Re: Minimal symbols and gdb's two-stage symbol reading

[Orjan Friberg <orjan dot friberg at axis dot com>]

Andrew Cagney wrote:
> 
> Orjan Friberg wrote:
> >
> > Would it be safe to rely on an absolute global symbol being read into
> > the minimal symbol table with the initial 'file' command (without using
> > the '-readnow' option) for an a.out target? From looking at
> > read_dbx_symtab it wasn't clear to me what happens.
> >
> > As discussed in a previous thread, I will use the symbol as an ABI
> > indicator, and I'm just wondering if gdb's two-stage symbol reading
> > could interfere with my intentions (i.e., the symbol not being there yet
> > when I want it to).
> 
> If I'm understanding things correct then I think Kevin's suggestion is
> preferable to PeterSs.
> 
> The arch_init() function shouldn't be relying on global state.  Rather
> it should only be looking to its immediate parameters for information.
> 

That sounds reasonable; thanks. bfd_symbol_lookup() also has the added
benefit of letting me detect the presence of the symbol I'm looking for
by the time gdbarch_init() is called as a result of the 'file' command.
(Previously, I wouldn't find the symbol until gdbarch_init() got called
_after_ the 'file' command was completed, for example as a result of a
user command that I've added.)

-- 
Orjan Friberg              E-mail: orjan.friberg@axis.com
Axis Communications AB     Phone:  +46 46 272 17 68