svn commit: r253380 - head/lib/libc/stdlib

[Andriy Gapon]

on 17/07/2013 08:27 Andrey Chernov said the following:
> On 17.07.2013 8:48, Andrey Chernov wrote:
>> On 17.07.2013 8:10, Andrey Chernov wrote:
>>> On 16.07.2013 11:26, Andriy Gapon wrote:
>>>> Modified: head/lib/libc/stdlib/getenv.c
>>>> ==============================================================================
>>>> --- head/lib/libc/stdlib/getenv.c	Tue Jul 16 06:50:22 2013	(r253379)
>>>> +++ head/lib/libc/stdlib/getenv.c	Tue Jul 16 07:26:46 2013	(r253380)
>>>> @@ -505,7 +505,7 @@ __setenv(const char *name, size_t nameLe
>>>>  		envVars[envNdx].valueSize = valueLen;
>>>>  
>>>>  		/* Save name of name/value pair. */
>>>> -		env = stpcpy(envVars[envNdx].name, name);
>>>> +		env = stpncpy(envVars[envNdx].name, name, nameLen);
>>>>  		if ((envVars[envNdx].name)[nameLen] != '=')
>>>>  			env = stpcpy(env, "=");
>>>>  	}
>>>>
>>>
>>> I am not sure what you are trying to fix, but you just made next line
>>> condition unpredictable random, since (envVars[envNdx].name)[nameLen] is
>>> never filled now and there is freshly malloced memory content, which is
>>> picked for != '=' comparison.
>>>
>>> Please back it out or fix.
>>
>> If I understand that right, correct version will be:
>>
>> env = stpncpy(envVars[envNdx].name, name, nameLen);
>> env = stpcpy(env, "=");
>>
>> Condition check is not needed.
> 
> Microoptimized:
> 
> env = stpncpy(envVars[envNdx].name, name, nameLen);
> *env++ = '=';
> 

In fact, I think that the currently committed code is not functionally broken,
just weird.
It is equivalent to:

x = unitialized_value;
if (x != X)
	x = X;
/*
else
	assert(x == X);
 */

which should be completely equivalent to just

x = X;

despite x starting out with random junk.

-- 
Andriy Gapon