Re: [patch] Support 64-bit constants/enums on 32-bit host [Re: [PATCH] Allow 64-bit enum values]

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Tue, 17 Apr 2012 17:09:23 +0200, Tom Tromey wrote:
> Due to the way that struct symbol is packed, I think this would only
> save sizeof(general_symbol_info) - sizeof(void*) bytes (32 bytes per
> symbol on x86-64).

Personally I still do not think we should care too much about
one 'struct symbol'.  There is more a problem we create
too many 'struct symbol' which are never used.


> This idea would make lazy CU expansion a bit faster, because you
> wouldn't have to re-scan the DIEs to make the symbol table.

I think this would be a win contrary to other disadvantages; I have sure no
technical arguments for that, other than accessing memory outside of the CPU
cache is terribly slow and several more bytes are worth the acceleration.


> and as you pointed out on irc, the result will
> still probably be slower than idb -- IOW, we're doing something really
> wrong, so why not start by finding that?

BTW without making some big decisions one should do some more serious test of
idb, or at least to be tested by a second person, I did just a quick test
myself (that GDB is 800%-times slower even with .gdb_index), I could make some
mistakes and I even no longer remember all the details how I specifically did
the test.


Thanks,
Jan