforms achieve a broader presence in servers or workstations.

DMA Between Devices and Arrow Lake HX Systems

Among the new features, NVIDIA Vera Rubin appears on the list of platforms compatible with peer-to-peer DMA, or P2PDMA. This function allows certain PCIe devices to exchange data directly, without relying on the conventional path through main memory, which can be useful in workloads that move large volumes of information between accelerators and other components.

Support on the kernel list does not guarantee that all Vera Rubin configurations will utilize P2PDMA identically or that any combination of devices will be compatible. The function requires coordination between hardware, drivers, and PCIe topology, so its availability represents a software foundation that specific systems will need to validate.

The Intel driver, for its part, receives support for VMCONFIG BUS_RESTRICT_CFG=3. This change allows listing descendant devices in Intel Arrow Lake HX systems, a necessary operation for the operating system to identify components connected behind certain segments or ports of the platform.

Enumeration is an essential stage of booting and managing PCIe devices, as the kernel needs to know what hardware exists and where it is located before allocating resources and loading drivers. In Arrow Lake HX, the new configuration seeks to resolve a specific discovery limitation, without the available information indicating broader changes in the overall functioning of those systems.

Warnings, Power, and Specific Fixes

Linux 7.3 will also show a warning when user space writes to configuration regions that are reserved exclusively for the kernel. This measure adds a diagnostic signal against operations that may interfere with normal device control, although the announcement does not detail that all such writes will be automatically blocked.

The separation between user space and kernel is especially important in device management, as improper modification of registers can produce hard-to-trace behaviors. A warning provides additional information to detect tools or workflows attempting to alter protected configurations, without attributing a universal blocking policy to this change.

Another adjustment allows the use of low power state D3 on root ports with native hot-plug capability on non-x86 platforms. State D3 reduces the power consumption of a PCIe device when it does not need to remain fully active, but support depends on the platform properly managing power transitions.

The change is not enabled for all x86 platforms, as some older x86 designs were not adequately validated for the low power state D3. This caution reflects that a feature available in the specification may not always be activated broadly without specific firmware, hardware, and recovery path testing.

Ultimately, the kernel will avoid the low power sleep state L0s on the Realtek RTS525A PCI Express card reader. On that device, using L0s ended up causing an AER interrupt storm, an excessive flow of alerts related to recovery and advanced error reporting of PCI Express.

By disabling that transition for the specific reader, Linux prioritizes stability over potential energy savings that resulted in negative consequences. The change falls into the category of specific kernel tweaks or quirks, where an exception for a particular model may be preferable to applying a general rule that harms other devices.

What the Changes Mean for Linux Users

The integration of these modifications primarily benefits those managing servers, workstations, and machines with recent hardware, rather than those using a desktop installation without specialized PCIe devices. The new identifiers may enhance the recognition of Intel platforms, while the reporting and power adjustments help investigate issues that previously could remain without visible explanation.

In environments with accelerators, fast storage, or multiple connected devices, P2PDMA and a more comprehensive enumeration may influence how the system leverages the PCIe topology. However, the source does not provide performance figures, so it would not be correct to anticipate concrete gains in speed, latency, or consumption from these compatibility changes.

Non-fatal warning errors should also be interpreted with caution: receiving more information does not mean the equipment is about to fail. Administrators and developers will need to distinguish between isolated alerts, recurring issues, and events associated with a specific device or driver, especially when the system maintains its operations normally.

As Linux 7.3 is still in development, details may receive new corrections before the final version. The integration described corresponds to the PCI/PCIe changes that were incorporated during the specified week, and its final adoption will depend on the kernel's review and stabilization cycle.

The set confirms a common trend in Linux development: much of the visible user improvements arise from very specific tweaks in identification, diagnostics, and power management. In this case, the goal is not a single high-profile feature, but a PCIe platform better prepared for emerging hardware and more transparent regarding errors that previously remained hidden.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.