Messages in this thread | | | From | Eric Dumazet <> | Date | Thu, 27 Jan 2022 18:53:14 -0800 | Subject | Re: [PATCH net] net: dev: Detect dev_hold() after netdev_wait_allrefs() |
| |
On Thu, Jan 27, 2022 at 6:48 PM Jann Horn <jannh@google.com> wrote:
> When someone is using NET_DEV_REFCNT_TRACKER for slow debugging, they > should also be able to take the performance hit of > CONFIG_PCPU_DEV_REFCNT and rely on the normal increment-from-zero > detection of the generic refcount code, right? (Maybe > NET_DEV_REFCNT_TRACKER should depend on !CONFIG_PCPU_DEV_REFCNT?)
NET_DEV_REFCNT_TRACKER is not slow, I think it has neglectable cost really. (I could not see any difference in my tests)
Also getting a trap at the exact moment of the buggy dev_hold_track() is somewhat better than after-fact checking.
In your case, linkwatch_add_event() already uses dev_hold_track() so my proposal would detect the issue right away.
> > My intent with the extra check in free_netdev() was to get some > limited detection for production systems that don't use > NET_DEV_REFCNT_TRACKER.
Understood
| |