competime_assert failure without -O3 optimization flag expected ?

competime_assert failure without -O3 optimization flag expected ?

[Alireza Haghdoost]

> +struct all_io_list *get_all_io_list(int save_mask, size_t *sz)
> +{
> +       struct all_io_list *rep;
> +       struct thread_data *td;
> +       size_t depth;
> +       void *next;
> +       int i, nr;
> +
> +       compiletime_assert(sizeof(struct all_io_list) == 8, "all_io_list");


I am getting compile time assertion failure on the following line of
code when I remove the optimization flag (-O3) in the Makefile. Is
this something expected ? I want to remove optimizations in order to
debug my code based on original source code line order not optimized
code order.


--Alireza

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 08:46 AM, Alireza Haghdoost wrote:
>> +struct all_io_list *get_all_io_list(int save_mask, size_t *sz)
>> +{
>> +       struct all_io_list *rep;
>> +       struct thread_data *td;
>> +       size_t depth;
>> +       void *next;
>> +       int i, nr;
>> +
>> +       compiletime_assert(sizeof(struct all_io_list) == 8, "all_io_list");
>
>
> I am getting compile time assertion failure on the following line of
> code when I remove the optimization flag (-O3) in the Makefile. Is
> this something expected ? I want to remove optimizations in order to
> debug my code based on original source code line order not optimized
> code order.

Yeah it's expected, but should be fixed. I just haven't looked into why 
the compiletime_assert() fails if optimizations are disabled. It really 
shouldn't, since the size of the struct is still 8 when disabled. So 
it's a bug in the compiletime_assert() code. Feel free to poke at it :-)

-- 
Jens Axboe

Re: competime_assert failure without -O3 optimization flag expected ?

[Alireza Haghdoost]

Sure, I will investigate it. Just want to confirm the issue is in the
compiletime_assert() code as you mentioned and it is not in that line
of code that I cite in my last email. I managed to remedy this issue a
little bit by using -O1 instead of -O3 but still the optimized program
order is not exactly similar to the default source code order.

--Alireza

On Mon, Jun 29, 2015 at 10:47 AM, Jens Axboe <axboe@kernel.dk> wrote:
> On 06/29/2015 08:46 AM, Alireza Haghdoost wrote:
>>>
>>> +struct all_io_list *get_all_io_list(int save_mask, size_t *sz)
>>> +{
>>> +       struct all_io_list *rep;
>>> +       struct thread_data *td;
>>> +       size_t depth;
>>> +       void *next;
>>> +       int i, nr;
>>> +
>>> +       compiletime_assert(sizeof(struct all_io_list) == 8,
>>> "all_io_list");
>>
>>
>>
>> I am getting compile time assertion failure on the following line of
>> code when I remove the optimization flag (-O3) in the Makefile. Is
>> this something expected ? I want to remove optimizations in order to
>> debug my code based on original source code line order not optimized
>> code order.
>
>
> Yeah it's expected, but should be fixed. I just haven't looked into why the
> compiletime_assert() fails if optimizations are disabled. It really
> shouldn't, since the size of the struct is still 8 when disabled. So it's a
> bug in the compiletime_assert() code. Feel free to poke at it :-)
>
> --
> Jens Axboe
>

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 09:53 AM, Alireza Haghdoost wrote:
> Sure, I will investigate it. Just want to confirm the issue is in the
> compiletime_assert() code as you mentioned and it is not in that line
> of code that I cite in my last email. I managed to remedy this issue a
> little bit by using -O1 instead of -O3 but still the optimized program
> order is not exactly similar to the default source code order.

It bothers me too, I'll turn off optimizations for debugging things too, 
and just end up uncommenting the lines to make that work. Apparently it 
hasn't bothered me enough to fix it up yet, but I would love to see it 
fixed.

-- 
Jens Axboe

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 09:55 AM, Jens Axboe wrote:
> On 06/29/2015 09:53 AM, Alireza Haghdoost wrote:
>> Sure, I will investigate it. Just want to confirm the issue is in the
>> compiletime_assert() code as you mentioned and it is not in that line
>> of code that I cite in my last email. I managed to remedy this issue a
>> little bit by using -O1 instead of -O3 but still the optimized program
>> order is not exactly similar to the default source code order.
>
> It bothers me too, I'll turn off optimizations for debugging things too,
> and just end up uncommenting the lines to make that work. Apparently it
> hasn't bothered me enough to fix it up yet, but I would love to see it
> fixed.

BTW, the issue is basically that without optimizations, gcc doesn't 
realize that the declared extern function doesn't get called. So asserts 
fail, doesn't matter if they are true or false.

-- 
Jens Axboe

Re: competime_assert failure without -O3 optimization flag expected ?

[Alireza Haghdoost]

Jens,

Sounds like the compiletime_assert() method was not function and is
relying on the optimizer performing dead code elimination to remove
the call to prefix ## suffix

How about this solution: I guess we have to work around it to make it
more portable since it works with C11.

diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
index 40e857c..7c9ba57 100644
--- a/fio-arh/compiler/compiler.h
+++ b/fio-arh/compiler/compiler.h
@@ -33,26 +33,6 @@
  1; \
 })

-#ifndef __compiletime_error
-#define __compiletime_error(message)
-#endif
-#ifndef __compiletime_error_fallback
-#define __compiletime_error_fallback(condition) do { } while (0)
-#endif
-
-#define __compiletime_assert(condition, msg, prefix, suffix) \
- do { \
- int __cond = !(condition); \
- extern void prefix ## suffix(void) __compiletime_error(msg); \
- if (__cond) \
- prefix ## suffix(); \
- __compiletime_error_fallback(__cond); \
- } while (0)
-
-#define _compiletime_assert(condition, msg, prefix, suffix) \
- __compiletime_assert(condition, msg, prefix, suffix)
-
-#define compiletime_assert(condition, msg) \
- _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
+#define compiletime_assert(condition, msg) _Static_assert(condition, msg)

 #endif

On Mon, Jun 29, 2015 at 11:04 AM, Jens Axboe <axboe@kernel.dk> wrote:
> On 06/29/2015 09:55 AM, Jens Axboe wrote:
>>
>> On 06/29/2015 09:53 AM, Alireza Haghdoost wrote:
>>>
>>> Sure, I will investigate it. Just want to confirm the issue is in the
>>> compiletime_assert() code as you mentioned and it is not in that line
>>> of code that I cite in my last email. I managed to remedy this issue a
>>> little bit by using -O1 instead of -O3 but still the optimized program
>>> order is not exactly similar to the default source code order.
>>
>>
>> It bothers me too, I'll turn off optimizations for debugging things too,
>> and just end up uncommenting the lines to make that work. Apparently it
>> hasn't bothered me enough to fix it up yet, but I would love to see it
>> fixed.
>
>
> BTW, the issue is basically that without optimizations, gcc doesn't realize
> that the declared extern function doesn't get called. So asserts fail,
> doesn't matter if they are true or false.
>
> --
> Jens Axboe
>

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 11:42 AM, Alireza Haghdoost wrote:
> Jens,
>
> Sounds like the compiletime_assert() method was not function and is
> relying on the optimizer performing dead code elimination to remove
> the call to prefix ## suffix
>
> How about this solution: I guess we have to work around it to make it
> more portable since it works with C11.
>
> diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
> index 40e857c..7c9ba57 100644
> --- a/fio-arh/compiler/compiler.h
> +++ b/fio-arh/compiler/compiler.h
> @@ -33,26 +33,6 @@
>    1; \
>   })
>
> -#ifndef __compiletime_error
> -#define __compiletime_error(message)
> -#endif
> -#ifndef __compiletime_error_fallback
> -#define __compiletime_error_fallback(condition) do { } while (0)
> -#endif
> -
> -#define __compiletime_assert(condition, msg, prefix, suffix) \
> - do { \
> - int __cond = !(condition); \
> - extern void prefix ## suffix(void) __compiletime_error(msg); \
> - if (__cond) \
> - prefix ## suffix(); \
> - __compiletime_error_fallback(__cond); \
> - } while (0)
> -
> -#define _compiletime_assert(condition, msg, prefix, suffix) \
> - __compiletime_assert(condition, msg, prefix, suffix)
> -
> -#define compiletime_assert(condition, msg) \
> - _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
> +#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
>
>   #endif

But now it's more compiler dependent, which is worse than before. At 
least it only broke if people fiddled with the optimizations before, 
otherwise it was fine.

Add a configure test for this, use _Static_assert() if it's available, 
the old method if not.

-- 
Jens Axboe

Re: competime_assert failure without -O3 optimization flag expected ?

[Alireza Haghdoost]

On Mon, Jun 29, 2015 at 12:50 PM, Jens Axboe <axboe@kernel.dk> wrote:
> On 06/29/2015 11:42 AM, Alireza Haghdoost wrote:
>>
>> Jens,
>>
>> Sounds like the compiletime_assert() method was not function and is
>> relying on the optimizer performing dead code elimination to remove
>> the call to prefix ## suffix
>>
>> How about this solution: I guess we have to work around it to make it
>> more portable since it works with C11.
>>
>> diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
>> index 40e857c..7c9ba57 100644
>> --- a/fio-arh/compiler/compiler.h
>> +++ b/fio-arh/compiler/compiler.h
>> @@ -33,26 +33,6 @@
>>    1; \
>>   })
>>
>> -#ifndef __compiletime_error
>> -#define __compiletime_error(message)
>> -#endif
>> -#ifndef __compiletime_error_fallback
>> -#define __compiletime_error_fallback(condition) do { } while (0)
>> -#endif
>> -
>> -#define __compiletime_assert(condition, msg, prefix, suffix) \
>> - do { \
>> - int __cond = !(condition); \
>> - extern void prefix ## suffix(void) __compiletime_error(msg); \
>> - if (__cond) \
>> - prefix ## suffix(); \
>> - __compiletime_error_fallback(__cond); \
>> - } while (0)
>> -
>> -#define _compiletime_assert(condition, msg, prefix, suffix) \
>> - __compiletime_assert(condition, msg, prefix, suffix)
>> -
>> -#define compiletime_assert(condition, msg) \
>> - _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>> +#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
>>
>>   #endif
>
>
> But now it's more compiler dependent, which is worse than before. At least
> it only broke if people fiddled with the optimizations before, otherwise it
> was fine.
>
> Add a configure test for this, use _Static_assert() if it's available, the
> old method if not.

Here you are:

diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
index 40e857c..93fdc56 100644
--- a/fio-arh/compiler/compiler.h
+++ b/fio-arh/compiler/compiler.h
@@ -1,5 +1,6 @@
 #ifndef FIO_COMPILER_H
 #define FIO_COMPILER_H
+#include <assert.h>

 #if __GNUC__ >= 4
 #include "compiler-gcc4.h"
@@ -33,6 +34,12 @@
  1; \
 })

+
+#if (__STDC_VERSION__ >= 201112L)
+#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
+#else
+
+
 #ifndef __compiletime_error
 #define __compiletime_error(message)
 #endif
@@ -55,4 +62,7 @@
 #define compiletime_assert(condition, msg) \
  _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)

+
+#endif
+
 #endif

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 12:25 PM, Alireza Haghdoost wrote:
> On Mon, Jun 29, 2015 at 12:50 PM, Jens Axboe <axboe@kernel.dk> wrote:
>> On 06/29/2015 11:42 AM, Alireza Haghdoost wrote:
>>>
>>> Jens,
>>>
>>> Sounds like the compiletime_assert() method was not function and is
>>> relying on the optimizer performing dead code elimination to remove
>>> the call to prefix ## suffix
>>>
>>> How about this solution: I guess we have to work around it to make it
>>> more portable since it works with C11.
>>>
>>> diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
>>> index 40e857c..7c9ba57 100644
>>> --- a/fio-arh/compiler/compiler.h
>>> +++ b/fio-arh/compiler/compiler.h
>>> @@ -33,26 +33,6 @@
>>>     1; \
>>>    })
>>>
>>> -#ifndef __compiletime_error
>>> -#define __compiletime_error(message)
>>> -#endif
>>> -#ifndef __compiletime_error_fallback
>>> -#define __compiletime_error_fallback(condition) do { } while (0)
>>> -#endif
>>> -
>>> -#define __compiletime_assert(condition, msg, prefix, suffix) \
>>> - do { \
>>> - int __cond = !(condition); \
>>> - extern void prefix ## suffix(void) __compiletime_error(msg); \
>>> - if (__cond) \
>>> - prefix ## suffix(); \
>>> - __compiletime_error_fallback(__cond); \
>>> - } while (0)
>>> -
>>> -#define _compiletime_assert(condition, msg, prefix, suffix) \
>>> - __compiletime_assert(condition, msg, prefix, suffix)
>>> -
>>> -#define compiletime_assert(condition, msg) \
>>> - _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>>> +#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
>>>
>>>    #endif
>>
>>
>> But now it's more compiler dependent, which is worse than before. At least
>> it only broke if people fiddled with the optimizations before, otherwise it
>> was fine.
>>
>> Add a configure test for this, use _Static_assert() if it's available, the
>> old method if not.
>
> Here you are:
>
> diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
> index 40e857c..93fdc56 100644
> --- a/fio-arh/compiler/compiler.h
> +++ b/fio-arh/compiler/compiler.h
> @@ -1,5 +1,6 @@
>   #ifndef FIO_COMPILER_H
>   #define FIO_COMPILER_H
> +#include <assert.h>
>
>   #if __GNUC__ >= 4
>   #include "compiler-gcc4.h"
> @@ -33,6 +34,12 @@
>    1; \
>   })
>
> +
> +#if (__STDC_VERSION__ >= 201112L)
> +#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
> +#else
> +
> +
>   #ifndef __compiletime_error
>   #define __compiletime_error(message)
>   #endif
> @@ -55,4 +62,7 @@
>   #define compiletime_assert(condition, msg) \
>    _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
>
> +
> +#endif
> +
>   #endif
>

This is not going to work for earlier compilers, in fact it breaks on 
even gcc 4.9 here. As I said, this needs to be a configure test. That is 
a lot more reliable than this sort of version checking.

-- 
Jens Axboe

Re: competime_assert failure without -O3 optimization flag expected ?

[Alireza Haghdoost]

[-- Attachment #1: Type: text/plain, Size: 1791 bytes --]

Make sens Jens. I did not get what you told me in the first place. Let
me know if this would fix it. It works for me with GCC 4.6 (patch file
is also attached):

diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
index 40e857c..c9f8776 100644
--- a/fio-arh/compiler/compiler.h
+++ b/fio-arh/compiler/compiler.h
@@ -1,5 +1,6 @@
 #ifndef FIO_COMPILER_H
 #define FIO_COMPILER_H
+#include <assert.h>

 #if __GNUC__ >= 4
 #include "compiler-gcc4.h"
@@ -33,6 +34,11 @@
  1; \
 })

+
+#if defined(CONFIG_STATIC_ASSERT)
+#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
+
+#else
 #ifndef __compiletime_error
 #define __compiletime_error(message)
 #endif
@@ -55,4 +61,6 @@
 #define compiletime_assert(condition, msg) \
  _compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)

+#endif
+
 #endif
diff --git a/fio-arh/configure b/fio-arh/configure
index e459d63..ab85936 100755
--- a/fio-arh/configure
+++ b/fio-arh/configure
@@ -1492,6 +1492,21 @@ if compile_prog "" "" "getmntinfo"; then
 fi
 echo "getmntinfo                    $getmntinfo"

+##########################################
+# Check whether we have _Static_assert
+static_assert="no"
+cat > $TMPC << EOF
+#include <assert.h>
+int main(int argc, char **argv)
+{
+  _Static_assert( 1 == 1 , "Check");
+  return 0 ;
+}
+EOF
+if compile_prog "" "" "static_assert"; then
+    static_assert="yes"
+fi
+echo "Static Assert                 $static_assert"
 #############################################################################

 if test "$wordsize" = "64" ; then
@@ -1671,6 +1686,9 @@ fi
 if test "$getmntinfo" = "yes" ; then
   output_sym "CONFIG_GETMNTINFO"
 fi
+if test "$static_assert" = "yes" ; then
+  output_sym "CONFIG_STATIC_ASSERT"
+fi

 if test "$zlib" = "no" ; then

[-- Attachment #2: patch --]
[-- Type: application/octet-stream, Size: 2828 bytes --]

diff --git a/fio-arh/compiler/compiler.h b/fio-arh/compiler/compiler.h
index 40e857c..c9f8776 100644
--- a/fio-arh/compiler/compiler.h
+++ b/fio-arh/compiler/compiler.h
@@ -1,5 +1,6 @@
 #ifndef FIO_COMPILER_H
 #define FIO_COMPILER_H
+#include <assert.h>
 
 #if __GNUC__ >= 4
 #include "compiler-gcc4.h"
@@ -33,6 +34,11 @@
 	1; \
 })
 
+
+#if defined(CONFIG_STATIC_ASSERT)
+#define compiletime_assert(condition, msg) _Static_assert(condition, msg)
+
+#else
 #ifndef __compiletime_error
 #define __compiletime_error(message)
 #endif
@@ -55,4 +61,6 @@
 #define compiletime_assert(condition, msg) \
 	_compiletime_assert(condition, msg, __compiletime_assert_, __LINE__)
 
+#endif 
+
 #endif
diff --git a/fio-arh/configure b/fio-arh/configure
index e459d63..ab85936 100755
--- a/fio-arh/configure
+++ b/fio-arh/configure
@@ -1492,6 +1492,21 @@ if compile_prog "" "" "getmntinfo"; then
 fi
 echo "getmntinfo                    $getmntinfo"
 
+##########################################
+# Check whether we have _Static_assert
+static_assert="no"
+cat > $TMPC << EOF
+#include <assert.h>
+int main(int argc, char **argv)
+{
+  _Static_assert( 1 == 1 , "Check");
+  return 0 ;
+}
+EOF
+if compile_prog "" "" "static_assert"; then
+    static_assert="yes"
+fi
+echo "Static Assert                 $static_assert"
 #############################################################################
 
 if test "$wordsize" = "64" ; then
@@ -1671,6 +1686,9 @@ fi
 if test "$getmntinfo" = "yes" ; then
   output_sym "CONFIG_GETMNTINFO"
 fi
+if test "$static_assert" = "yes" ; then
+  output_sym "CONFIG_STATIC_ASSERT"
+fi
 
 if test "$zlib" = "no" ; then
   echo "Consider installing zlib-dev (zlib-devel), some fio features depend on it."
diff --git a/fio-arh/t/axmap b/fio-arh/t/axmap
deleted file mode 100755
index 2facabe..0000000
Binary files a/fio-arh/t/axmap and /dev/null differ
diff --git a/fio-arh/t/fio-btrace2fio b/fio-arh/t/fio-btrace2fio
deleted file mode 100755
index 13addba..0000000
Binary files a/fio-arh/t/fio-btrace2fio and /dev/null differ
diff --git a/fio-arh/t/fio-dedupe b/fio-arh/t/fio-dedupe
deleted file mode 100755
index 2570081..0000000
Binary files a/fio-arh/t/fio-dedupe and /dev/null differ
diff --git a/fio-arh/t/fio-genzipf b/fio-arh/t/fio-genzipf
deleted file mode 100755
index d133a8a..0000000
Binary files a/fio-arh/t/fio-genzipf and /dev/null differ
diff --git a/fio-arh/t/ieee754 b/fio-arh/t/ieee754
deleted file mode 100755
index 3c7e924..0000000
Binary files a/fio-arh/t/ieee754 and /dev/null differ
diff --git a/fio-arh/t/lfsr-test b/fio-arh/t/lfsr-test
deleted file mode 100755
index 602ea8d..0000000
Binary files a/fio-arh/t/lfsr-test and /dev/null differ
diff --git a/fio-arh/t/stest b/fio-arh/t/stest
deleted file mode 100755
index f4743f3..0000000
Binary files a/fio-arh/t/stest and /dev/null differ

Re: competime_assert failure without -O3 optimization flag expected ?

[Jens Axboe]

On 06/29/2015 12:58 PM, Alireza Haghdoost wrote:
> Make sens Jens. I did not get what you told me in the first place. Let
> me know if this would fix it. It works for me with GCC 4.6 (patch file
> is also attached):

This is perfect! Thanks, committed:

http://git.kernel.dk/cgit/fio/commit/?id=5018f79f6d8da82b6fafbbeeebdecb3799788bc3

-- 
Jens Axboe