Re^5: getlogin works on debian but not ubuntu?

Works for me even on Ubuntu PP:

mb@dennis ~ $ perl -E'say getlogin'
mb
mb@dennis ~ $ cat gl.c
#include <unistd.h>
#include <stdio.h>

int main(void) {
   puts(getlogin());
   return 0;
}
mb@dennis ~ $ gcc gl.c
mb@dennis ~ $ ./a.out
mb
[download]

Comment on Re^5: getlogin works on debian but not ubuntu? Download Code

Replies are listed 'Best First'.
Re^6: getlogin works on debian but not ubuntu? by bcrowell2 (Friar) on Jan 07, 2013 at 04:23 UTC
Looks like an OS or library bug. `$ cat >gl.c #include <unistd.h> #include <stdio.h> int main(void) { puts(getlogin()); return 0; } ---- rintintin ~ $ gcc gl.c ---- rintintin ~ $ ./a.out Segmentation fault (core dumped)` [download] Maybe it's the same thing as this? https://bugs.launchpad.net/ubuntu/+source/openssh/+bug/1067779	[reply] [d/l]
Re^7: getlogin works on debian but not ubuntu? by mbethke (Hermit) on Jan 07, 2013 at 16:45 UTC
Maybe it's the same thing as this? https://bugs.launchpad.net/ubuntu/+source/openssh/+bug/1067779 Strange. My netbook has the very same distribution, my /proc/self/loginuid contains UINT_MAX, but stuff still works. strace tells me that instead of reading /proc/self/loginuid (which on some Linuxes such as my Gentoo doesn't even exist) it checks the /proc/self/fd/0 link to find the terminal the program is running on and then checks /var/run/utmpx or /var/run/utmp for the user logged in on that terminal. The weird thing is that there is no trace of getutent() or something in my glibc's getlogin() implementation. I haven't found whether it gets vendor-patched in somewhere or why else that might be.	[reply]