|
From: Yeshurun, M. <mei...@in...> - 2005-08-17 11:51:41
|
I could be wrong of course. Perhaps you could try to make Valgrind generate an error for printf by passing weird arguments, and then check in the stack trace if printf is in a.out or in a so. If it's in a so, it would definitely rule out what I suggested. Regards, Meir -----Original Message----- From: Rob Holland [mailto:ti...@ge...]=20 Sent: Wednesday, August 17, 2005 2:44 PM To: Yeshurun, Meir Cc: valgrind-users Subject: RE: [Valgrind-users] Replacing printf On Wed, 2005-08-17 at 14:38 +0300, Yeshurun, Meir wrote: > It's possible that printf is a small inline function or a template > function (which then turns around and calls fprintf) that is compiled > into your program, not one of the so's. You need to find a way to leave > the code for printf out of your program, so your program will have to > call the code for printf that is in one of the so's.=20 Hmm. nm a.out reports: U printf@@GLIBC_2.2.5 Doesn't that mean it will trying to find it from a shared library? If I use fprintf, I get U fprintf@@GLIBC_2.2.5 Which implies that they aren't different in how they are resolved. Cheers, Rob --=20 |