Hi,
On 21-04-19 16:38, Wolfram Sang wrote:
On Sun, Apr 21, 2019 at 04:37:24PM +0200, Wolfram Sang wrote:
Hi,
Adding relevant people to CC.
Now really!
Original bug report: https://bugs.archlinux.org/task/62391
Thanks for the report!
May be related to this email: https://lkml.org/lkml/2019/4/4/518,
unconfirmed.
Yes, that's the one. Hans, there are two problems with the dev_err here.
One: It is not ratelimited. Two: We don't see which driver tried to
transfer something with a suspended adapter. My suggestion: Use
dev_WARN_ONCE(). And it seems we should think again about ratelimiting
in the core, too.
As the related hardware is quite specific, my question is more about
limiting the ammount of output the i2c_designware module produces when
encounters such errors.
Valid request. If we decide to move to dev_WARN_*, then I'd like to ask
you to run the fix again because behind all this, there is another
driver which needs fixing, and the WARN will point to it hopefully.
I fully agree with all of the above, the i2c-designware and the i2c-core
check should both be changed to dev_WARN_ONCE().
Once this is done, please reproduce the bug again (with a kernel using
dev_WARN_ONCE() for the i2c-designware check) and then please provide
the oops/backtrace the dev_WARN_ONCE() generates.
Regards,
Hans
