| |||||||||||||
64-bit division on 32-bit arch - where's the compiler error?64-bit division on 32-bit arch - where's the compiler error?Posted Aug 30, 2011 3:34 UTC (Tue) by nybble41 (subscriber, #55106)In reply to: 64-bit division on 32-bit arch - where's the compiler error? by giraffedata Parent article: Quotes of the week
> ... we've been saying this is problem with 64 bit numbers on a 32 bit CPU. But __udivdi3() is the divide function for "unsigned long" operands, i.e. on a 32 bit machine it is a 32 bit divide. The compiler would use __udivti3() instead if it wanted to divide 64 bit numbers.
That's not quite correct. The preface to the runtime library section[1] of the GCC Internals manual states that the C types are "for illustrative purposes", and that "these routines take arguments and return values of a specific machine mode, not a specific C type." The Machine Modes section[2] defines SImode, DImode, and TImode as a four-byte integer, an eight-byte integer, and a 16-byte integer, respectively. In other words, in the hypothetical architecture invented to illustrate these library routines, the "int" in __udivsi3() is 32-bit, the "long" in __udivdi3() is 64-bit, and the "long long" in __udivti3() is 128-bit.
A simple test program confirms that 64-bit division generates a call to __udivdi3() in 32-bit mode. Similarly, division of unsigned __int128 values in 64-bit mode generates a call to __udivti3().
[1] http://gcc.gnu.org/onlinedocs/gccint/Libgcc.html
(Log in to post comments)
|
|||||||||||||