Robert Elz writes: > Date: Sun, 04 Oct 2015 11:52:18 +1100 > From: matthew green <m...@eterna.com.au> > Message-ID: <24080.1443919...@splode.eterna.com.au> > > > | how about this: > > Looks right to me, though gcc might produce a (bogus) warning about a > potentially uninitialised use of uha on the kmem_free() line. > (uh was always set, so does not suffer from that problem.)
yeah - it might trigger at -O3, like dozens of other lines of code in our tree (hundreds of dozens?) i used to build my kernels with -O3, but it gives so many false positives like this i gave up. years and years ago, and last i tried, it was no better. it does compile in GENERIC in -current. i'll try to commit soon, i would rather find a tester, but i agree that it's clearly and obviously broken as-is. .mrg.
