Hello,
My company is working on creating a robust system that can handle our Xilinx Kintex 7 going offline without taking Linux with it. Unfortunately, If we take a working system and unconfigure the fpga while writing to the pcie space we get a series of events that cause the cpu to hang. Any help is greatly appreciated!
Fault sequence.
1)The Fpga is unconfigured
2)The mmaped pcie space is no longer valid and throws a imprecise external abort on the next write.
3)The default abort handler is called in drivers/pci/host/pci-imx6.c (imx6q_pcie_abort_handler()).
4)The default abort handler returns 0 "abort Handled".
5)The next write to the fpga's pcie space causes the cpu to hang.
We are working on tracking this down with the Dstream debugger but so far no luck.
CPU: Arm IMX6Q
FPGA: Xilinx Kintex 7
Kernel: freescale 3.10.53
Hello Cole Peterson,
did you managed to solve this issue?
We have a similar problem with iMX6Q hanging itself while we take down EP without notice.
Any advice would be welcomed.
Regards,
Primož
Hi Cole
one can check patch below for imprecise abort issue
[PATCH v6 3/3] PCI: imx6: Add support for i.MX6 PCIe controller
Best regards
igor
Hello Igor,
Thank-you for the response but unfortunately the patch doesn't help. The abort handler is not called on step 5 when the cpu hangs.
-Cole