On Sun, Sep 05, 2021 at 05:12:39PM +0200, Greg KH wrote: > On Sun, Sep 05, 2021 at 09:56:58AM -0500, Ian Pilcher wrote: > > On 9/5/21 9:51 AM, Greg KH wrote: > > > On Sun, Sep 05, 2021 at 09:39:57AM -0500, Ian Pilcher wrote: > > > > On 9/4/21 1:01 AM, Greg KH wrote: > > > > > Please never use WARN_ON() in new code unless the machine is really > > > > > broken and you can not do anything else here. > > > > > > > > Wait what? I thought that was BUG_ON. > > > > > > Not whan panic-on-warn is set, which is getting more and more common > > > these days. > > > > Fair enough. What is the recommend approach to reporting a "this should > > never" happen situation these days? > > dev_err() and handle the error properly. > > WARN_ON is the right choice for reporting recoverable kernel bugs, and BUG_ON for unrecoverable ones; see the two comments in include/asm-generic/bug.h which explain this. Please don't use dev_err() if it's a kernel bug (and not just unexpected input from userspace or hardware behaving weirdly), as that prevents the bug from being reported if it occurs. Greg, you've been corrected on this before, e.g. https://lore.kernel.org/linux-fsdevel/20210707023548.15872-1-desmondcheongzx@xxxxxxxxx/T/#u. Please stop spreading false information as it is destroying your credibility :-( - Eric
