[DdhD]

Summary

Use p->bdf.domainno instead of hardcoded PCI domain 0 when
constructing PciHandleType in probe_capability_pci_express().

This preserves the device's complete PCI address and makes capability
probing consistent with probe_pci(), which already uses the actual
PCI domain.

Problem

On my dual-socket Intel Xeon 6746E (Sierra Forest) Linux system,
an Ethernet controller exists at 0001:60:00.0, while
0000:60:00.0 does not exist.

PCM discovers the device in domain 0001, but the hardcoded domain
causes capability probing to access 0000:60:00.0 instead.
As a result, pcm-iio exits with:

PCM ERROR. Exception PCM error: can't open PciHandle 0:96:0:0

Validation

I tested this one-line change locally on the affected system.

Before the change, sudo ./pcm-iio 1 failed with the exception above.
After applying the change and rebuilding, pcm-iio ran successfully
without that exception.

This validates the fix for the reported device-discovery failure,
not the accuracy of the bandwidth measurements.

Fixes #1055

[Copilot]

Copilot review overview

🟢 Approval recommended

The focused change correctly fixes multi-domain PCI capability access without affecting domain-zero devices.

Review effort: Balanced
Findings: None

What changed in this PR

Fixes #1055 by preserving the PCI domain during PCIe capability probing.

Changes:

• Replaces hardcoded domain 0 with the device’s actual domain.
• Aligns capability probing with probe_pci().

File	Description
src/​lspci.cpp	Uses the complete PCI address when opening the device.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.