Re: [PATCH] BUILD_BUG_ON sucks

[Andrew Morton <akpm@linux-foundation.org>]

On Sun, 17 Aug 2008 08:19:06 -0400 Theodore Tso <tytso@mit.edu> wrote:

> On Sat, Aug 16, 2008 at 10:46:58AM -0700, Andrew Morton wrote:
> > >       what this code is supposed to do?
> > > 
> > > 	journal = handle->h_transaction->t_journal;
> > > 	if (!journal_set_features(journal, 0, 0, JFS_FEATURE_INCOMPAT_REVOKE)) {
> > > 		J_ASSERT (!"Cannot set revoke feature!");
> > > 			 ^^^^
> > 
> > lol.  It's been there since I merged ext3 in 2.4.15.  Probably it was
> > in sct's ext3 patches in the RH kernel.
> > 
> > Don't change it - it might be important!
> 
> Heh!  Well, it does the right thing, and doesn't take any extra text
> space assuming a vaguely competent C compiler optimizer.  :-)
> 
> I'm pretty sure that back in the 2.4 days, we didn't have BUG_ON.  We
> should do a s/J_ASSERT/BUG_ON/g pass over all of fs/jbd and fs/jbd2.
> I'll submit patches for application when the 2.6.27 merge window opens
> up --- or is this an obvious enough and safe enough transformation
> that it will get accepted mainline at this point?
> 

May as well get it over and done with.

We presently have a mix of J_ASSERT, J_ASSERT_JH and J_ASSERT_BH.  That
will become BUG_ON, J_ASSERT_JH and J_ASSERT_BH.  Which a is slightly
unpleasing loss of consistency but whatever.