On Sun, Nov 18, 2012 at 12:02:17PM +0800, Fengguang Wu wrote:
> On Sun, Nov 18, 2012 at 10:30:57AM +1100, Dave Chinner wrote:
> > On Fri, Nov 16, 2012 at 02:42:42PM +0800, kbuild test robot wrote:
> > > tree: git://oss.sgi.com/xfs/xfs for-next
> > > head: 1813dd64057490e7a0678a885c4fe6d02f78bdc1
> > > commit: 612cfbfe174a89d565363fff7f3961a2dda5fb71 [68/70] xfs: add
> > > pre-write metadata buffer verifier callbacks
> > >
> > >
> > > sparse warnings:
> > >
> > > + fs/xfs/xfs_ialloc_btree.c:186:1: sparse: symbol 'xfs_inobt_verify' was
> > > not declared. Should it be static?
> > > fs/xfs/xfs_ialloc_btree.c:227:1: sparse: symbol 'xfs_inobt_read_verify'
> > > was not declared. Should it be static?
> > These are debug build only, and intended to be globally visible so
> > they can be changed with a debugger. And sparse didn't warn about
> > them, because you didn't do a debug build, so the script shouldn't
> > be changing them.(*)
> > Cheers,
> > Dave.
> > (*) Please don't do debug XFS builds with this automated checker. We
> > pretty much turn all static functions to global "noinline" functions
> > for debug builds, and sparse will throw hundreds of useless warnings.
> Dave, thanks for the explanations. And sorry for the noises!
> I'll disable the "Should it be static" checks against the xfs tree.
I didn't say that - I'm suggesting that you should be more selective
about what the bot throws out as a fix. i.e. only fix the newly
discovered problem rather than everything that is found....