On Tue, 2 Jun 2026 12:12:39 +0100 Yury M. wrote:

On 6/2/26 03:50, Jakub Kicinski wrote:

quoted

On Wed, 27 May 2026 11:55:35 +0000 Yury Murashka wrote:  

quoted

During PCIe hot-plug events, uncorrectable errors can be reported and
AER recovery for the tg3 device is initiated by the AER kernel driver.
The tg3_io_error_detected function is the AER error recovery handler.  

IDK, I really hate this NAPI specific tracking. Can we add a callback
to struct pci_error_handlers to let the driver know that things went
fully belly up and the driver should pack up its toys? Sounds to me
like the key issue here is that there's a side exit thru the state
machine.  

1. We already have a similar check in the bnxt driver. Please check the 
BNXT_STATE_NAPI_DISABLED flag.

I'm painfully aware.

2. I'm trying to fix a specific issue in this specific driver.
3. Your suggested changes are a long-term option for the fix. They 
require a redesign of the AER feature. Plus, I'm not sure they will be 
accepted.

There's only one way to find out.