50 lines
2.4 KiB
TeX
50 lines
2.4 KiB
TeX
% Created 2025-06-30 Mon 16:42
|
|
% Intended LaTeX compiler: pdflatex
|
|
\documentclass[11pt]{article}
|
|
\usepackage[utf8]{inputenc}
|
|
\usepackage[T1]{fontenc}
|
|
\usepackage{graphicx}
|
|
\usepackage{longtable}
|
|
\usepackage{wrapfig}
|
|
\usepackage{rotating}
|
|
\usepackage[normalem]{ulem}
|
|
\usepackage{amsmath}
|
|
\usepackage{amssymb}
|
|
\usepackage{capt-of}
|
|
\usepackage{hyperref}
|
|
\author{Akilan}
|
|
\date{\today}
|
|
\title{}
|
|
\hypersetup{
|
|
pdfauthor={Akilan},
|
|
pdftitle={},
|
|
pdfkeywords={},
|
|
pdfsubject={},
|
|
pdfcreator={Emacs 30.1 (Org mode 9.7.11)},
|
|
pdflang={English}}
|
|
\begin{document}
|
|
|
|
\tableofcontents
|
|
|
|
\section{Discussion with Adam:}
|
|
\label{sec:org9f7a2c6}
|
|
Recommendation to adding a few pages to the thesis mentioning that Security is not the goal of the current research.
|
|
\begin{itemize}
|
|
\item To mention aspects such as timing attacks was not evaluated.
|
|
\begin{itemize}
|
|
\item When storing bounds next to each other attacks such hammering DRAM to cause bitflips (Rowhammer attack) is a possibility.
|
|
\end{itemize}
|
|
\end{itemize}
|
|
|
|
We spoke in detail about the use case of Libc against Kernel modules. He asked me why contig malloc was not used to which I mentioned the underlying behavior of it.
|
|
|
|
Adam recommended for the evaluation section to have scatter plot show casing the numbers for multiple runs (This is a bit tricky in our case since we are doing a percentage difference).
|
|
\begin{itemize}
|
|
\item We spoke about the performance runs in detail. For instance what ARM counters are missing from the Morello implementation and how we used to the other performance counters to cover them.
|
|
\item We spoke about the wide use of Jemalloc but the lack of Academic sources to the Jemalloc implementation. It was good to know as someone working in performance side and Adam coming from the Security side we both faced similar rabbit holes when it came to dealing with finding academic algorithmic references to Jemalloc.
|
|
\item Adam understood my reasoning on why I could not make certain trivial assumptions about cache misses with Jemalloc with Contigious memory vs Standard Jemalloc. He recommended that the Thesis should have this (Appreciated that these aspects are considered).
|
|
\end{itemize}
|
|
|
|
In regards to the timetable he mentioned that it's a good decision sticking to the RISCV simulator and not focusing on the Uni-kernel experiment. He recommended that to keep slippage time until max February or March 2026 as a hard deadline for experiments (But motivated to finish them by December).
|
|
\end{document}
|