PERFORCE change 55970 for review
David Xu
davidxu at freebsd.org
Mon Jun 28 05:20:28 PDT 2004
OK, I am glad to see you are working on libthread_db, but
I have already a libthread_db tree in ksedbg branch.
I am current making libthread_db.so in two levels:
the first level is a umbrella, the second level is a driver,
every thread library will have a driver, I found you were trying
to mix three thread libraries code into in same functions,
things like following code looks strange for me, how can you
include three different thr_pirvate.h in same .c, and compile
them ?
+td_ta_map_lwp2thr(const td_thragent_t *ta, lwpid_t lwp, td_thrhandle_t *th)
> {
> - return (TD_OK);
> + ps_err_e err;
> +
> + th->th_ta = ta;
> +
> + switch (ta->ta_lib) {
> + case PTHREAD_LIBC_R:
> + err = ps_pread(ta->ta_ph, ta->ta.libc_r.thread_run,
> + &th->th_thread, sizeof(psaddr_t));
> + return (err == PS_OK) ? TD_OK : TD_ERR;
> + case PTHREAD_LIBKSE:
> + lwp = 0;
> + return (TD_NOTHR);
> + case PTHREAD_LIBTHR:
> + return (TD_NOTHR);
> + }
> + return (TD_ERR);
> }
Marcel Moolenaar wrote:
> http://perforce.freebsd.org/chv.cgi?CH=55970
>
> Change 55970 by marcel at marcel_nfs on 2004/06/28 04:12:48
>
> Submit the result of me fucking around a bit:
> o Add td_ta_delete()
> o Make this WARNS 4 clean. This basicly means adding __unused
> to most of the argument of the functions I haven't implemented
> yet.
> o Detect libc_r and libthr. I don't worry about libkse, because
> davidxu@ is playing with that. We probably need to combine our
> efforts...
> o Implement some of the functions that GDB already needs. In
> particular I see td_ta_thr_iter() play a pivoting role...
>
> Affected files ...
More information about the p4-projects
mailing list