First, non-html is enough in this mailing list.
Used correctly,
Ptranal.resolve_funptr
does what you say your are looking for. BTW, small alias sets
are your friends and this holds true for function pointers in
particular.
/Chris
Alex,
On Fri, Sep 20, 2013 at 09:12:50AM +0300, Alex Susu wrote:
> The partial modules is very interesting. Was it used for a publication?
> If so, could
> you please tell me some reference(s).
The partial feature was written by the
original CIL crew Necula, McPeak, Weimer. It
look
Gabriel -
On Wed, Sep 18, 2013 at 08:41:04AM +0100, Gabriel Kerneis wrote:
> Do you know why Golf is not enabled by default?
No, I don't. Could be the dreaded
backwards-compatability whatever, though.
I don't see any reason not to hard-code `Golf'.
It never produced any false positives
Alex -
On Tue, Sep 17, 2013 at 04:21:43PM +0300, Alex Susu wrote:
> Did anybody use Dead code elimination (DCE)?
We used it, but that dates back almost a
decade, when we analyzed several large projects
in embedded environments.
Let me summarize the problems I still remember.
(1) CI
On Wed, Aug 07, 2013 at 07:13:18AM +0100, Gabriel Kerneis wrote:
> No, the conversion basically calls Cil.constFold which fails on anything but
> ints and enums.
Once upon a long ago (~10y) we already
had this discussion with respect to the
`Partial' module.
Back then someone (probably mo
On Mon, May 21, 2012 at 05:56:01PM +0530, srikanth vaindam wrote:
> When I changed the line "module A = Olf" to "module A = Golf" in ptranal.ml I
> get link error saying
>
> Linking bytecode obj/x86_LINUX/cilly.byte.exe
> Error while linking obj/x86_LINUX/ptranal.cmo:
> Reference to undefined globa
Hi -
It looks like we are exactly at a border here. ;-)
My super-simplified code based on Dany's snipped is
struct foo {int bar;};
typedef struct foo foo_t;
foo_t foos[] = {{-1}, {0}, {1}};
foo_t more_foos[] = {{0}, {8}, {15}, {}};
Now, gcc(1) shrieks
$ gcc --version
Hello!
On Wed, May 04, 2011 at 02:29:32PM +0800, haihao shen wrote:
> I found CIL treats the following two functions differently.
> void loop_test1()
> while((get_a()<1) && (get_b()<1))
> void loop_test2()
> while(!(get_a()<1) && !(get_b()<1))
Your conditions that control the `w
On Tue, Apr 26, 2011 at 07:46:16PM -0400, Feng Zhu wrote:
> I am a new user of CIL and I try to use the pointer-analysis module.
I hope you're aware that you must change the
module A = Olf
line in "ptranal.ml" to
module A = Golf
and recompile CIL to actually use the
generalized
Hi Vijayaraghavan,
On Tue, Mar 29, 2011 at 08:31:55PM +0800, Vijayaraghavan Murali wrote:
> I'm using CIL's dopartial to play with constant propagation. It's very
> painful to look through manually in the output program to see if
> constant propagation has been applied somewhere, and where.
On Sat, Feb 05, 2011 at 06:59:06PM -0500, Dazhi Zhang wrote:
> the declaration of "static int a; " is not matched by any case. It can be
> caught if there is not "static" ahead.
I cannot confirm your finding. My Cil matches
static int n;
in the
Cil.GVar (variable, initialization,
Hello Olivier -
On Fri, Jul 16, 2010 at 05:50:17PM +0200, Olivier Crameri wrote:
> is there any documentation or paper somewhere that describes in more details
> the points-to analysis module of CIL ?
I'm afraid there is none; at least none
I'm aware of. However, you can refer to the
off
Hi!
On Mon, Jul 12, 2010 at 08:12:08PM +0300, Zakkak Foivos wrote:
> i need some help inserting some preprocessor directives between some
> generated (with Cil) stmts.
Some time ago, I used #pragma directives
to communicate extra information to and from a
cil run. Of course your cil-exte
Chuck -
I can reproduce your findings and I too
think Cil does not follow the C-standard here,
though IMHO the relevant part is not 7.15.1.1:2,
but 6.5.2.2:7. Also see: Harbison/Steele
Sec. 6.3.5.
On Thu, May 20, 2010 at 12:10:49PM -0500, Chucky Ellison wrote:
> The work that I am doing
On Tue, Apr 27, 2010 at 11:15:14AM +0200, Dany Vereertbrugghen wrote:
> Maybe it's because we're still on 1.3.6?
Maybe, it is because I'm running a
64-bit system? Sorry, but I forgot to mention
that in my initial post.
/Chris
-
Hi Dany!
On Tue, Apr 27, 2010 at 08:42:05AM +0200, Dany Vereertbrugghen wrote:
> We're currently running into a problem where CIL seems to transform
>
> (mainValue < (0xULL)/1000)
> to
> (mainValue < 0ULL)
Excuse me, but I cannot reproduce your problem
here. My "cil-c
Hello Cole -
> I am trying to use the Ptranal points-to analysis (with Golf) in CIL
> 1.3.7 to compute points-to sets containing dynamically allocated memory
> locations, but the points-to analysis seems to consider all pointers to
> any malloc'd memory as pointing to the same location (specifical
Steven -
On Thu, Mar 04, 2010 at 08:45:05PM +0100, Steven Van Acker wrote:
> I would like to use #pragma directives to let my cilly extension skip
> certain functions like so:
> #pragma myskip somefunc
> The problem I'm running in to, is that pragma attributes need to be
> formatted in a s
On Wed, Sep 09, 2009 at 10:38:14AM -0400, Stephen Magill wrote:
> I haven't tried this, but I suspect a pointer analysis could help you obtain
> a more general solution. Take a look at ptranal.ml in the cil src/ext/pta
> directory.
The default pointer analysis module is
Olf, because "ptra
John -
On Thu, Aug 27, 2009 at 04:59:45PM -0600, John Regehr wrote:
> >unsigned long a;
> >unsigned long *p;
> >p = (unsigned long *)&p;
> >a = ++(*p);
>
> Gabriel, since p refers to itself this code looks to me like an
> occurrence of the second clause of this type of undefined be
Hi Elnatan!
> CIL mistakenly casts the operands of && and || to intType, when in
> fact they are only required to have scalar type. This can lead to a
> bug if the operands have type larger than int and have values which
> truncate to 0 when cast to an int (and if useLogicalOperators is set
> to t
Hi Vu,
On Wed, Jul 08, 2009 at 12:39:22AM -0600, ThanhVu (Vu) Nguyen wrote:
> Hi, can I find out what kind of statement (e.g., a single statement, a
> block of statements, etc) from a given Cil.stmtkind?
To distinguish between the kinds of
statments (Instr, Return, Goto, etc.) you can
j
On Tue, May 26, 2009 at 10:45:30PM +0200, Gabriel Kerneis wrote:
> I consider switching to Tailor the day I get fed up to manually tracking it.
I have been using the following Tailor
configuration to mirror the CIL repository for
quite a while now. It has been working
flawlessly so far.
Gabriel -
On Tue, May 26, 2009 at 07:02:16PM +0200, Gabriel Kerneis wrote:
> I've been suffering performance issues with CIL recently. 30% of the
> time was spent in garbage collection.
A while ago a faced similar problems.
That time, I tried three different approaches to
speed up the ana
On Tue, Apr 21, 2009 at 12:57:07PM -0400, Divya Krishnan wrote:
> Basically if my input is:
> int c[10];
> I want my output to be
> volatile int c[10];
We may not like it, but for a C-compiler
the order of the type modifier and the type
identifier does not matter.
You want an array of vol
Benjamin -
On Wed, Mar 25, 2009 at 01:50:58PM -0700, Benjamin Ylvisaker wrote:
> 2) In C or bash or whatever, implement basically the same thing, but
> externally at the process level.
>
> Has anyone else dealt with this issue? Any ideas about what would be
> best?
There are several load
On Wed, Mar 11, 2009 at 01:30:45PM +, Samin Ishtiaq wrote:
> I run this function after cpp, and before I call Frontc.parse,
> to remove these annotations:
>
> --- snip ---
If you go for the "find-replace" approach the freaky directives could
be translated into gcc attributes ("__attribute__((.
27 matches
Mail list logo
