Re: bug report...

> There's this new _IO_new_fclose() again. > What platform is this? My guess is that it is a Linux box with a broken glibc-2.1.1 library (ie. RH6). Stig and I chased this one down a while ago. Well, I did the chasing, Stig did the fixing.. ;) As far as I recall, it is a weird one. It only happens when PHP is loaded as a DSO. What happens is that when you pass a pointer to a libc function (as we do when we register fclose() as the destructor for open files) and then try to call this function from that pointer, something in libc trashes the stack. Our workaround was to change this: GLOBAL(le_fp) = register_list_destructors(fclose,NULL); to this: static void php3i_destructor_fclose(FILE *fp) { (void)fclose(fp); } GLOBAL(le_fp) = register_list_destructors(php3i_destructor_fclose,NULL); Looks weird, I know, but it actually fixed it. I don't see it in the ChangeLog, but we did this in early June so it went into either 3.0.9 or 3.0.10. My guess is that this guy is using an earlier version. -Rasmus