Hi Simon M.,

On Wednesday December 2, 2009, Simon Marlow wrote:
> On 01/12/2009 06:49, Howard B. Golden wrote:
> > Mon Nov 30 22:12:34 PST 2009  [email protected]
> >    * Fix GHC ticket 2615 (linker scripts in .so files)
> >    This patch only applies to systems that use ELF format files.
> >    The patch modifies the addDLL function so that it recognizes
> >    "invalid ELF header" errors. If these occur, the file that was
> > opened is scanned for a linker script GROUP ( ... ) directive. If
> > found, the first file inside the GROUP ( ... ) will be sent to
> > dlopen. Any errors reported by dlopen then will be reported  to the
> > caller.
> 
> Thanks Howard.  I think the patch could do with a round of tweaks
>  before being pushed, comments below.

I will make the changes you suggest. Then I'll follow the full 
validation process. I have done much of this already. I noticed that 
there is only 1 failure: T1969. I don't think I can do anything about 
that, so I plan to disregard it. Please confirm.

> Can we rely on the availability of regex.h and POSIX regexes?  Does
>  this need a configure test?

I don't know the answer to this. Isn't POSIX part of the baseline 
assumption? If not, I would need some help with the autoconf changes, 
since I haven't done any before.

> We don't generally use atexit() in the RTS, all the cleanup has to be
> done by hs_exit().  Typically each subsystem has an exitFoo() or
> freeFoo() function called by hs_exit().

I'm uncertain about what I should do here. Should I simply remove the 
atexit() and the callback code. Or should I somehow link it up to 
hs_exit()?

> I think you left some debug output in here.  Suggest protecting it
>  with IF_DEBUG(linker, ...).

I already discovered this when I ran the validation script.

I'm not sure how to write a test case that can be added to the 
validation tests. By _hand_ I have been testing by running:

$ ghc-stage2 --interactive -lpcre

On my system (Gentoo), /usr/lib64/libpcre.so is a linker script. 
However, I don't think this will always be true on other systems. I'd 
appreciate some suggestions about how to write the test.

> Also I suggest adding a comment referring to the ticket number, and
>  an brief description of the problem and solution.

I will add this.

Thanks for your thorough review! Now that I understand the QC process 
(validation scripts) I will try to use it as intended.

Regards,

Howard

_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc

Reply via email to