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: [RFC][PATCH 0/4][kprobe](djprobe) Direct jump optimized kprobes


Frank Ch. Eigler wrote:
> mingo wrote:
> 
>> [...]  I'm wondering whether it could be made a 100% transparent
>> speedup to kprobes: how hard would it be to do a simplified
>> disassembly of the target address to automate the 'this kprobe can
>> safely be turned into a djprobe transparently' step [...]
> 
> The entire criterion is not easy to check at the binary point.  In
> particular, it is hard to tell whether some part of the overlaid
> instruction sequence is the possible target of a branch elsewhere.

Yes. This is the problem. We couldn't find anything good way to
ensure the safety of branch target without debuginfo.

Now we are developing safety check tool that bases on
elfutils.

And during this safety check process, userspace tool must count
the length of replaced instructions. So we chose the length as
a trigger to enable djprobes. It is not the problem to count 
the length of instructions inside kernel. 

If you have any suggestion on this, we appreciate them.


Satoshi


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