review LADADL13

citations.bib

@@ -1,3 +1,15 @@
+@inproceedings{evolution,
+	author = {Lu, Lanyue and Andrea C. Arpaci-Dusseau and Remzi H. Arpaci-Dusseau and Lu, Shan},
+	title = {A Study of Linux File System Evolution},
+	booktitle = {Proceedings of the 11th Conference on File and Storage Technologies (FAST {\textquoteright}13)},
+	series = {FAST'13},
+	year = {2013},
+	month = {February},
+	address = {San Jose, California},
+	url = {http://research.cs.wisc.edu/adsl/Publications/fsstudy-fast13.pdf},
+}
+
+
 @inproceedings{quark,
 	author = {Jang, Dongseok and Tatlock, Zachary and Lerner, Sorin},
 	title = {Establishing browser security guarantees through formal shim verification},

related.tex

@@ -109,6 +109,12 @@ \subsection{Bug Fixing}
 
 
 \section{Other}
+\subsection{Bug Arch\ae{}ology}
+\begin{itemize}
+	\item \cite{evolution} is a study of linux file system evolution. They analysed over 5000 patches on six common linux filesystems: ext3, ext4, reiserfs, jfs, xfs, and btrfs, and classified them according to multiple metrics: bugfix/performance/enhancement/reliability, common-case/failure-path, type of data structure involved, etc.
+		Main observations include: 40\% of bugs were in failure-path code, semantic bugs (i.e. domain knowledge to fix) are over 50\% of bugs (with concurrency bugs in second place - 20\%), 50\% of all patches are maintenance, and B-trees are relatively unlikely to have bugs compared to other data structures.
+		The intended impact of the paper is not crystal-clear, though they claim it should influence tool- and language-designers and developers.
+\end{itemize}
 
 \section{Future Reading}
 \begin{itemize}