Re: [RFA] Add interface for registering JITed code

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Reid" == Reid Kleckner <rnk@mit.edu> writes:

Tom> Yeah.  This would work if you had two JITs in a process, say loaded
Tom> dynamically, and the various __jit symbols had hidden visibility.

Reid> I think if someone has that use case, then they can write a client
Reid> side library that wraps the __jit symbols and make the two JITs call
Reid> that.

Yeah, I changed my mind, let's not worry about it.  If it ever comes up,
we can deal with it then without breaking compatibility.

Reid> +  /* Remember a mapping from entry_addr to objfile.  */
Reid> +  set_objfile_data (objfile, jit_objfile_data, (void*) entry_addr);

Tom> I don't think you need the cast here.  There are a few of these.

Reid> Actually, it is, because entry_addr is a CORE_ADDR.  CORE_ADDR fits
Reid> into a void* right?  I'm just doing it to avoid mallocing a single
Reid> CORE_ADDR.

Oh, I see.  It is better to just do the allocation; I think we try to
avoid casting between CORE_ADDR and void*.

Reid> +      case BPSTAT_WHAT_CHECK_JIT:
Reid> +        if (debug_infrun)
Reid> +          fprintf_unfiltered (gdb_stdlog, "infrun: BPSTAT_WHAT_CHECK_JIT\n");
Reid> +
Reid> +        /* Switch terminal for any messages produced by breakpoint_re_set.  */
Reid> +        target_terminal_ours_for_output ();

Is the comment here really valid?
I think it should probably mention jit_event_handler, not
breakpoint_re_set.

Reid> +//#include "assert.h"
Reid> +//#include "string.h"

Just delete these.
(And just FYI, gdb doesn't allow "//" comments.)

Reid> +/* Openning the file is a no-op.  */

Typo, should be "opening".

Reid> +  /* If we already found the symbols and successfully set the breakpoint, don't
Reid> +     do it again.  */
Reid> +  if (jit_descriptor_addr != 0)
Reid> +    return;

I'm sorry I didn't notice this before.

This approach won't work if you have address space randomization enabled
and you re-run the inferior.

Tom