Hello Jonathan, On Sat, Jan 25, 2025 at 02:48:28PM +0000, Jonathan Cameron wrote: > Fully agree the change makes sense. Though the floating point example > is unlikely to bite us in kernel! > > Did you see an actual failure to match? If not I'd be tempted > to take this for next cycle rather than as a fix to rush in. > I'd be surprised if current code doesn't happen to work. No, I didn't see a failure, I noticed that while working on the driver source. Also in this case nothing severe happens. The check might only result in failure to detect that a setup could be reused and uses a different one then. In the worst case that increases pressure to rotate the setup configuration in a timely manner. David also pointed out in a private conversion that the bool issue probably isn't relevant because the compiler (at least gcc, didn't check clang) only uses values 0 and 1 and the padding ideally should be all zero. Anyhow, I think comparing the actual byte representation is a bad pattern that shouldn't be used even if in that case it might work. So I'm not aware of a problem that would justify getting this in early as a fix and the next cycle it totally fine. Best regards Uwe
Attachment:
signature.asc
Description: PGP signature
