libstdc++/9246: Bad handling of nul characters in codecvt<wchar_t, char, mbstate_t>

libstdc++/9246: Bad handling of nul characters in codecvt<wchar_t, char, mbstate_t>

[peturr02]

>Number:         9246
>Category:       libstdc++
>Synopsis:       Bad handling of nul characters in codecvt<wchar_t, char, mbstate_t>
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Jan 09 07:56:03 PST 2003
>Closed-Date:
>Last-Modified:
>Originator:     peturr02@ru.is
>Release:        gcc-3.2.1
>Organization:
>Environment:
Red Hat Linux 8.0
>Description:
If the input buffer for codecvt<wchar_t, char, mbstate_t>::do_out contains nul characters, from_next is not set properly and the function may return partial even though there is plenty of space left in the output buffer.
>How-To-Repeat:
See attachment.
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: text/plain; name="nulbug.cc"
Content-Disposition: inline; filename="nulbug.cc"

#include <locale>
#include <cwchar>
#include <cassert>
#include <cstring>

void test01()
{
	using namespace std;
	typedef codecvt<wchar_t, char, mbstate_t> w_codecvt;

	const wchar_t wnul[] = { L'\0' };
	const wchar_t* from_next;

	char buffer[1024];
	char* to_next;

	locale loc;
	const w_codecvt& cvt = use_facet<w_codecvt>(loc);
	
	mbstate_t state;
	memset(&state, 0, sizeof(state));
	
	codecvt_base::result res =
		cvt.out(state, wnul, wnul + 1, from_next, buffer,
			buffer + sizeof(buffer), to_next);
	
	assert(res != codecvt_base::noconv);
	
	if (res == codecvt_base::ok)
	{
		assert(from_next == wnul + 1);
		assert(to_next > buffer);
		assert(to_next <= buffer + sizeof(buffer));
	}
	else if (res == codecvt_base::partial)
	{
		assert(to_next == buffer);
		assert(from_next == wnul);
	}
	else if (res == codecvt_base::error)
	{
		assert(to_next == buffer);
		assert(from_next == wnul);
	}
}

void test02()
{
	using namespace std;
	typedef codecvt<wchar_t, char, mbstate_t> w_codecvt;

	const wchar_t wstr[] = { L'a', L'\0' };
	const wchar_t* from_next;

	char buffer[1024];
	char* to_next;

	locale loc;
	const w_codecvt& cvt = use_facet<w_codecvt>(loc);
	
	mbstate_t state;
	memset(&state, 0, sizeof(state));
	
	codecvt_base::result res =
		cvt.out(state, wstr, wstr + 2, from_next, buffer,
			buffer + sizeof(buffer), to_next);
	
	assert(res != codecvt_base::noconv);
	
	if (res == codecvt_base::ok)
	{
		assert(from_next == wstr + 2);
		assert(to_next > buffer);
		assert(to_next <= buffer + sizeof(buffer));
	}
	else if (res == codecvt_base::partial)
	{
		assert(to_next > buffer);
		assert(to_next <= buffer + sizeof(buffer));
		assert(from_next >= wstr);
		assert(from_next < wstr + 2);
	}
	else if (res == codecvt_base::error)
	{
		assert(to_next > buffer);
		assert(to_next <= buffer + sizeof(buffer));
		assert(from_next >= wstr);
		assert(from_next < wstr + 2);
	}
}

int main()
{
	//test01();
	test02();

	return 0;
}