------- Comment #6 from sezeroz at gmail dot com 2009-10-19 11:11 ------- (In reply to comment #5) > function is left, so chances are you are refering to a variable later on even > after it went out of scope.
By a closer look, the function is called twice, first by its argument set to true in which case it assigns the pointer, then with the argument set as false where the code seems to assume that the pointer is still valid. So, you are right, the code is buggy and I apologize for the noise. I am amazed, however, how the x86 compilations and/or optimized compilations worked flawlessly since ten years of time... Any hints? -- sezeroz at gmail dot com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|WAITING |RESOLVED Resolution| |INVALID http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41751