On Sun, 2026-02-08 at 18:38 +0530, Ridham Khurana via Gcc wrote:
> hello,
> my name is Ridham Khurana, currently in 4th year of Integrated M.Tech
> (CSE)
> at University of hyderabad. I am highly interested in system
> programming
> and compiler engineering. I am interested in the “extend the static
> analysis pass” project of gsoc 2026.

Hi Ridham; thanks for your interest in GCC, and in the static analysis
pass.  I would be the mentor of such a project.

Note that I don't know yet if GCC has yet been officially accepted by
GSoC as a mentoring project for 2026.

> 
> I already began the study of analyzer's internals starting from the
> links
> provided on the selected-projects page. i also previously made some
> PRs in
> the QEMU and RISC-V system. My experience in compilers is from
> building 2
> compilers from scratch, RMc4 (made in C), compiling a subset of C,
> and
> RMc7(C++17) , a big in-progress project targeting 90-95% ISO C
> coverage.
> Working on these 2 compilers helped me gain knowledge of compiler
> internals
> like ast, IR , semantic analysis ,  optimizaiton and low- level code
> generation. these projects and my contributions helped me to learn
> navigating large codebases and handling C/C++ environments.

This sounds great.

> 
> i saw there are 4 different idea under this project, so i wanted to
> ask
> like if there are any preference which idea should be implemented or
> worked
> on first or any suggestions regarding this.

My preference is the first one listed: format-string support.  This RFE
is bug 107017 in our bugzilla instance.  But if a candidate has a
strong interest in adding checking for a particular API (say POSIX-
related), I'll consider that too.

>  i would appreciate some
> guidance on how to continue this study of the architecture and any
> suggestions for me to focus on.

I would suggest building gcc from source and stepping through it on
some trivial C files, in particular, stepping through the analyzer on,
say, a trivial double-free example, and seeing the various the stages C
it goes through (all the way from frontend through to the analyzer, and
perhaps beyond to asm if you want a challenge!)

If you're interested in the format-string one, you might also want to
take a look at the two existing implementation of this within GCC (see
the bug I mentioned above).  We currently have two implementations of
format-string handling in GCC, and I'd much prefer that some kind of
refactoring out of the common logic happened so that they share code,
rather than just adding a 3rd implementation :-/

> 
> github link : https://github.com/rmkhurana28

Thanks; I enjoyed looking through your C compiler implementations.

Hope this is helpful
Dave

Reply via email to