Several lines in the kernel PKGBUILD capture the checksums of /lib/modules/$pkgver/modules.symbols. My assumption has been that if the checksums of this file change from kernel version foo.1 to foo.2, that modules built against foo.1 would not work under foo.2 and would be used by the maintainers to trigger a rebuilt against foo.2. Conversely, if the checksums stay the same between version foo.1 and foo.2 then modules built against foo.1 would work just fine running under foo.2 thus not requiring a rebuild against foo.2. Is my understanding of the utility of checksumming /lib/modules/$pkgver/modules.symbols correct? If not, what are these sums used for?
# module.symbols md5sums # x86_64 # 2fd43e3edc671c61e043a5c0b3b2a1f0 /lib/modules/3.12.0-1-ARCH/modules.symbols # i686 # e98940249665dbfa380cfdbbacf6c6b8 /lib/modules/3.12.0-1-ARCH/modules.symbols
Last edited by graysky (2013-11-06 19:51:32)
Did you trying searching the web for this? If so, did you turn up the same sort of positively insane forum threads I found? I don't think I've ever had a single search turn up such a high proportion of weirdness.
Mostly interactions between an extraordinarily paranoid Ubuntu user and other non-paranoid Ubuntu users about LUKS being co-opted in some way by a remote attacker when the OP tried to read the logs (which showed the machine receiving IP addresses which the user believed to be part of an attack by strangely named users - actually LAN addresses requested by avahi etc. when the network interface went down). What was striking was the way in which every aspect of the computer's operation was reinterpreted as evidence of a hack - from a refusal to copy to a non-existent destination all the way through to boot failures. No explanations of any sort budged the OP one iota. There were some less colourful interactions of the same sort but slightly less extreme. I guess there's a security tool you can install which generates a bunch of false positives including the file name modules.symbols.
I didn't find anything relevant either. I just was somewhat surprised by the proportion of paranoia the search turned up. Sorry, I don't mean to take your thread OT. Especially since I now am curious about the answer!
Last edited by cfr (2013-11-06 00:28:10)
In our PKGBUILDs it is used as an indicator of modules needing rebuilt.
This is correct most times this works as indicator for external module rebuilds.
Thanks for the replies, guys. I second guessed its purpose when some users of my unofficial repo reported that some modules failed to insert on a minor kernel version bump whose md5sums did not change. I guess we can conclude that it is a good but not absolute method to trigger the rebuild.