+ Alex
On 4/22/2024 8:34 PM, Thinh Tran wrote:
The patch fixes an issue when repeated EEH reports with a single error
on the bus of Intel X710 4-port 10G Base-T adapter, in the MSI domain
causing the devices to be permanently disabled. It fully resets and
restart the devices when handling the PCI EEH error.
Two new functions, i40e_io_suspend() and i40e_io_resume(), have been
introduced. These functions were refactor from the existing
i40e_suspend() and i40e_resume() respectively. This refactoring was
done due to concerns about the logic of the I40E_SUSPENSED state, which
caused the device not able to recover. The functios are now used in the
EEH handling for device suspend/resume callbacks.
- In the PCI error detected callback, replaced i40e_prep_for_reset()
with i40e_io_suspend(). The chance is to fully suspend all I/O
operations
- In the PCI error slot reset callback, replaced pci_enable_device_mem()
with pci_enable_device(). This change enables both I/O and memory of
the device.
- In the PCI error resume callback, replace i40e_handle_reset_warning()
with i40e_io_resume(). This change allows the system to resume I/O
operations
You don't mark a target tree, I believe you're sending this as a bug
fix? If so, can you mark it with '[Patch iwl-net]' and provide a Fixes:
tag.
Thanks,
Tony
Signed-off-by: Thinh Tran <thin...@linux.ibm.com>
Tested-by: Robert Thomas <rob.tho...@ibm.com>
