On Wed, Nov 6, 2019 at 5:17 PM Mark Brown broonie@kernel.org wrote:
On Wed, Nov 06, 2019 at 04:07:52PM +0000, Robin Murphy wrote:
FWIW this smells like a builtin driver had its of_device_id table marked __init, leaving drv->of_match_table as a dangling pointer to freed memory by this point.
Indeed, in fact I sent a fix for this to Linus Walleij yesterday having seen the relevant build warning when testing -next. Someone already reported that it fixed the boot issues. Hopefully Linus will pick it up soon :/
Yeah picked it up and pushed out now. I wish I'd been quicker with it but the patch spot activity has been high. (Bad signal-to-noise ratio on the mailing lists.)
I wonder if it's worth to look at the static checkers like checkpatch to warn for this?
There is always a bit of delicate balance between just fixing some weird one-off problems and making sure they never happen again.
Yours, Linus Walleij