On Thu, Jun 04, 2020 at 10:23:06AM -0300, Jason Gunthorpe wrote:

On Wed, Jun 03, 2020 at 04:32:02PM -0700, Kees Cook wrote:

quoted

Using uninitialized_var() is dangerous as it papers over real bugs[1]
(or can in the future), and suppresses unrelated compiler warnings
(e.g. "unused variable"). If the compiler thinks it is uninitialized,
either simply initialize the variable or make compiler changes.

I preparation for removing[2] the[3] macro[4], remove all remaining
needless uses with the following script:

git grep '\buninitialized_var\b' | cut -d: -f1 | sort -u | \
	xargs perl -pi -e \
		's/\buninitialized_var\(([^\)]+)\)/\1/g;
		 s:\s*/\* (GCC be quiet|to make compiler happy) \*/$::g;'

drivers/video/fbdev/riva/riva_hw.c was manually tweaked to avoid
pathological white-space.

No outstanding warnings were found building allmodconfig with GCC 9.3.0
for x86_64, i386, arm64, arm, powerpc, powerpc64le, s390x, mips, sparc64,
alpha, and m68k.

At least in the infiniband part I'm confident that old gcc versions
will print warnings after this patch.

As the warnings are wrong, do we care? Should old gcc maybe just -Wno-
the warning?

I *think* a lot of those are from -Wmaybe-uninitialized, but Linus just
turned that off unconditionally in v5.7:
78a5255ffb6a ("Stop the ad-hoc games with -Wno-maybe-initialized")

I'll try to double-check with some older gcc versions. My compiler
collection is mostly single-axis: lots of arches, not lots of versions. ;)

Otherwise the IB bits look ok to me

Acked-by: Jason Gunthorpe <redacted>

Thanks!

-- 
Kees Cook