[linux-next:master] BUILD REGRESSION 6dc544b66971c7f9909ff038b62149105272d26a

[David Sterba]

On Wed, May 29, 2024 at 02:19:47AM +0800, kernel test robot wrote:
> tree/branch: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> branch HEAD: 6dc544b66971c7f9909ff038b62149105272d26a  Add linux-next specific files for 20240528
> 
> Error/Warning reports:
> 
> https://lore.kernel.org/oe-kbuild-all/202405282036.maEDO54Q-lkp@intel.com
> https://lore.kernel.org/oe-kbuild-all/202405282148.jaF0FLhu-lkp@intel.com
> https://lore.kernel.org/oe-kbuild-all/202405282308.UEzt6hqC-lkp@intel.com
> 
> Error/Warning: (recently discovered and may have been fixed)
> 
> drivers/dma-buf/udmabuf.c:45:(.text+0x140): undefined reference to `vmf_insert_pfn'
> fs/btrfs/fiemap.c:822:26: warning: 'last_extent_end' may be used uninitialized [-Wmaybe-uninitialized]

The report says it's gcc 13.2, that one I use (and expect others as well
as it's a recent one) and we also have -Wmaybe-uninitialized enabled in
fs/btrfs/ to catch such warnings. Yet this is reported on mips64, is
there something special about that compiler+architecture?

The warning is IMO a false positive, the maybe-uninitialized variable is
passed as pointer but initialized on success and never used on failure.
We can safely silence the warning by initializing the variable to 0 but
this may be pointing to a problem with mips64+gcc namely because other
compiler+host combinations do not warn abou that.