[Bug c/60846] New: Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[Bug c/60846] New: Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[jirik.svoboda at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

            Bug ID: 60846
           Summary: Add 128-bit integer types for general use on
                    32-bit/64-bit CPUs
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jirik.svoboda at seznam dot cz

(u)int64_t is available on all targets (32-bit and 64-bit) but __int128 isn't
(according to http://gcc.gnu.org/onlinedocs/gcc/_005f_005fint128.html __int128
is only available if long long is 128-bits wide which doesn't seem to be the
case for 32-bit/64-bit Intel targets).

Please add something like (u)int128_t, i.e. 128-bit signed and unsigned integer
types available on 32-bit and 64-bit compiler targets. This would allow, for
example, representing an IPv6 address as a single scalar value. Address masking
could be performed with a simple bitwise operator.

Since 32-bit and 64-bit CPUs typically cannot operate on 128-bit integers
directly, multiple instructions need to be generated for a single arithmetic
operation.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |WORKSFORME

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
It's already available.  The sentence you quote just says you can't write a
literal 128bit constant in source.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hp at gcc dot gnu.org

--- Comment #2 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
Reporter: do you mean just x86_64 / x86 targets or all 32- and 64-bit targets?
(The implied reply from the description is yes; I only ask since you mentioned
"32-bit/64-bit Intel" and people commonly think that's the whole world, ...)

Point: the reporter is correct in the general case (confirmed for cris-elf) and
if that was the actual request, the WORKSFORME would be incorrect.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

--- Comment #3 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
(In reply to Hans-Peter Nilsson from comment #2)
> (The implied reply from the description is yes

Sorry, I meant "the implied reply from the description is 'all targets'"...
Bah.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[jirik.svoboda at seznam dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Jiri Svoboda <jirik.svoboda at seznam dot cz> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |UNCONFIRMED
         Resolution|WORKSFORME                  |---

--- Comment #4 from Jiri Svoboda <jirik.svoboda at seznam dot cz> ---
As an example:

jirka@omelette:/tmp> echo "__int128 i;" > test.c
jirka@omelette:/tmp> gcc -o test -m64 test.c
/usr/lib64/gcc/x86_64-suse-linux/4.8/../../../../lib64/crt1.o: In function
`_start':
/home/abuild/rpmbuild/BUILD/glibc-2.18/csu/../sysdeps/x86_64/start.S:118:
undefined reference to `main'
collect2: error: ld returned 1 exit status
jirka@omelette:/tmp> gcc -c -m64 test.c
jirka@omelette:/tmp> echo "__int128 i;" > test.c
jirka@omelette:/tmp> gcc -c -m64 test.c
jirka@omelette:/tmp> gcc -c -m32 test.c
test.c:1:1: error: ‘__int128’ is not supported for this target
 __int128 i;
 ^
jirka@omelette:/tmp> gcc --version
gcc (SUSE Linux) 4.8.1 20130909 [gcc-4_8-branch revision 202388]
Copyright (C) 2013 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

jirka@omelette:/tmp> gcc -dumpmachine
x86_64-suse-linux
jirka@omelette:/tmp> 

So it works for x86_64 64-bit target, but not for 32-bit target.
And I mean I would like to have this functionality in all 32-bit and 64-bit
targets, not just in some. Since I'm developping a cross-platform network stack
which runs on Intel (32-bit, 64-bit), PowerPC (32-bit), Itanium (64-bit), ARM
(32-bit), MIPS (32-bit), etc.
>From gcc-bugs-return-449571-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Tue Apr 22 13:32:15 2014
Return-Path: <gcc-bugs-return-449571-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org>
Delivered-To: listarch-gcc-bugs@gcc.gnu.org
Received: (qmail 6010 invoked by alias); 22 Apr 2014 13:32:15 -0000
Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-bugs.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-help@gcc.gnu.org>
Sender: gcc-bugs-owner@gcc.gnu.org
Delivered-To: mailing list gcc-bugs@gcc.gnu.org
Received: (qmail 5956 invoked by uid 55); 22 Apr 2014 13:32:12 -0000
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/60849] [4.7/4.8 Regression] bogus comparison result type
Date: Tue, 22 Apr 2014 13:32:00 -0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 4.9.0
X-Bugzilla-Keywords:
X-Bugzilla-Severity: normal
X-Bugzilla-Who: rguenth at gcc dot gnu.org
X-Bugzilla-Status: ASSIGNED
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: rguenth at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 4.7.4
X-Bugzilla-Flags:
X-Bugzilla-Changed-Fields:
Message-ID: <bug-60849-4-xDJsdRW6xZ@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-60849-4@http.gcc.gnu.org/bugzilla/>
References: <bug-60849-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-SW-Source: 2014-04/txt/msg01591.txt.bz2
Content-length: 787

http://gcc.gnu.org/bugzilla/show_bug.cgi?id`849

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
Author: rguenth
Date: Tue Apr 22 13:31:41 2014
New Revision: 209632

URL: http://gcc.gnu.org/viewcvs?rev 9632&root=gcc&view=rev
Log:
2014-04-22  Richard Biener  <rguenther@suse.de>

    Backport from mainline
    2014-04-17  Richard Biener  <rguenther@suse.de>

    PR middle-end/60849
    * tree-ssa-propagate.c (valid_gimple_rhs_p): Only allow effective
    boolean results for comparisons.

    * g++.dg/opt/pr60849.C: New testcase.

Added:
    branches/gcc-4_9-branch/gcc/testsuite/g++.dg/opt/pr60849.C
Modified:
    branches/gcc-4_9-branch/gcc/ChangeLog
    branches/gcc-4_9-branch/gcc/testsuite/ChangeLog
    branches/gcc-4_9-branch/gcc/tree-ssa-propagate.c

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[hp at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Hans-Peter Nilsson <hp at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-04-23
     Ever confirmed|0                           |1

--- Comment #5 from Hans-Peter Nilsson <hp at gcc dot gnu.org> ---
Confirmed then, for e.g. cris-elf at r209710.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[mirh at protonmail dot ch]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

mirh at protonmail dot ch changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mirh at protonmail dot ch

--- Comment #6 from mirh at protonmail dot ch ---
Any update?
Of course performance is going to take a hit, but if you are working with
128bit data it's not like you have much other choices (other than painstakingly
working this around yourself).

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[feliks314159 at yahoo dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Bill Gray <feliks314159 at yahoo dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |feliks314159 at yahoo dot com

--- Comment #7 from Bill Gray <feliks314159 at yahoo dot com> ---
I'd really like to see 128-bit ints made available on 32-bit targets.  I
realize it wouldn't be fast for multiplying/dividing,  but almost all of my use
cases (and most uses in general,  I'd bet,  except for crypto) involve bitwise
operations.

For that matter,  I'd expect crypto folks would welcome 256- and higher-count
integers (get a SHA256 hash in a single scalar value).  But perhaps not;  they
already have workarounds for such issues.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note with __builtin_*_overflow, implementing 128bit integer (or larger) as a
class for C++ should be easier now.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

--- Comment #9 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
Note that supporting on any given 32-bit architecture requires making a 
choice of what the function-calling ABI should be for such types (and 
corresponding complex types) as function arguments and return values; 
architecture maintainers would need to make such a choice, in consultation 
with any group there may be that maintains the ABI for that architecture.

[Bug c/60846] Add 128-bit integer types for general use on 32-bit/64-bit CPUs

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60846

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #10 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
If a particular 32-bit arch agrees on ABI issues, one could use _BitInt(128) or
unsigned _BitInt(128) for it in GCC 14+, though currently only x86-64 64-bit
enables it.