GSoC 2026 Inquiry: Non-recursive demangler for libiberty

 Dear GCC Mentors,
My name is Peeyush, and I am a student at Lovely Professional University. I
am writing to express my strong interest in the GSoC 2026 project: "Make
cp-demangle non-recursive."
I have spent the last few days setting up my development environment on
Windows 11 using MSYS2. I have successfully built libiberty and verified
the baseline by running the existing demangler tests (test-demangle.exe:
420 tests, 0 failures, also some error which when I googled said not to
worry about, please do refer to the attached image).
[image: image.png]
I am currently drafting my proposal and focusing on a design that uses an
explicit, heap-allocated stack (LIFO worklist) to replace the recursive
calls in
  d_print_comp. My goal is to address the long-standing stack overflow
vulnerabilities (like CVE-2018-18484) caused by deeply nested mangled names.
I have been studying the cp-demangle.c source code and the Itanium ABI
mangling grammar. If there are any specific technical constraints or
preferred
approaches I should consider while finalizing my proposal over the next 48
hours, I would greatly appreciate your guidance. I would like to apologize
for replying and starting late.

Thank you for your time and for this opportunity to contribute to the GCC
project.

Best regards,
Peeyush Maurya