This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB 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: [PATCH 0/9] Add support for DTrace USDT probes to gdb


Hi Sergio.

    > This patch series introduces support in GDB for a new type of probe:
    > DTrace USDT probes.
    >
    > The first three patches do some changes to the existing probe.[ch]
    > code, fixing some minor problems associated to support several probe
    > types, having several probes of different types defined in the same
    > object and supporting the notion of enabling and disabling probes.
    >
    > The rest of the patches are the implementation of the new probe type,
    > including target support for x86_64 targets, a testsuite and
    > documentation.
    
    OK, I think I reviewed all the patches (except the docs).  First of all,
    thanks for doing this!

Heh, thanks for the throughout review!
    
    I have mixed feelings about the inclusion of this feature.  While I
    obviously agree that support for a new probe type is good, I also don't
    like the fact that we are talking about a feature that relies on
    proprietary software (Oracle dtrace) to operate.

As you know I am sympathetic with your concerns.  Was I in your position
I would raise exactly the same point.

    You mentioned that you were going to try to come up with some way to
    test this feature using a tweaked asm source.  I think this is a good
    progress already, and I am looking forward to seeing this.

Yes, I am working in a .s file that, once compiled and linked, will
provide the DOF program corresponding to the testcase.  That will not
replace dtrace -G for the general usage, and will have to be updated
should the testcase change, but at least will allow to test the probe
support without having to install proprietary software.
    
    Either way, if we decide to include this feature on GDB, you (or someone
    else from Oracle) will be responsible for it, because we outside Oracle
    have no way to make sure that it doesn't break.

I can certainly take responsibility on it.
    
    Having said that, I will wait until you post updated patches with our
    comments addressed (and hopefully with a nice testcase :-P), and then we
    can continue this discussion :-).

I will post a corrected version of the patch series soon.
Thanks again!


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