On Wed, 11 Sep 2019, Daniel Kahn Gillmor wrote:
But hm, maybe the 4.19.0-5 ABI wasn't actually stable?
Thinking about this a bit more ... the issue wasn't a symvers mismatch,
which is what I'd have expected if the ABI was not actually stable, and
in my case the module headers exactly matched the running kernel
anyways.
The error was about a bad relocation. IIRC, the bad relocation would
imply that the module's own compilation said "edit <address> _in the
module_ with <change>", but the place it was supposed to edit wasn't all
zeros.
Which sounds to me like the compiled module binary being inconsistent
with itself, not with the running kernel?
My understanding of relocations and how they work with kernel modules as
opposed to normal userspace binaries is ... not strong, so I may be
barking up the wrong tree here.
--
-- Matt
"Reality is that which, when you stop believing in it, doesn't go away".
-- Philip K. Dick
GPG fingerprint: 0061 15DF D282 D4A9 57CE 77C5 16AF 1460 4A3C C4E9
