Patchwork Use volatile global counters in test-tgmath.c.

login
register
mail settings
Submitter Stefan Liebler
Date April 18, 2018, 11:51 a.m.
Message ID <d31b50d1-dcd2-c0b6-8a55-5f5e7f31d9e4@linux.vnet.ibm.com>
Download mbox | patch
Permalink /patch/26775/
State New
Headers show

Comments

Stefan Liebler - April 18, 2018, 11:51 a.m.
Hi,

If build with -Os on s390x, the test-tgmath fails with:
float functions not called often enough (-10000)

Within compile_testf(), the counter (count_float) is saved
before the complex functions are called.
Afterwards the saved counter differs to the current-counter.

But the tests with the complex functions do not increment count_float!
Instead count_float is saved to a register before calling totalorder
and totalordermag which both increment count_float.
The compiler is allowed to do that as totalorderf and totalordermagf
is declared with __attribute__ ((__const__)) in math/bits/mathcalls.h.

Thus this patch adjusts the global counters to be volatile.
Then count_float is saved after totalordermag.

Okay to commit?

Bye
Stefan

---
ChangeLog:

	* math/test-tgmath.c (count_double, count_float,
	count_ ldouble,	count_cdouble, count_cfloat,
	count_cldouble): Use volatile int.
Joseph Myers - April 18, 2018, 4:05 p.m.
On Wed, 18 Apr 2018, Stefan Liebler wrote:

> 	* math/test-tgmath.c (count_double, count_float,
> 	count_ ldouble,	count_cdouble, count_cfloat,
> 	count_cldouble): Use volatile int.

OK.

Patch

commit 03fd4e7e0fd1d2f6dedee49bdbefde7b23a644e1
Author: Stefan Liebler <stli@linux.vnet.ibm.com>
Date:   Mon Apr 16 16:55:25 2018 +0200

    Use volatile global counters in test-tgmath.c.
    
    If build with -Os on s390x, the test-tgmath fails with:
    float functions not called often enough (-10000)
    
    Within compile_testf(), the counter (count_float) is saved
    before the complex functions are called.
    Afterwards the saved counter differs to the current-counter.
    
    But the tests with the complex functions do not increment count_float!
    Instead count_float is saved to a register before calling totalorder
    and totalordermag which both increment count_float.
    The compiler is allowed to do that as totalorderf and totalordermagf
    is declared with __attribute__ ((__const__)) in math/bits/mathcalls.h.
    
    Thus this patch adjusts the global counters to be volatile.
    Then count_float is saved after totalordermag.
    
    ChangeLog:
    
    	* math/test-tgmath.c (count_double, count_float,
    	count_ ldouble,	count_cdouble, count_cfloat,
    	count_cldouble): Use volatile int.

diff --git a/math/test-tgmath.c b/math/test-tgmath.c
index 6d8e7fd..2b0787c 100644
--- a/math/test-tgmath.c
+++ b/math/test-tgmath.c
@@ -45,12 +45,12 @@  complex float fz;
 complex double dz;
 complex long double lz;
 
-int count_double;
-int count_float;
-int count_ldouble;
-int count_cdouble;
-int count_cfloat;
-int count_cldouble;
+volatile int count_double;
+volatile int count_float;
+volatile int count_ldouble;
+volatile int count_cdouble;
+volatile int count_cfloat;
+volatile int count_cldouble;
 
 #define NCALLS     134
 #define NCALLS_INT 4