[
https://issues.apache.org/jira/browse/STDCXX-841?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12588028#action_12588028
]
Eric Lemings commented on STDCXX-841:
-------------------------------------
Some sort of arithmetic overflow evaluating this test:
{noformat}
file stdcxx/trunk/tests/utilities/20.temp.buffer.cpp:
...
459 test_failure ((BigStruct<MAX_SIZE / 2>*)0, 0);
...
{noformat}
The terminating condition of the loop in test_failure() is not catching it:
{noformat}
file stdcxx/trunk/tests/utilities/20.temp.buffer.cpp:
...
267 for (std::size_t i = 2; i < sizeof (T); i <<= 1) {
268 // exercise arithmetic overflow (not to be confused
269 // with the out-of-memory tests below)
270
271 // attempts to allocate space for an array of elements
272 // with a total size that exceeds SIZE_MAX must fail
273 nelems = _RWSTD_PTRDIFF_MAX / i + 1;
274
275 pa = std::get_temporary_buffer<T>(nelems);
276
277 rw_assert (0 == pa.first && 0 == pa.second, 0, __LINE__,
278 "get_temporary_buffer<%s>(%ld) == { 0, 0 }, "
279 "got { %#p, %td }",
280 tname, nelems, pa.first, pa.second);
281 }
...
{noformat}
Here's what I observed cycling near what should be the end of the loop:
{noformat}
...
(dbx) cont
[EMAIL PROTECTED] ([EMAIL PROTECTED]) stopped in
test_failure<BigStruct<4611686018427387903UL> > at line 273 in file
"20.temp.buffer.cpp"
273 nelems = _RWSTD_PTRDIFF_MAX / i + 1;
i = 9223372036854775808U
(dbx) cont
[EMAIL PROTECTED] ([EMAIL PROTECTED]) stopped in
test_failure<BigStruct<4611686018427387903UL> > at line 273 in file
"20.temp.buffer.cpp"
273 nelems = _RWSTD_PTRDIFF_MAX / i + 1;
i = 0
{noformat}
> [Sun C++] SIGFPE in 20.temp.buffer in LP64
> ------------------------------------------
>
> Key: STDCXX-841
> URL: https://issues.apache.org/jira/browse/STDCXX-841
> Project: C++ Standard Library
> Issue Type: Bug
> Components: Tests
> Affects Versions: 4.2.0
> Environment: Sun C++
> Reporter: Martin Sebor
> Assignee: Eric Lemings
> Priority: Minor
> Fix For: 4.2.1
>
> Original Estimate: 2h
> Remaining Estimate: 2h
>
> When compiled with Sun C++ (any version) in LP64 mode, the test
> [20.temp.buffer.cpp|http://svn.apache.org/viewvc/stdcxx/trunk/tests/utilities/20.temp.buffer.cpp?view=markup]
> abends with SIGFPE.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
