Re: Link Errors -- Relocation Overflow
Re: Link Errors -- Relocation Overflow
- Subject: Re: Link Errors -- Relocation Overflow
- From: Bob Clark <email@hidden>
- Date: Sun, 17 Jul 2005 12:05:44 -0700
Since it's been awhile, my original message is at:
<http://lists.apple.com/archives/Darwin-dev/2005/Jun/msg00120.html>.
<http://lists.apple.com/archives/darwin-development/2001/May/
msg00620.html> suggests that data sizes bigger than 32MB won't work;
hmmm. None of the individual libraries are larger than 32MB, but
cumulatively they probably are. But if that's related, then why is the
link failing for (as far as I can tell) all static libraries being
linked? I don't think (although I could be wrong) that this is the
issue.
This looks like a good avenue for attack. Run size on everything,
then try
linking with fewer libraries until you get unresolved symbol errors
instead.
Then see what the total sizes are for the set that breaks it.
Thanks for the suggestion, Peter.
(Sorry about the delay -- this is a back-burner project and there was a
lot of monkey-work that I did to
try to attack from this perspective.)
The punchline: this didn't seem to work.
As soon as I pull a single static library I get unresolved symbol
errors. (If I attack from the opposite direction and start out with
zero static libraries and just add the static libraries suggested by
the unresolved symbol errors, then as soon as I reach zero unresolved
symbol errors, the "relocation overflow" errors pop up.)
I even threw out the static libraries altogether and started adding the
original source files to my project, as needed to fix the
unresolved-symbol errors. Again, just as I dropped the last source file
to my project and the link was able to resolve all the symbols, the
mysterious (and infuriating) "relocation overflow" errors all popped
up.
Furthermore, I ran size on all the libraries -- the sum of all the TEXT
segments was well under the 32MB barrier; in fact, the sum is just over
4 MB. (The file size of the static libraries also sums to around 4 MB.)
(There were no other wonky-looking segment sizes either.)
So: it looks to me like it's a different link "phase" that's causing
the "relocation overflow" errors, although not knowing the internal
workings of the linker I'm just speculating.
Any further suggestions will fall on receptive and appreciative ears.
:-)
--Bob
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden