Re: [CIL users] Is it possible to use comments to ``direct'' a CIL transformation?

2011-11-24 Thread Pascal Cuoq
On Nov 24, 2011, at 10:49 AM, Gabriel Kerneis wrote: > On Thu, Nov 24, 2011 at 10:38:34AM +0100, Dimiter Milushev wrote: > >> I imagine one way is to possibly use CIL on a file that is not >> preprocessed? >> The reason I am asking is that I need a transformation that will keep >> some structur

Re: [CIL users] Patch to consider Addrof/StartOf as a definition

2011-01-26 Thread Pascal Cuoq
On Jan 26, 2011, at 8:24 AM, Piramanayagam Arumuga Nainar wrote: > I am doing a "definitely-uninitialized" analysis (i.e. no potential > initialization should be missed). In this case, I fear the Program Dependency Graph that the use-defs I illustrated are a component of will not be so useful to y

[CIL users] Misleading example in CIL documentation at http://www.eecs.berkeley.edu/~necula/cil/

2010-09-29 Thread Pascal Cuoq
Hello, I just noticed a bug in http://www.eecs.berkeley.edu/~necula/cil/cil016.html#toc32 : In the second example, (1 - sizeof(int)) >> 32) should be for instance (1 - sizeof(int)) >> 16) >> 16, or perhaps just (1 - sizeof(int)) >> 31). The current example is undefined behavior, and will

Re: [CIL users] missing implicit casts

2010-06-22 Thread Pascal Cuoq
Hello, On Jun 22, 2010, at 11:13 AM, Marco Trudel wrote: > void foo(int i, char c) { > foo('.', 2); > i = c; > c = i; > } > > becomes: > > void foo(int i, char c) { > foo('.', (char)2); > i = (int)c; > c = (char)i; > } > > Where I guess the call to foo should be "foo((int)'.',