Yeah that was the one piece I was thinking of, Ed. (You'll recall I broke one of your scenarios when I did that very thing - expecting 1+ section headers before I'd return a proper architecture for the elf).
In Matthew's first patch, all it did was say "hey if you do have section headers already parsed, we know we've done the note parsing too and short circuit the bits we've already run. I think the only time we'd get in trouble with this is if we're somehow building and extending an ObjectFileELF in stages and really do need to parse them multiple times. I was about to give his second patch a look now. When you looked at it, you were looking at his first one, is that correct? Any thoughts on the second one? -Todd On Fri, Jul 11, 2014 at 7:24 AM, Ed Maste <[email protected]> wrote: > On 10 July 2014 21:48, Todd Fiala <[email protected]> wrote: > > > > Matthew, > > > > Your patch looks good to me. I would like Greg and/or Ed to have a look > at it too. If they don’t object, I can submit it. > > > > I ran tests on Linux and MacOSX with the patch applied > > The change LGTM and it passes on FreeBSD. > > One little point to keep in mind is that there are ELF files that have > no section headers. We should avoid (further) conflating > m_section_headers.empty() with having not yet parsed the header. > -- Todd Fiala | Software Engineer | [email protected] | 650-943-3180
_______________________________________________ lldb-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
