LWN: Comments on "Irked by NO_IRQ"

Irked by NO_IRQ

Tov — Thu, 15 Dec 2011 17:03:34 +0000

Wouldn't they be "other positives" instead of just "false positives"?
In what type of robust code is (var < 0) considered a valid comparison on an unsigned value?

Irked by NO_IRQ

clugstj — Thu, 08 Dec 2011 18:32:04 +0000

And then you'd have to sift through the 5 bazillion false positives.

Irked by NO_IRQ

nteon — Thu, 08 Dec 2011 16:02:24 +0000

Wouldn't changing all the places irq's are stored to unsigned allow gcc's -Wsign-compare to point out where (irq < 0) happens? It seems like this would be fairly straightforward for structures.

Irked by NO_IRQ

xav — Thu, 08 Dec 2011 15:43:47 +0000

That semantic search-and-replace looks like a perfect job for Coccinelle[1] to me.

[1]: https://lwn.net/Articles/315686/