On Wed, Apr 09 2025 at 08:49, James Bottomley wrote: > On Tue, 2025-04-08 at 22:46 +0200, Thomas Gleixner wrote: > [...] >> - retain_ptr(dev); >> msix_update_entries(dev, entries); >> + retain_ptr(dev); >> return 0; > > Heh, well, that scores -6 on the Rusty scale "the name tells you how > not to use it" because retain_ptr would commonly be read to mean the > pointer retains its value. It would be nice if KASAN had compile time > markers that would cause a use after free build failure for this, but, > apparently, it doesn't. The cleanup annotations don't let us remove > the scope free function, which would otherwise let us match the use to > the name, so I think the name might have to change to something like > retain_and_null_ptr(). Yeah. When I saw it I immediately rumaged for a brown paperbag :)
