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: [RFA/commit/dwarf] Create partial symbols for nested subprograms


>>>>> "Joel" == Joel Brobecker <brobecker@adacore.com> writes:

Joel> but the one thing that I am wondering is how users will respond
Joel> to an unexpectedly long delay when entering a command that is
Joel> usually fast and yet takes a noticeably long time that one
Joel> time. For intance, imagine the following scenario:

Joel>   % gdb my_300MB_exe
Joel>   (gdb) break main
Joel>   [12 secs later...]
Joel>   Breakpoint 1, ...

Yeah, I wonder that too.

One thing to note is that, on average, we should expect the delay here
to be half of the startup delay that users experience today.  My
reasoning is that, on average, we should only have to load half the
psymtabs before finding a symbol.  (Of course, the worst case does
happen regularly, so this is not entirely satisfactory.)

I will do some more tests and fixing of this patch sometime soon.  If
this is a big problem perhaps I'll reopen the pubnames investigation.
Or maybe we can use the minimal symbols as a guide somehow, in some
common cases (or maybe we do already, I did not look at this in
depth).  (FWIW I'm also curious about lazily reading minimal symbols.)

Tom


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