On 19. 04. 24, 17:38, Tony Nguyen wrote:
On 4/19/2024 2:03 AM, Jiri Slaby wrote:
On 19. 04. 24, 10:44, Jiri Slaby wrote:
This crashes the kernel as a spinlock is held upper in the call stack
in e1000_watchdog_task():
spin_lock(&adapter->stats64_lock);
e1000e_update_stats(adapter);
-> e1000e_update_phy_stats()
-> e1000e_read_phy_reg_mdic()
-> usleep_range() ----> Boom.
It was reported to our bugzilla:
https://bugzilla.suse.com/show_bug.cgi?id=1223109
I believe, the mainline has the same bug.
Any ideas?
Obviously change the usleeps back to udelays? Or maybe only when
retry_enabled is set?
Hi Jiri,
Should be the former. Could you give this patch a try?
https://lore.kernel.org/intel-wired-lan/20240417190320.3159360-1-vitaly.lifshits@xxxxxxxxx/T/#u
Thanks, now pushed to the openSUSE kernel.
--
js
suse labs
