Messages in this thread | | | Subject | Re: [PATCH] mm: thp: avoid uninitialized variable use | From | "Yang Shi" <> | Date | Sun, 17 Dec 2017 06:25:17 +0800 |
| |
On 12/16/17 4:24 AM, Arnd Bergmann wrote: > On Fri, Dec 15, 2017 at 7:01 PM, Yang Shi <yang.s@alibaba-inc.com> wrote: >> >> >> On 12/15/17 4:51 AM, Arnd Bergmann wrote: >>> >>> When the down_read_trylock() fails, 'vma' has not been initialized >>> yet, which gcc now warns about: >>> >>> mm/khugepaged.c: In function 'khugepaged': >>> mm/khugepaged.c:1659:25: error: 'vma' may be used uninitialized in this >>> function [-Werror=maybe-uninitialized] >> >> >> Arnd, >> >> Thanks for catching this. I'm wondering why my test didn't catch it. It >> might be because my gcc is old. I'm using gcc 4.8.5 on centos 7. > > Correct, gcc-4.8 and earlier have too many false-positive warnings with > -Wmaybe-uninitialized, so we turn it off on those versions. 4.9 is much > better here, but I'd recommend using gcc-6 or gcc-7 when you upgrade, > they have a much better set of default warnings besides producing better > binary code.
Thanks, I just upgraded gcc to 6.4 on my cetnos 7 machine. But, I ran into a build error with 4.15-rc3 kernel, but 4.14 is fine. I bisected to a commit in Makefile. I will email my bug report to the mailing list.
Regards, Yang
> > See http://git.infradead.org/users/segher/buildall.git for a simple way > to build toolchains suitable for building kernels in varying architectures > and versions. > > Arnd >
| |