No because it doesn't happen on any system I have useful access to and all
the traces show is someone previously (any time previously) stamped on
memory. You cn try enabling SKB debugging in include/linux/skbuff.h and
rebuilding from scratch to look for networking buffer errors, but so far
nobody has found any. I have no idea who or what is corrupting memory nor
found any common links in the reports, so it could be a bug anywhere in
the kernel or in a gcc version
[Reference: I build/run tests with gcc 2.5.8 building -m386 binaries for
a pentium, a 486 and a 386. I don't use gcc 2.6.x or 2.7.0 because I have
too many user programs these gcc versions compile wrong]. Thats not to say
building with gcc2.5.8 is the answer either.
Alan