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: RFC: lazily call save_current_space_and_thread


>>>>> "Pedro" == Pedro Alves <pedro@codesourcery.com> writes:

Pedro> Okay.  In any case, what I was aluding to, is that
Pedro> in non-stop mode gdb is making sure the user selected
Pedro> frame is constant between events (fetch_inferior_event),
Pedro> so in that case, there's a selected frame to restore back
Pedro> to.  But thinking a bit more, that obviously doesn't matter
Pedro> because if the user had a frame selected, then the debug info
Pedro> for that frame has already been read anyway, and there's nothing
Pedro> to optimize.

Also, this code will still do the right thing if there actually is a
selected frame.

Moving the selected frame to be per-thread would also help here.
But I am going to guess that this is tricky, for the same reasons that
making the regcache per-thread is tricky.

Pedro> Error in re-setting breakpoint 1: Cannot access memory at address 0x45e22a
Pedro> Error in re-setting breakpoint 2: Cannot access memory at address 0x4e2e5a
Pedro> warning: Error removing breakpoint 1
Pedro> warning: Error removing breakpoint 2

I didn't run into this, but I have seen other weird things, like errors
from libpthread_db.  I haven't investigated yet.

Tom


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