[llvm] r216268 - [Support] Fix the overflow bug in ULEB128 decoding.

[David Blaikie]

So this scares me a bit - not because this fix isn't right (it looks
right to me) but because this isn't the only code that does (U)LEB
conversions. I think there's a couple of copies in debug info code...

Could you check for them? Or I can try to dredge them up & take a glance.

On Fri, Aug 22, 2014 at 9:29 AM, Alex Lorenz <arphaman at gmail.com> wrote:
> Author: arphaman
> Date: Fri Aug 22 11:29:45 2014
> New Revision: 216268
>
> URL: http://llvm.org/viewvc/llvm-project?rev=216268&view=rev
> Log:
> [Support] Fix the overflow bug in ULEB128 decoding.
>
> Differential Revision: http://reviews.llvm.org/D5029
>
> Modified:
>     llvm/trunk/include/llvm/Support/LEB128.h
>     llvm/trunk/unittests/Support/LEB128Test.cpp
>
> Modified: llvm/trunk/include/llvm/Support/LEB128.h
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/include/llvm/Support/LEB128.h?rev=216268&r1=216267&r2=216268&view=diff
> ==============================================================================
> --- llvm/trunk/include/llvm/Support/LEB128.h (original)
> +++ llvm/trunk/include/llvm/Support/LEB128.h Fri Aug 22 11:29:45 2014
> @@ -82,7 +82,7 @@ inline uint64_t decodeULEB128(const uint
>    uint64_t Value = 0;
>    unsigned Shift = 0;
>    do {
> -    Value += (*p & 0x7f) << Shift;
> +    Value += uint64_t(*p & 0x7f) << Shift;
>      Shift += 7;
>    } while (*p++ >= 128);
>    if (n)
>
> Modified: llvm/trunk/unittests/Support/LEB128Test.cpp
> URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/unittests/Support/LEB128Test.cpp?rev=216268&r1=216267&r2=216268&view=diff
> ==============================================================================
> --- llvm/trunk/unittests/Support/LEB128Test.cpp (original)
> +++ llvm/trunk/unittests/Support/LEB128Test.cpp Fri Aug 22 11:29:45 2014
> @@ -106,6 +106,7 @@ TEST(LEB128Test, DecodeULEB128) {
>    EXPECT_DECODE_ULEB128_EQ(0xffu, "\xff\x01");
>    EXPECT_DECODE_ULEB128_EQ(0x100u, "\x80\x02");
>    EXPECT_DECODE_ULEB128_EQ(0x101u, "\x81\x02");
> +  EXPECT_DECODE_ULEB128_EQ(4294975616ULL, "\x80\xc1\x80\x80\x10");
>
>    // Decode ULEB128 with extra padding bytes
>    EXPECT_DECODE_ULEB128_EQ(0u, "\x80\x00");
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits