Re: [PATCH RFC 0/2] PCI: rockchip-host: Retry training and hot-reset a JMB585 on RK3399

[Igor Velkov]

On Wed, 28 Oct 2026 17:31:17 +0000, Enrique Hernández Bello wrote:
> pin enumeration to an A53 and see whether your aborts all become synchronous.

Helios64, v7.3 with Shawn's hook registered before pci_host_probe(), no
other PCIe patches, a 1 s startup delay on the PCIe power regulator (DT
overlay), s2idle for 30 s:
- all CPUs: 2 aborts in 40 cycles, on CPU0 and CPU1;
- maxcpus=4 (A53 only): 2 aborts in 40 cycles, on CPU1 and CPU0.
All four are synchronous (ESR 0x96000210) in rockchip_pcie_rd_conf(),
from pci_power_up() in pci_pm_resume_noirq(). The hook caught no SError
in 80 cycles, and no abort landed on an A72.