This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: test case bin3 failed in my x86_64 box


Frank Ch. Eigler wrote:

If you meant that as an additional question, then the answers to my
questions probably imply a reasonable answer to yours.

If you meant that rhetorically (to answer my question about endianness),
I would expect that a program that processes binary traces of the realm
of complexity that we are contemplating generating (varying records types
and lengths) would either
- require metadata to describe records, which then might as well include
 an identification of native endianness
- be smart enough to deduce it, perhaps based on a single magic number
 similar to the ELF header word
- be sadly inflexible in these ways, in which case the trace data is likely
 to contain trivial fixed-sized bits, which would be produced from a small
 number of script probes/functions, which then might as well hard-code
 endianness in the formatting directives.

I know, "not necessarily" is a come-back to all of the above, but I
think the generalities hold.

- FChE


It's just a question? I probably should say "data captured on one machine and analyzed on another machine would be the case when endianess becomes an issue".


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