Messages in this thread | | | Date | Mon, 8 Jun 2015 17:26:16 +0300 | Subject | Re: [PATCH] debug: Deprecate BUG_ON() use in new code, introduce CRASH_ON() | From | Alexey Dobriyan <> |
| |
Ingo Molnar wrote:
> These days a BUG_ON() is almost always fatal due to unreleased locks, > plus we still don't print kernel crashes to the graphical console, > so they are silent hard lockups in 99% of the cases.
Is it a regression?
I've just tested with "echo c >/proc/sysrq-trigger" and kernel.panic_on_oops=1 and the box just hanged. But I clearly remember it started to work at some point and it was a miracle. Recently a colleague of mine show such graphical panic (most of it as always scrolled away BTW).
It's usual setup with i915 driver, X and fbconsole which is never used.
As for BUG_ON() I hope rule will be relaxed for "not under locks" situations and something will be done with inevitable miryads of trivial patches.
Count them:
$ chgrep -e BUG_ON -w -n | wc -l 9660
| |