Finding and understanding bugs in C compilers [pdf] (2011)

原始链接: https://www.cse.unr.edu/~fredh/class/460/S2013/class/Papers/tanna.pdf

%PDF-1.5 % 484 0 obj > endobj 485 0 obj ] /Prev 635255 /W [1 4 1] /Length 400>> stream xKHQЙҹkeo0q(y!ڹQ0+Z c$j*(-t7\hZ"H,pEb#D!-y5 fa9Wx&oq\7XD`'naZ~n 'cN:"|[f$@DMC$Q1FrZ74?YiX縀2T}Ckd! a;.b,ia&Y ڦI el^>-W' /ʍ]!KG~̭nnK#WbJ>?)_Ur-pEN?7sNSpdyˈzsT|sy;$븣}k#JF6J>Xv(c endstream endobj 486 0 obj > stream

The Hacker News post discusses a 2011 paper on finding C compiler bugs using randomly generated C code. The core challenge is ensuring the generated code adheres to C's defined behavior to avoid false positives. A commenter, pfdietz, identifies the paper as related to Csmith and highlights its importance in inspiring further work. He notes that differential testing of compilers predates the paper, but C's prevalence of undefined behavior makes generating valid and diverse test cases difficult. Pfdietz also references a 2012 paper on creduce, a tool for minimizing failing test inputs while preserving valid C syntax, which is useful in conjunction with Csmith and other languages. Another user, anitil, expresses intrigue with the complexity of compiler bug hunting.
相关文章

原文
%PDF-1.5 % 484 0 obj > endobj 485 0 obj ] /Prev 635255 /W [1 4 1] /Length 400>> stream xKHQЙҹkeo0q(y!ڹQ0+Z c$j*(-t7\hZ"H,pEb#D!-y5 fa9Wx&oq\7XD`'naZ~n 'cN:"|[f$@DMC$Q1FrZ74?YiX縀2T}Ckd! a;.b,ia&Y ڦI el^>-W' /ʍ]!KG~̭nnK#WbJ>?)_Ur-pEN?7sNSpdyˈzsT|sy;$븣}k#JF6J>Xv(c endstream endobj 486 0 obj > stream
联系我们 contact @ memedata.com