liblzma: Avoid null pointer + 0 (undefined behavior in C).

In the C99 and C17 standards, section 6.5.6 paragraph 8 means that
adding 0 to a null pointer is undefined behavior. As of writing,
"clang -fsanitize=undefined" (Clang 15) diagnoses this. However,
I'm not aware of any compiler that would take advantage of this
when optimizing (Clang 15 included). It's good to avoid this anyway
since compilers might some day infer that pointer arithmetic implies
that the pointer is not NULL. That is, the following foo() would then
unconditionally return 0, even for foo(NULL, 0):

    void bar(char *a, char *b);

    int foo(char *a, size_t n)
    {
        bar(a, a + n);
        return a == NULL;
    }

In contrast to C, C++ explicitly allows null pointer + 0. So if
the above is compiled as C++ then there is no undefined behavior
in the foo(NULL, 0) call.

To me it seems that changing the C standard would be the sane
thing to do (just add one sentence) as it would ensure that a huge
amount of old code won't break in the future. Based on web searches
it seems that a large number of codebases (where null pointer + 0
occurs) are being fixed instead to be future-proof in case compilers
will some day optimize based on it (like making the above foo(NULL, 0)
return 0) which in the worst case will cause security bugs.

Some projects don't plan to change it. For example, gnulib and thus
many GNU tools currently require that null pointer + 0 is defined:

    https://lists.gnu.org/archive/html/bug-gnulib/2021-11/msg00000.html

    https://www.gnu.org/software/gnulib/manual/html_node/Other-portability-assumptions.html

In XZ Utils null pointer + 0 issue should be fixed after this
commit. This adds a few if-statements and thus branches to avoid
null pointer + 0. These check for size > 0 instead of ptr != NULL
because this way bugs where size > 0 && ptr == NULL will likely
get caught quickly. None of them are in hot spots so it shouldn't
matter for performance.

A little less readable version would be replacing

    ptr + offset

with

    offset != 0 ? ptr + offset : ptr

or creating a macro for it:

    #define my_ptr_add(ptr, offset) \
            ((offset) != 0 ? ((ptr) + (offset)) : (ptr))

Checking for offset != 0 instead of ptr != NULL allows GCC >= 8.1,
Clang >= 7, and Clang-based ICX to optimize it to the very same code
as ptr + offset. That is, it won't create a branch. So for hot code
this could be a good solution to avoid null pointer + 0. Unfortunately
other compilers like ICC 2021 or MSVC 19.33 (VS2022) will create a
branch from my_ptr_add().

Thanks to Marcin Kowalczyk for reporting the problem:
https://github.com/tukaani-project/xz/issues/36
This commit is contained in:
Lasse Collin 2023-02-21 22:57:10 +02:00
parent d9445b5b2d
commit 09363bea46
9 changed files with 72 additions and 22 deletions

View File

@ -123,7 +123,10 @@ block_decode(void *coder_ptr, const lzma_allocator *allocator,
return LZMA_DATA_ERROR;
}
if (!coder->ignore_check)
// Don't waste time updating the integrity check if it will be
// ignored. Also skip it if no new output was produced. This
// avoids null pointer + 0 (undefined behavior) when out == 0.
if (!coder->ignore_check && out_used > 0)
lzma_check_update(&coder->check, coder->block->check,
out + out_start, out_used);

View File

@ -77,8 +77,11 @@ block_encode(void *coder_ptr, const lzma_allocator *allocator,
// checked it at the beginning of this function.
coder->uncompressed_size += in_used;
lzma_check_update(&coder->check, coder->block->check,
in + in_start, in_used);
// Call lzma_check_update() only if input was consumed. This
// avoids null pointer + 0 (undefined behavior) when in == 0.
if (in_used > 0)
lzma_check_update(&coder->check, coder->block->check,
in + in_start, in_used);
if (ret != LZMA_STREAM_END || action == LZMA_SYNC_FLUSH)
return ret;

View File

@ -289,13 +289,21 @@ lzma_code(lzma_stream *strm, lzma_action action)
strm->next_in, &in_pos, strm->avail_in,
strm->next_out, &out_pos, strm->avail_out, action);
strm->next_in += in_pos;
strm->avail_in -= in_pos;
strm->total_in += in_pos;
// Updating next_in and next_out has to be skipped when they are NULL
// to avoid null pointer + 0 (undefined behavior). Do this by checking
// in_pos > 0 and out_pos > 0 because this way NULL + non-zero (a bug)
// will get caught one way or other.
if (in_pos > 0) {
strm->next_in += in_pos;
strm->avail_in -= in_pos;
strm->total_in += in_pos;
}
strm->next_out += out_pos;
strm->avail_out -= out_pos;
strm->total_out += out_pos;
if (out_pos > 0) {
strm->next_out += out_pos;
strm->avail_out -= out_pos;
strm->total_out += out_pos;
}
strm->internal->avail_in = strm->avail_in;

View File

@ -200,9 +200,16 @@ index_decode(void *coder_ptr, const lzma_allocator *allocator,
}
out:
// Update the CRC32,
coder->crc32 = lzma_crc32(in + in_start,
*in_pos - in_start, coder->crc32);
// Update the CRC32.
//
// Avoid null pointer + 0 (undefined behavior) in "in + in_start".
// In such a case we had no input and thus in_used == 0.
{
const size_t in_used = *in_pos - in_start;
if (in_used > 0)
coder->crc32 = lzma_crc32(in + in_start,
in_used, coder->crc32);
}
return ret;
}

View File

@ -153,8 +153,15 @@ index_encode(void *coder_ptr,
out:
// Update the CRC32.
coder->crc32 = lzma_crc32(out + out_start,
*out_pos - out_start, coder->crc32);
//
// Avoid null pointer + 0 (undefined behavior) in "out + out_start".
// In such a case we had no input and thus out_used == 0.
{
const size_t out_used = *out_pos - out_start;
if (out_used > 0)
coder->crc32 = lzma_crc32(out + out_start,
out_used, coder->crc32);
}
return ret;
}

View File

@ -325,9 +325,16 @@ lzma_index_hash_decode(lzma_index_hash *index_hash, const uint8_t *in,
}
out:
// Update the CRC32,
index_hash->crc32 = lzma_crc32(in + in_start,
*in_pos - in_start, index_hash->crc32);
// Update the CRC32.
//
// Avoid null pointer + 0 (undefined behavior) in "in + in_start".
// In such a case we had no input and thus in_used == 0.
{
const size_t in_used = *in_pos - in_start;
if (in_used > 0)
index_hash->crc32 = lzma_crc32(in + in_start,
in_used, index_hash->crc32);
}
return ret;
}

View File

@ -42,7 +42,12 @@ delta_decode(void *coder_ptr, const lzma_allocator *allocator,
in, in_pos, in_size, out, out_pos, out_size,
action);
decode_buffer(coder, out + out_start, *out_pos - out_start);
// out might be NULL. In that case size == 0. Null pointer + 0 is
// undefined behavior so skip the call in that case as it would
// do nothing anyway.
const size_t size = *out_pos - out_start;
if (size > 0)
decode_buffer(coder, out + out_start, size);
return ret;
}

View File

@ -63,7 +63,12 @@ delta_encode(void *coder_ptr, const lzma_allocator *allocator,
const size_t out_avail = out_size - *out_pos;
const size_t size = my_min(in_avail, out_avail);
copy_and_encode(coder, in + *in_pos, out + *out_pos, size);
// in and out might be NULL. In such cases size == 0.
// Null pointer + 0 is undefined behavior so skip
// the call in that case as it would do nothing anyway.
if (size > 0)
copy_and_encode(coder, in + *in_pos, out + *out_pos,
size);
*in_pos += size;
*out_pos += size;
@ -78,7 +83,10 @@ delta_encode(void *coder_ptr, const lzma_allocator *allocator,
in, in_pos, in_size, out, out_pos, out_size,
action);
encode_in_place(coder, out + out_start, *out_pos - out_start);
// Like above, avoid null pointer + 0.
const size_t size = *out_pos - out_start;
if (size > 0)
encode_in_place(coder, out + out_start, size);
}
return ret;

View File

@ -139,9 +139,11 @@ simple_code(void *coder_ptr, const lzma_allocator *allocator,
return ret;
}
// Filter out[].
// Filter out[] unless there is nothing to filter.
// This way we avoid null pointer + 0 (undefined behavior)
// when out == NULL.
const size_t size = *out_pos - out_start;
const size_t filtered = call_filter(
const size_t filtered = size == 0 ? 0 : call_filter(
coder, out + out_start, size);
const size_t unfiltered = size - filtered;