This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: Masked bitfield comparisons may yield incorrect results



Sorry for not getting back to this sooner.  Since nobody was
responding to my mail, it fell off my queue.

With the 3 patches I sent, all problems I'm aware of in that code path
have been fixed.


Regarding the DHCP problem, that was a server-side issue.  It had to
do with the third case, where you're comparing fields that are
adjacent in both structures but in different positions, and that
causes lnbitsize and rnbitsize to be different.  The symptom was that
a comparison always failed in dhcprequest() and the server never sent
ACKs.

(Aside: That in turn caused a misconfiguration on another DHCP server
on the same subnet to rear its ugly head, and the client I was trying
to boot got in a REQUEST/NAK war with the other server, flooding the
network and causing the router to become overloaded, which yielded to
some stupid politics.  This caused me coin the term `Rube Goldbug'.)



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]