On Fri, Feb 07, 2025 at 07:39:03AM +0100, Markus Elfring wrote: > > I don't *really* like guard() anyway because it's kind of magic in > > that the unlock doesn't actually appear in the code, and it's kind of > > hard to unravel what guard() is and how it works. But I guess that's > > mostly because it's just a new idiom that takes time to internalize. > > How will the circumstances evolve further for growing applications of > scope-based resource management? I'm sure it will evolve to become the typical style. Right now, it's not quite there yet, as evidenced by the fact that the only reference to them in Documentation/ is this somewhat ambivalent note: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/maintainer-netdev.rst?id=v6.13#n380 We do already have a few uses of guard() and scoped_guard() in drivers/pci, and I don't really object to more, including in this amd-mdb case. Whatever we do, I *would* want to do it consistently throughout the file. Bjorn
