[
https://issues.apache.org/jira/browse/THRIFT-3821?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16122756#comment-16122756
]
ASF GitHub Bot commented on THRIFT-3821:
----------------------------------------
Github user jeking3 commented on a diff in the pull request:
https://github.com/apache/thrift/pull/1326#discussion_r132611133
--- Diff: lib/cpp/test/TMemoryBufferTest.cpp ---
@@ -117,4 +117,17 @@ BOOST_AUTO_TEST_CASE(test_exceptions) {
BOOST_CHECK_NO_THROW(buf2.write((const uint8_t*)"bar", 3));
}
+#ifndef _WIN32
+// We can't allocate 1 GB of memory in 32-bit environments.
--- End diff --
This makes me wonder if the TBufferTransport should have a size limit that
is configurable, with a default of INT32_MAX, and then the test can make a
smaller one like 4KB, and write 4KB and then one byte more, instead of using up
2GB of memory.
> TMemoryBuffer buffer may overflow when resizing
> -----------------------------------------------
>
> Key: THRIFT-3821
> URL: https://issues.apache.org/jira/browse/THRIFT-3821
> Project: Thrift
> Issue Type: Bug
> Components: C++ - Library
> Affects Versions: 0.9.3
> Reporter: Huaisi Xu
>
> In ensurecanwrite():
> {code}
> uint32_t new_size = bufferSize_;
> while (len > avail) {
> new_size = new_size > 0 ? new_size * 2 : 1;
> avail = available_write() + (new_size - bufferSize_);
> }
> // Allocate into a new pointer so we don't bork ours if it fails.
> uint8_t* new_buffer = static_cast<uint8_t*>(std::realloc(buffer_,
> new_size));
> if (new_buffer == NULL) {
> throw std::bad_alloc();
> }
> rBase_ = new_buffer + (rBase_ - buffer_);
> rBound_ = new_buffer + (rBound_ - buffer_);
> wBase_ = new_buffer + (wBase_ - buffer_);
> wBound_ = new_buffer + new_size;
> buffer_ = new_buffer;
> bufferSize_ = new_size;
> {code}
> If old bufferSize_ is lager than 2gb, then calculating new size will overflow.
> i.e. if bufferSize_ = 3355443200, then new buffer size will be 2415919104,
> which is less than old size.
> However,
> {code}
> avail = available_write() + (new_size - bufferSize_)
> {code}
> overflows again, so we will end up with an shrinked buffer.
> What is worse is that after
> {code}
> wBase_ = new_buffer + (wBase_ - buffer_);
> wBound_ = new_buffer + new_size;
> {code}
> wBase_ stays the same, but wBound_ becomes lower than it. What happens next
> is that uint32_t avail = available_write() may overflow every time
> subsequently. and thrift writes to unknown memory.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
