There are no WiFi adaptors in my system. I have observed this bug on both my hardwired 100/1000 NICs.
Note all my failures have been on 2.6.24-16. (Changed title) I did not run the -14 kernel.
I tried reproducing yesterday by pounding on the CPU with infinite make -j4 kernel compiles.
Both cores in my CPU were saturated for 3+ hours. No failures during this test.
Crashed once during the 24 hours, but I still can't figure out a good way to reproduce.
My best guess is this is triggered by an unusual condition in the networking stack, e.g. segmentation, checksum, or MTU misalignment.
Has anyone tried seeing if a serial console still responds after failure?
There are no WiFi adaptors in my system. I have observed this bug on both my hardwired 100/1000 NICs.
Note all my failures have been on 2.6.24-16. (Changed title) I did not run the -14 kernel.
I tried reproducing yesterday by pounding on the CPU with infinite make -j4 kernel compiles.
Both cores in my CPU were saturated for 3+ hours. No failures during this test.
Crashed once during the 24 hours, but I still can't figure out a good way to reproduce.
My best guess is this is triggered by an unusual condition in the networking stack, e.g. segmentation, checksum, or MTU misalignment.
Has anyone tried seeing if a serial console still responds after failure?