GSoC 2015

[Dmitry V. Levin]

On Sun, Mar 22, 2015 at 09:10:34AM +0530, Prateek Gupta wrote:
> So, is the idea to remove all these code redundancies and provide a
> structured way to support multiarchitecture in a new way or to extend the
> current implementation?
> 
> Regards,
> Prateek

Do you really think I'd propose extending the current implementation
(which I labelled "inherently unreliable") as a GSoC project?

P.S.  Top posting was not a good idea.

> On Sun, Mar 22, 2015 at 5:11 AM, Dmitry V. Levin wrote:
> 
> > On Sun, Mar 22, 2015 at 12:34:55AM +0530, Prateek Gupta wrote:
> > > As far as I percieve, the multiarchitecture support project is about
> > making
> > > a hierarchial set of the parsers to firstly compiling all general parser
> > > required by every arch and then compiling only those that are required by
> > > the arch that the user's system has.
> >
> > No.  Have a look at the code.  For example, "struct old_sigaction32" and
> > "struct new_sigaction32" in signal.c, there is a lot of code duplication
> > of this kind all over strace code, and some structures are not yet
> > duplicated thus leading to incorrect multiarch decoding.  I think we
> > cannot duplicate all structure definitions, thus I call the current
> > multiarch support inherently unreliable.
> >
> > The idea of this project is to implement correct multiarch decoding
> > in a way that would not require manual source code duplication.


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20150324/aed8d30c/attachment.bin>