Re: Experiencing Segfault in CryptoPP::StreamTransformationFilter in Linux build

I don't even understand the code - but based on your symptoms (works until you increase the level of compiler optimization), I'd say it's a compiler bug. More so, a bug in a compiler that's most likely been fixed a long time ago.

Thus, incentive to mess with something that long lost its relevance is pretty low. 

Feel free to create and post a reproducer - a small example that compiles and demonstrates the problem, if you want to pursue this. Or just upgrade the compiler (should be done regardless - your development is insecure).

> On Apr 18, 2021, at 23:07, Sunandan Nandi <[email protected]> wrote:
> 
> 
> Hi Uri,
> 
> Thanks.OS version and clang version is project binding right now. But certainly, will check.
> 
> Do you see any obvious problem in the code snippet I have added in the above post?
> 
> Regards,
> Sunandan
> 
>> On Sun, Apr 18, 2021 at 5:45 PM Uri Blumenthal <[email protected]> wrote:
>> Considering that the current stable version of Clang is 11, would or not make sense to upgrade your compiler?
>> 
>> Not to mention that your OS probably does not get any security patches now - so, perhaps, moving from 18.04LTS to 20.04 LTS would be a good idea?
>> 
>> I regularly build with Clang and GCC, no crashes.
>> 
>>>> On Apr 18, 2021, at 08:02, Sunandan Nandi <[email protected]> wrote:
>>>> 
>>> Hi,
>>> 
>>> I am experiencing Segfault inside CryptoPP::StreamTransformationFilter.
>>> 
>>> Overview:
>>> I have downloaded and built CryptoPP as an external CMake module. 
>>> Problem:
>>> This segfault is happening only in Linux(Ubuntu 18.04.05LTS) Release build.
>>> 64-bit,compiler clang 6.0
>>> 
>>> GDB BT
>>> Thread 9 "***" received signal SIGSEGV, Segmentation fault.
>>> [Switching to Thread 0x7fffdb7fe700 (LWP 5835)]
>>> 0x0000000000000000 in ?? ()
>>> (gdb) bt
>>> #0  0x0000000000000000 in ?? ()
>>> #1  0x0000000000b7559d in CryptoPP::StreamTransformationFilter::StreamTransformationFilter (this=0x7fffdb7fd1c0, c=..., attachment=<optimized out>,
>>>     padding=<optimized out>) at filters.cpp:593
>>> #2  0x00000000004ff4fd in ***::AESEncryptDecryptUtility::decrypt(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&) ()
>>> 
>>> Trials to resolve this issue:
>>> If I build my target where I am using CryptoPP( static lib) with optimization level O1 or less, then no segfault observed. But this issue is with optimization level O2 or more.
>>> 
>>> Code snippet:
>>> std::string AESEncryptDecryptUtility::decrypt(const std::string& encrypted_text) { std::string decrypted_text; CryptoPP::AES::Decryption aesDecryption(key_.data(), KEY_LENGTH); CryptoPP::CBC_Mode_ExternalCipher::Decryption cbcDecryption(aesDecryption, iv_.data()); CryptoPP::StreamTransformationFilter stfDecryptor(cbcDecryption, new CryptoPP::StringSink(decrypted_text)); stfDecryptor.Put(reinterpret_cast<const unsigned char*>(encrypted_text.c_str()), encrypted_text.size()); stfDecryptor.MessageEnd(); return decrypted_text; 
>>> 
>>> Inside StreamTransformationFilter() where it's crashing, the line highlighted in Red is pointing in GDB as the last point before the crash(segfault)
>>> 
>>> StreamTransformationFilter::StreamTransformationFilter(StreamTransformation &c, BufferedTransformation *attachment, BlockPaddingScheme padding)
>>>     : FilterWithBufferedInput(attachment), m_cipher(c), m_padding(DEFAULT_PADDING)
>>> {
>>>     CRYPTOPP_ASSERT(c.MinLastBlockSize() == 0 || c.MinLastBlockSize() > c.MandatoryBlockSize());
>>> 
>>>     const bool authenticatedFilter = dynamic_cast<AuthenticatedSymmetricCipher *>(&c) != NULLPTR;
>>>     if (authenticatedFilter)
>>>         throw InvalidArgument("StreamTransformationFilter: please use AuthenticatedEncryptionFilter and AuthenticatedDecryptionFilter for AuthenticatedSymmetricCipher");
>>> 
>>>     // InitializeDerivedAndReturnNewSizes may override some of these
>>>     m_mandatoryBlockSize = m_cipher.MandatoryBlockSize();
>>> 
>>> Please let me know if any other information needed.
>>> 
>>> Regards,
>>> Sunandan
>>> 
>>>  
>>> -- 
>>> You received this message because you are subscribed to the Google Groups "Crypto++ Users" group.
>>> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
>>> To view this discussion on the web visit https://groups.google.com/d/msgid/cryptopp-users/e6c860f6-1e74-4372-b4de-691e0692747bn%40googlegroups.com.
>> 
>> -- 
>> You received this message because you are subscribed to the Google Groups "Crypto++ Users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
>> To view this discussion on the web visit https://groups.google.com/d/msgid/cryptopp-users/17A2AD5B-284E-406B-8C8E-EAB3F80F85B5%40gmail.com.
> 
> -- 
> You received this message because you are subscribed to the Google Groups "Crypto++ Users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To view this discussion on the web visit https://groups.google.com/d/msgid/cryptopp-users/CAKPYA22Gv%3Dr_HNg3j3Dom_E4yRcv6eH6sDX%2B2qbRrkT9vT4W1A%40mail.gmail.com.

-- 
You received this message because you are subscribed to the Google Groups "Crypto++ Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
To view this discussion on the web visit https://groups.google.com/d/msgid/cryptopp-users/432B0DA5-BC99-4F43-887C-3C8BA578B9D0%40gmail.com.