GNU Guile 2.1.6 released (beta)

GNU Guile 2.1.6 released (beta)

[Andy Wingo]

We are pleased to announce GNU Guile release 2.1.6.

Guile 2.1.6 is the sixth pre-release in what will eventually become the
2.2 release series.  We encourage you to test this release and provide
feedback to guile-devel@gnu.org.

This is a bug-fix release, mostly fixing bugs related to threads and
interrupts.  It also includes support for fast floating-point
comparisons; special thanks to new Guile committer David Thompson for
this feature.  See the full NEWS below, for details.

The Guile web page is located at http://gnu.org/software/guile/, and
among other things, it contains a copy of the Guile manual and pointers
to more resources.

Guile is an implementation of the Scheme programming language, with
support for many SRFIs, packaged for use in a wide variety of
environments.  In addition to implementing the R5RS Scheme standard,
Guile includes a module system, full access to POSIX system calls,
networking support, multiple threads, dynamic linking, a foreign
function call interface, and powerful string processing.

Guile can run interactively, as a script interpreter, and as a Scheme
compiler to VM bytecode.  It is also packaged as a library so that
applications can easily incorporate a complete Scheme interpreter/VM.
An application can use Guile as an extension language, a clean and
powerful configuration language, or as multi-purpose "glue" to connect
primitives provided by the application.  It is easy to call Scheme code
From C code and vice versa.  Applications can add new functions, data
types, control structures, and even syntax to Guile, to create a
domain-specific language tailored to the task at hand.

Guile 2.1.6 can be installed in parallel with Guile 2.0.x; see
http://www.gnu.org/software/guile/manual/html_node/Parallel-Installations.html.

A more detailed NEWS summary follows these details on how to get the
Guile sources.

Here are the compressed sources:
  http://alpha.gnu.org/gnu/guile/guile-2.1.6.tar.gz   (18MB)
  http://alpha.gnu.org/gnu/guile/guile-2.1.6.tar.xz   (11MB)

Here are the GPG detached signatures[*]:
  http://alpha.gnu.org/gnu/guile/guile-2.1.6.tar.gz.sig
  http://alpha.gnu.org/gnu/guile/guile-2.1.6.tar.xz.sig

Use a mirror for higher download bandwidth:
  http://www.gnu.org/order/ftp.html

Here are the SHA256 checksums:

  6f3f34eb2d4b2c6e458c4103e731ce51479c0a4713cf4cfcc29844ec17f0305f  guile-2.1.6.tar.gz
  f3c25770a34bdd4391593f793107991a57b88350f2dcb947e9924d6522595d59  guile-2.1.6.tar.xz

[*] Use a .sig file to verify that the corresponding file (without the
.sig suffix) is intact.  First, be sure to download both the .sig file
and the corresponding tarball.  Then, run a command like this:

  gpg --verify guile-2.1.6.tar.gz.sig

If that command fails because you don't have the required public key,
then run this command to import it:

  gpg --keyserver keys.gnupg.net --recv-keys FF478FB264DE32EC296725A3DDC0F5358812F8F2

and rerun the 'gpg --verify' command.

This release was bootstrapped with the following tools:
  Autoconf 2.69
  Automake 1.15
  Libtool 2.4.6
  Gnulib v0.1-800-g68b6ade
  Makeinfo 6.1


Changes in 2.1.6 (changes since the 2.1.5 alpha release):

* New interfaces
** suspendable-continuation?

This predicate returns true if the delimited continuation captured by
aborting to a prompt would be able to be resumed.  See "Prompt
Primitives" in the manual for more.

** scm_c_prepare_to_wait_on_fd, scm_c_prepare_to_wait_on_cond,
** scm_c_wait_finished

See "Interrupts" in the manual for more.

* Performance improvements

** Support unboxed floating-point comparisons

Thanks to David Thompson for this work.

* Incompatible changes

** Rename new array functions

See "Arrays as arrays of arrays" in the manual for more.

* Bug fixes

** `scm_gc_warn_proc' writes directly to stderr

The garbage collector sometimes has warnings to display to the user.
Before, Guile would see if the current warning port was a file port, and
in that case write the warning to that file, and otherwise default to
stderr.  Now Guile just writes to stderr, fixing a bug where determining
the current warning port would allocate and thus deadlock as the GC
warnings are issued with the GC lock held.

** Fix miscompilation in significant-bits computation for loop vars
** Fix many threading bugs
** Fix macOS portability bugs
Thanks to Matt Wette!

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 508 bytes --]


> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> We are pleased to announce GNU Guile release 2.1.6.
> 
> Guile 2.1.6 is the sixth pre-release in what will eventually become the
> 2.2 release series.  We encourage you to test this release and provide
> feedback to guile-devel@gnu.org.

Saw this one last round.  Mac OS, now gcc-6.3.0:

;;; ("#i1@-0" 1.0 -0.0)
FAIL: numbers.test: string->number: valid complex number strings

I am going to see if I can generate the assembly.

Matt


[-- Attachment #2: Type: text/html, Size: 1906 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 910 bytes --]


> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> We are pleased to announce GNU Guile release 2.1.6.
> 
> Guile 2.1.6 is the sixth pre-release in what will eventually become the
> 2.2 release series.  We encourage you to test this release and provide
> feedback to guile-devel@gnu.org.

I don’t see problems with this but I think it is a bug: should use labs for longs.

--- libguile/numbers.c.orig     2016-11-21 13:56:23.000000000 -0800
+++ libguile/numbers.c  2017-01-20 06:04:53.000000000 -0800
@@ -9951,7 +9951,7 @@
   long n_size = scm_to_long (scm_integer_length (n));
   long d_size = scm_to_long (scm_integer_length (d));
 
-  if (abs (n_size - d_size) > 1)
+  if (labs (n_size - d_size) > 1)
     return (scm_difference (log_of_exact_integer (n),
                            log_of_exact_integer (d)));
   else if (scm_is_false (scm_negative_p (n)))


[-- Attachment #2: Type: text/html, Size: 4116 bytes --]

Re: GNU Guile 2.1.6 released (beta) [numbers.c]

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 2296 bytes --]


> On Jan 19, 2017, at 7:01 PM, Matt Wette <matt.wette@gmail.com> wrote:
> 
> 
>> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com <mailto:wingo@pobox.com>> wrote:
>> 
>> We are pleased to announce GNU Guile release 2.1.6.
>> 
>> Guile 2.1.6 is the sixth pre-release in what will eventually become the
>> 2.2 release series.  We encourage you to test this release and provide
>> feedback to guile-devel@gnu.org <mailto:guile-devel@gnu.org>.
> 
> Saw this one last round.  Mac OS, now gcc-6.3.0:
> 
> ;;; ("#i1@-0" 1.0 -0.0)
> FAIL: numbers.test: string->number: valid complex number strings
> 
> I am going to see if I can generate the assembly.

Short story: scm_c_make_polar is broken for the Mac.  Guile needs to decide if it want to use __sincos() on Mac, or suppress optimization, or ...

I was able to get the above to work (guile-2.1.5) by using
 SCM
+#ifdef __APPLE__
+__attribute__((optimize("O0")))
+#endif
 scm_c_make_polar (double mag, double ang)


In scm_c_make_polar, “gcc -O2”  turns sin(), cos() into cexp(), since cexp(i*x) = cos(x) + i*sin(x):

gcc -O0 =>
LM4339:
	movq	-32(%rbp), %rax
	movd	%rax, %xmm0
	call	_sin
	movd	%xmm0, %rax
	movq	%rax, -8(%rbp)
LM4340:
	movq	-32(%rbp), %rax
	movd	%rax, %xmm0
	call	_cos
	movd	%xmm0, %rax
	movq	%rax, -16(%rbp)

gcc -O2 =>
	pxor	%xmm0, %xmm0
LVL2703:
	call	_cexp

I wrote a little C program to show that cexp() does not preserve the zero-signed-ness:

 cos,sin: +1.000000 -0.000000
__sincos: +1.000000 -0.000000
    cexp: +1.000000 +0.000000

The scm_c_make_polar will use sincos() if available, but macOS does not have sincos(), it has __sincos().

#include <stdio.h>
#include <stdint.h>
#include <math.h>
#include <complex.h>

extern double z, p, n;

int main() {
  double complex c;
  double d, sine, cosine;

  d = z*n;
  printf(" cos,sin: %+f %+f\n", cos(d), sin(d));
  __sincos(d, &sine, &cosine);
  printf("__sincos: %+f %+f\n", cosine, sine);
  c = cexp(CMPLX(0.0, d));
  printf("    cexp: %+f %+f\n", creal(c), cimag(c));
}

double z = 0.0, p = +1.0, n = -1.0;

Incidentally, the above program will not compile on my machine w/ gcc-6.3.0.  “gcc -std=c11” or “gcc -std=c99” will not recognize the standard macro CMPLX().




[-- Attachment #2: Type: text/html, Size: 13644 bytes --]

Re: GNU Guile 2.1.6 released (beta) [numbers.c]

[Amirouche]

[-- Attachment #1: Type: text/plain, Size: 2423 bytes --]

Le 20/01/2017 à 20:01, Matt Wette a écrit :
>
>> On Jan 19, 2017, at 7:01 PM, Matt Wette <matt.wette@gmail.com 
>> <mailto:matt.wette@gmail.com>> wrote:
>>
>>
>>> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com 
>>> <mailto:wingo@pobox.com>> wrote:
>>>
>>> We are pleased to announce GNU Guile release 2.1.6.
>>>
>>> Guile 2.1.6 is the sixth pre-release in what will eventually become the
>>> 2.2 release series.  We encourage you to test this release and provide
>>> feedback to guile-devel@gnu.org <mailto:guile-devel@gnu.org>.
>>
>> Saw this one last round.  Mac OS, now gcc-6.3.0:
>>
>> ;;; ("#i1@-0" 1.0 -0.0)
>> FAIL: numbers.test: string->number: valid complex number strings
>>
>> I am going to see if I can generate the assembly.
>
> Short story: scm_c_make_polar is broken for the Mac.  Guile needs to 
> decide if it want to use __sincos() on Mac, or suppress optimization, 
> or ...
>
> I was able to get the above to work (guile-2.1.5) by using
>  SCM
> +#ifdef __APPLE__
> +__attribute__((optimize("O0")))
> +#endif
>  scm_c_make_polar (double mag, double ang)
>
>
> In scm_c_make_polar, “gcc -O2”  turns sin(), cos() into cexp(), since 
> cexp(i*x) = cos(x) + i*sin(x):
>
> gcc -O0 =>
> LM4339:
> movq-32(%rbp), %rax
> movd%rax, %xmm0
> call_sin
> movd%xmm0, %rax
> movq%rax, -8(%rbp)
> LM4340:
> movq-32(%rbp), %rax
> movd%rax, %xmm0
> call_cos
> movd%xmm0, %rax
> movq%rax, -16(%rbp)
>
> gcc -O2 =>
> pxor%xmm0, %xmm0
> LVL2703:
> call_cexp
>
> I wrote a little C program to show that cexp() does not preserve the 
> zero-signed-ness:
>
>  cos,sin: +1.000000 -0.000000
> __sincos: +1.000000 -0.000000
>     cexp: +1.000000 +0.000000
>
> The scm_c_make_polar will use sincos() if available, but macOS does 
> not have sincos(), it has __sincos().
>
> #include <stdio.h>
> #include <stdint.h>
> #include <math.h>
> #include <complex.h>
>
> extern double z, p, n;
>
> int main() {
>   double complex c;
>   double d, sine, cosine;
>
>   d = z*n;
>   printf(" cos,sin: %+f %+f\n", cos(d), sin(d));
>   __sincos(d, &sine, &cosine);
>   printf("__sincos: %+f %+f\n", cosine, sine);
>   c = cexp(CMPLX(0.0, d));
>   printf("    cexp: %+f %+f\n", creal(c), cimag(c));
> }
>
> double z = 0.0, p = +1.0, n = -1.0;
>
> Incidentally, the above program will not compile on my machine w/ 
> gcc-6.3.0.  “gcc -std=c11” or “gcc -std=c99” will not recognize the 
> standard macro CMPLX().
>
>
>
wow!

[-- Attachment #2: Type: text/html, Size: 17990 bytes --]

Re: GNU Guile 2.1.6 released (beta) [numbers.c]

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 1379 bytes --]


> On Jan 20, 2017, at 11:01 AM, Matt Wette <matt.wette@gmail.com> wrote:
> 
>> 
>> On Jan 19, 2017, at 7:01 PM, Matt Wette <matt.wette@gmail.com <mailto:matt.wette@gmail.com>> wrote:
>> 
>> 
>>> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com <mailto:wingo@pobox.com>> wrote:
>>> 
>>> We are pleased to announce GNU Guile release 2.1.6.
>>> 
>>> Guile 2.1.6 is the sixth pre-release in what will eventually become the
>>> 2.2 release series.  We encourage you to test this release and provide
>>> feedback to guile-devel@gnu.org <mailto:guile-devel@gnu.org>.
>> 
>> Saw this one last round.  Mac OS, now gcc-6.3.0:
>> 
>> ;;; ("#i1@-0" 1.0 -0.0)
>> FAIL: numbers.test: string->number: valid complex number strings
>> 
>> I am going to see if I can generate the assembly.
> 
> Short story: scm_c_make_polar is broken for the Mac.  Guile needs to decide if it want to use __sincos() on Mac, or suppress optimization, or ...
> 

a patch, with which make check passes all tests on macOS 10.12.2, aka Sierra:

--- libguile/numbers.c.orig     2017-01-19 16:57:30.000000000 -0800
+++ libguile/numbers.c  2017-01-19 20:20:58.000000000 -0800
@@ -9099,6 +9099,9 @@
 #undef FUNC_NAME
 
 SCM
+#if defined(__APPLE__) && defined(__GNUC__)
+__attribute__((optimize("O0")))
+#endif
 scm_c_make_polar (double mag, double ang)
 {
   double s, c;


[-- Attachment #2: Type: text/html, Size: 6227 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 914 bytes --]


> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> We are pleased to announce GNU Guile release 2.1.6.
> 
> Guile 2.1.6 is the sixth pre-release in what will eventually become the
> 2.2 release series.  We encourage you to test this release and provide
> feedback to guile-devel@gnu.org.

So I am trying to compile using clang.  This has detected issues with atomics-internal.h.  I can fix most issues with casts:
 static inline uint32_t
 scm_atomic_subtract_uint32 (uint32_t *loc, uint32_t arg)
 {
-  return atomic_fetch_sub (loc, arg);
+  return atomic_fetch_sub ((_Atomic uint32_t *)loc, arg);
 }

But problems remain with scm_atomic_set_pointer and scm_atomic_ref_pointer which use void** and is apparently not defined for use with atomics.

Apparently gcc does not have a problem with this use but I’m guessing the C11 standard may have an issue with it.

Matt


[-- Attachment #2: Type: text/html, Size: 3029 bytes --]

Re: GNU Guile 2.1.6 released (beta) [numbers.c]

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 1313 bytes --]


> On Jan 20, 2017, at 12:23 PM, Matt Wette <matt.wette@gmail.com> wrote:
> 
>> 
>> On Jan 20, 2017, at 11:01 AM, Matt Wette <matt.wette@gmail.com <mailto:matt.wette@gmail.com>> wrote:
>> 
>>> 
>>> On Jan 19, 2017, at 7:01 PM, Matt Wette <matt.wette@gmail.com <mailto:matt.wette@gmail.com>> wrote:
>>> 
>>> 
>>>> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com <mailto:wingo@pobox.com>> wrote:
>>>> 
>>>> We are pleased to announce GNU Guile release 2.1.6.
>>>> 
>>>> Guile 2.1.6 is the sixth pre-release in what will eventually become the
>>>> 2.2 release series.  We encourage you to test this release and provide
>>>> feedback to guile-devel@gnu.org <mailto:guile-devel@gnu.org>.
>>> 
>>> Saw this one last round.  Mac OS, now gcc-6.3.0:
>>> 
>>> ;;; ("#i1@-0" 1.0 -0.0)
>>> FAIL: numbers.test: string->number: valid complex number strings
>>> 
>>> I am going to see if I can generate the assembly.
>> 
>> Short story: scm_c_make_polar is broken for the Mac.  Guile needs to decide if it want to use __sincos() on Mac, or suppress optimization, or ...
>> 
> 

Attached is another patch.  This one adds check for __sincos() to configure.am, configure and #ifdef to numbers.c to use __sincos() on the Mac.  I don’t know if __APPLE__ is required anymore or not.

Matt


[-- Attachment #2.1: Type: text/html, Size: 3936 bytes --]

[-- Attachment #2.2: patch.__sincos --]
[-- Type: application/octet-stream, Size: 1848 bytes --]

--- libguile/numbers.c-orig	2017-01-27 16:44:27.000000000 -0800
+++ libguile/numbers.c	2017-01-27 16:47:49.000000000 -0800
@@ -9109,6 +9109,8 @@
      details.  */
 #if (defined HAVE_SINCOS) && (defined __GLIBC__) && (defined _GNU_SOURCE)
   sincos (ang, &s, &c);
+#elif (defined HAVE___SINCOS) && (defined __APPLE__)
+  __sincos (ang, &s, &c);
 #else
   s = sin (ang);
   c = cos (ang);
@@ -9951,7 +9953,7 @@
   long n_size = scm_to_long (scm_integer_length (n));
   long d_size = scm_to_long (scm_integer_length (d));
 
-  if (abs (n_size - d_size) > 1)
+  if (labs (n_size - d_size) > 1)
     return (scm_difference (log_of_exact_integer (n),
 			    log_of_exact_integer (d)));
   else if (scm_is_false (scm_negative_p (n)))
--- configure.ac-orig	2017-01-27 16:42:08.000000000 -0800
+++ configure.ac	2017-01-27 16:42:45.000000000 -0800
@@ -1152,8 +1152,9 @@
 #   asinh, acosh, atanh, trunc - C99 standard, generally not available on
 #                                older systems
 #   sincos - GLIBC extension
+#   __sincos - APPLE extension
 #
-AC_CHECK_FUNCS(asinh acosh atanh copysign finite sincos trunc)
+AC_CHECK_FUNCS(asinh acosh atanh copysign finite sincos __sincos trunc)
 
 # C99 specifies isinf and isnan as macros.
 # HP-UX provides only macros, no functions.
--- configure-orig	2017-01-27 16:42:14.000000000 -0800
+++ configure	2017-01-27 16:56:43.000000000 -0800
@@ -52786,8 +52786,9 @@
 #   asinh, acosh, atanh, trunc - C99 standard, generally not available on
 #                                older systems
 #   sincos - GLIBC extension
+#   __sincos - APPLE extension
 #
-for ac_func in asinh acosh atanh copysign finite sincos trunc
+for ac_func in asinh acosh atanh copysign finite sincos __sincos trunc
 do :
   as_ac_var=`$as_echo "ac_cv_func_$ac_func" | $as_tr_sh`
 ac_fn_c_check_func "$LINENO" "$ac_func" "$as_ac_var"

[-- Attachment #2.3: Type: text/html, Size: 237 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 619 bytes --]


> On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com> wrote:
> 
> We are pleased to announce GNU Guile release 2.1.6.
> 
> Guile 2.1.6 is the sixth pre-release in what will eventually become the
> 2.2 release series.  We encourage you to test this release and provide
> feedback to guile-devel@gnu.org.

I agree with Triple-X’s comment near line 391 in doc/ref/guile.texi:

@c XXX: Would be nicer if it were close to the (sxml simple) documentation.
@include sxml-match.texi

@include scsh.texi
@include curried.texi

@include statprof.texi
@include sxml.texi
@include texinfo.texi
Matt


[-- Attachment #2: Type: text/html, Size: 4404 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Andy Wingo]

Hi,

Picking up this topic...

On Wed 25 Jan 2017 04:35, Matt Wette <matt.wette@gmail.com> writes:

>  On Jan 18, 2017, at 6:26 PM, Andy Wingo <wingo@pobox.com> wrote:
>
>  We are pleased to announce GNU Guile release 2.1.6.
>
>  Guile 2.1.6 is the sixth pre-release in what will eventually become the
>  2.2 release series. We encourage you to test this release and provide
>  feedback to guile-devel@gnu.org.
>
> So I am trying to compile using clang. This has detected issues with atomics-internal.h. I can fix most issues with casts:
> static inline uint32_t
> scm_atomic_subtract_uint32 (uint32_t *loc, uint32_t arg)
> {
> - return atomic_fetch_sub (loc, arg);
> + return atomic_fetch_sub ((_Atomic uint32_t *)loc, arg);
> }
>
> But problems remain with scm_atomic_set_pointer and scm_atomic_ref_pointer which use void** and is apparently not defined for use with atomics.
>
> Apparently gcc does not have a problem with this use but I’m guessing the C11 standard may have an issue with it.

WDYT about this?

Andy

diff --git a/libguile/atomics-internal.h b/libguile/atomics-internal.h
index f2d17e1..da3ebd8 100644
--- a/libguile/atomics-internal.h
+++ b/libguile/atomics-internal.h
@@ -31,46 +31,56 @@
 #ifdef HAVE_STDATOMIC_H
 
 #include <stdatomic.h>
+
 static inline uint32_t
 scm_atomic_subtract_uint32 (uint32_t *loc, uint32_t arg)
 {
-  return atomic_fetch_sub (loc, arg);
+  atomic_uint_least32_t *a_loc = (atomic_uint_least32_t *) loc;
+  return atomic_fetch_sub (a_loc, arg);
 }
 static inline _Bool
 scm_atomic_compare_and_swap_uint32 (uint32_t *loc, uint32_t *expected,
                                     uint32_t desired)
 {
-  return atomic_compare_exchange_weak (loc, expected, desired);
+  atomic_uint_least32_t *a_loc = (atomic_uint_least32_t *) loc;
+  return atomic_compare_exchange_weak (a_loc, expected, desired);
 }
 static inline void
 scm_atomic_set_pointer (void **loc, void *val)
 {
-  atomic_store (loc, val);
+  atomic_uintptr_t *a_loc = (atomic_uintptr_t *) loc;
+  atomic_store (a_loc, (uintptr_t) val);
 }
 static inline void *
 scm_atomic_ref_pointer (void **loc)
 {
-  return atomic_load (loc);
+  atomic_uintptr_t *a_loc = (atomic_uintptr_t *) loc;
+  return (void *) atomic_load (a_loc);
 }
 static inline void
 scm_atomic_set_scm (SCM *loc, SCM val)
 {
-  atomic_store (loc, val);
+  atomic_uintptr_t *a_loc = (atomic_uintptr_t *) loc;
+  atomic_store (a_loc, SCM_UNPACK (val));
 }
 static inline SCM
 scm_atomic_ref_scm (SCM *loc)
 {
-  return atomic_load (loc);
+  atomic_uintptr_t *a_loc = (atomic_uintptr_t *) loc;
+  return SCM_PACK (atomic_load (a_loc));
 }
 static inline SCM
 scm_atomic_swap_scm (SCM *loc, SCM val)
 {
-  return atomic_exchange (loc, val);
+  return SCM_PACK (atomic_exchange (loc, val));
 }
 static inline _Bool
 scm_atomic_compare_and_swap_scm (SCM *loc, SCM *expected, SCM desired)
 {
-  return atomic_compare_exchange_weak (loc, expected, desired);
+  atomic_uintptr_t *a_loc = (atomic_uintptr_t *) loc;
+  return atomic_compare_exchange_weak (a_loc,
+                                       (uintptr_t *) expected,
+                                       SCM_UNPACK (desired));
 }
 #else /* HAVE_STDATOMIC_H */
 

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 1609 bytes --]


> On Mar 9, 2017, at 1:33 PM, Andy Wingo <wingo@pobox.com> wrote:
> On Wed 25 Jan 2017 04:35, Matt Wette <matt.wette@gmail.com <mailto:matt.wette@gmail.com>> writes:
>> So I am trying to compile using clang. This has detected issues with atomics-internal.h. I can fix most issues with casts:
>> static inline uint32_t
>> scm_atomic_subtract_uint32 (uint32_t *loc, uint32_t arg)
>> {
>> - return atomic_fetch_sub (loc, arg);
>> + return atomic_fetch_sub ((_Atomic uint32_t *)loc, arg);
>> }
>> 
>> But problems remain with scm_atomic_set_pointer and scm_atomic_ref_pointer which use void** and is apparently not defined for use with atomics.
>> 
>> Apparently gcc does not have a problem with this use but I’m guessing the C11 standard may have an issue with it.
> 
> WDYT about this?
> 
> Andy

Have not tried to chase this down but ...

In file included from async.c:27:
../libguile/atomics-internal.h:75:20: error: address argument to atomic operation must be a pointer to _Atomic type ('SCM *' (aka 'struct scm_unused_struct **') invalid)
  return SCM_PACK (atomic_exchange (loc, val));
                   ^                ~~~
/opt/local/libexec/llvm-3.9/bin/../lib/clang/3.9.1/include/stdatomic.h:137:42: note: expanded from macro 'atomic_exchange'
#define atomic_exchange(object, desired) __c11_atomic_exchange(object, desired, __ATOMIC_SEQ_CST)
                                         ^                     ~~~~~~
../libguile/tags.h:105:32: note: expanded from macro 'SCM_PACK'
#   define SCM_PACK(x) ((SCM) (x))
                               ^
1 error generated.


[-- Attachment #2: Type: text/html, Size: 8641 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

[-- Attachment #1: Type: text/plain, Size: 2103 bytes --]


> On Mar 9, 2017, at 6:01 PM, Matt Wette <matt.wette@gmail.com> wrote:
> 
> 
>> On Mar 9, 2017, at 1:33 PM, Andy Wingo <wingo@pobox.com <mailto:wingo@pobox.com>> wrote:
>> On Wed 25 Jan 2017 04:35, Matt Wette <matt.wette@gmail.com <mailto:matt.wette@gmail.com>> writes:
>>> So I am trying to compile using clang. This has detected issues with atomics-internal.h. I can fix most issues with casts:
>>> static inline uint32_t
>>> scm_atomic_subtract_uint32 (uint32_t *loc, uint32_t arg)
>>> {
>>> - return atomic_fetch_sub (loc, arg);
>>> + return atomic_fetch_sub ((_Atomic uint32_t *)loc, arg);
>>> }
>>> 
>>> But problems remain with scm_atomic_set_pointer and scm_atomic_ref_pointer which use void** and is apparently not defined for use with atomics.
>>> 
>>> Apparently gcc does not have a problem with this use but I’m guessing the C11 standard may have an issue with it.
>> 
>> WDYT about this?
>> 
>> Andy
> 
> Have not tried to chase this down but ...
> 
> In file included from async.c:27:
> ../libguile/atomics-internal.h:75:20: error: address argument to atomic operation must be a pointer to _Atomic type ('SCM *' (aka 'struct scm_unused_struct **') invalid)
>   return SCM_PACK (atomic_exchange (loc, val));
>                    ^                ~~~
> /opt/local/libexec/llvm-3.9/bin/../lib/clang/3.9.1/include/stdatomic.h:137:42: note: expanded from macro 'atomic_exchange'
> #define atomic_exchange(object, desired) __c11_atomic_exchange(object, desired, __ATOMIC_SEQ_CST)
>                                          ^                     ~~~~~~
> ../libguile/tags.h:105:32: note: expanded from macro 'SCM_PACK'
> #   define SCM_PACK(x) ((SCM) (x))
>                                ^
> 1 error generated.
> 

With this change it compiles under clang-3.9, though generates warning, and gets through make check (with required other unrelated patches) OK:

static inline SCM
scm_atomic_swap_scm (SCM *loc, SCM val)
{
  volatile atomic_uintptr_t *a_loc = (volatile atomic_uintptr_t *)loc;
  return SCM_PACK (atomic_exchange (a_loc, val));
}



[-- Attachment #2: Type: text/html, Size: 9624 bytes --]

Re: GNU Guile 2.1.6 released (beta)

[Andy Wingo]

On Fri 10 Mar 2017 03:01, Matt Wette <matt.wette@gmail.com> writes:

> In file included from async.c:27:
> ../libguile/atomics-internal.h:75:20: error: address argument to atomic operation must be a pointer to _Atomic type ('SCM *' (aka 'struct scm_unused_struct **') invalid)
> return SCM_PACK (atomic_exchange (loc, val));

Right, forgot to fix this one.  Following up on your other message where
you had success with clang, I fixed this and pushed.  We'll see in the
next prerelease then :)

Andy

Re: GNU Guile 2.1.6 released (beta)

[Matt Wette]

> On Mar 10, 2017, at 12:13 AM, Andy Wingo <wingo@pobox.com> wrote:
> 
> On Fri 10 Mar 2017 03:01, Matt Wette <matt.wette@gmail.com> writes:
> 
>> In file included from async.c:27:
>> ../libguile/atomics-internal.h:75:20: error: address argument to atomic operation must be a pointer to _Atomic type ('SCM *' (aka 'struct scm_unused_struct **') invalid)
>> return SCM_PACK (atomic_exchange (loc, val));
> 
> Right, forgot to fix this one.  Following up on your other message where
> you had success with clang, I fixed this and pushed.  We'll see in the
> next prerelease then :)

FYI, I rebuilt with gcc: gcc is generating a warning on my code above. — Matt