Chris Down writes:
That article you reference states the opposite of what you said. And I got
burnt by it before. Because Linus stated, if it is available for users, it
is an ABI.
Hmm, even in 2011 after that article there were discussions about
debugfs explicitly being the "wild west"[0], no? I heard the same
during LSFMM discussions during recent years as well. Although I
confess that I am not frequently in discussions about debugfs so I
don't really know where the majority opinion is nowadays.
Either way, as discussed the contents wouldn't be the ABI (as with my
/proc/self/smaps allusion), the file format would be, so it wouldn't
imply that printk() calls themselves or their locations become an ABI.
0: https://lwn.net/Articles/429321/
(To be clear, that article basically says that debugfs should become less of a
"wild west", of course. I mostly just am not sure that it's so clear to
everyone what the ABI status of debugfs currently is, although probably your
recent experience with Linus is a reasonable thermometer on the whole thing.)
