[Bug libc/18396] New: timezone internal code for time_t_max and time_t_min is UB

["nszabolcs at gmail dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=18396

            Bug ID: 18396
           Summary: timezone internal code for time_t_max and time_t_min
                    is UB
           Product: glibc
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: libc
          Assignee: unassigned at sourceware dot org
          Reporter: nszabolcs at gmail dot com
                CC: drepper.fsp at gmail dot com
  Target Milestone: ---

recent gcc change exposed an invalid -1 << n in timezone/private.h
if time_t is signed:

/* The minimum and maximum finite time values.  */
static time_t const time_t_min =
  (TYPE_SIGNED(time_t)
   ? (time_t) -1 << (CHAR_BIT * sizeof (time_t) - 1)
   : 0);
static time_t const time_t_max =
  (TYPE_SIGNED(time_t)
   ? - (~ 0 < 0) - ((time_t) -1 << (CHAR_BIT * sizeof (time_t) - 1))
   : -1);

the gcc warning may get silenced as it caused other problems:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66066

but the UB should be fixed.

There is no simple solution: the above code seems to want to support
non-two's-complement signed int representations too.

I think a solution based on (uintmax_t)-1 >> n can work for time_t_max
with reasonable assumptions, otherwise arch specific soulution can be used.

-- 
You are receiving this mail because:
You are on the CC list for the bug.