[lldb-dev] UnicodeDecodeError for serialize SBValue description

Enrico Granata via lldb-dev lldb-dev at lists.llvm.org
Thu Apr 7 10:29:13 PDT 2016


> On Apr 7, 2016, at 9:51 AM, Jim Ingham <jingham at apple.com> wrote:
> 
> I don't think Enrico was suggesting that we maintain a bunch of third party data formatters in the lldb source base.  

That depends - if this std::string implementation is part of a publicly available STL implementation, it might make sense for us to “know about it” out of the box in the same way we know about libstdc++ and libc++
If it is an internal-only string class, then, yes, I would definitely not suggest putting this inside the LLDB core

> He was giving C++ examples (using the lldb_private API's) because the STL formatters are in C++, so that's what he had on hand to demonstrate the kinds of algorithms you would use to dig into these complex structures.  For the most part the lldb_private API's used in Enrico's examples are mirrored in the SB API's pretty directly, so this isn't a terrible source for examples.
> 
> Note, it used to be possible to write C++ based data formatters, build them in a shared library and load them with the "plugin load" command.  These have the advantage of working on systems that don't support Python.  Not sure what the state of that is these days, however.

It might or might not work. If it didn’t work and somebody wanted to fix that, I suspect we would gladly accept their patches.

>  But even if you were going to write C++ formatters you'd be better off using the SB API's not the lldb_private API's since then your plugins would have a longer useful life-cycle.
> 
> Jim
> 
> 
>> On Apr 7, 2016, at 2:45 AM, Tamas Berghammer via lldb-dev <lldb-dev at lists.llvm.org> wrote:
>> 
>> LLDB supports adding data formatters without modifying the source code and I would strongly prefer to go that way as we don't want each user of LLDB to start adding data formatters to their own custom types. We have a pretty detailed (but possible a bit outdated) description about how they work and how you can add a new one here: http://lldb.llvm.org/varformats.html
>> 
>> Enrico: Is there any reason you suggested the data formatters written inside LLDB over the python based ones?
>> 
>> On Thu, Apr 7, 2016 at 3:31 AM Jeffrey Tan via lldb-dev <lldb-dev at lists.llvm.org> wrote:
>> Thanks Enrico. This is very detailed! I will take a look. 
>> Btw: originally, I was hoping that data formatter can be added without changing the source code. Like giving a xml/json format file telling lldb the memory layout/structure of the data structure, lldb can parse the xml/json and deduce the formatting. This is approach used by data visualizer in VS debugger: https://msdn.microsoft.com/en-us/library/jj620914.aspx
>> This will make adding data formatter more extensible/flexible. Any reason we did not take this approach? 
>> 
>> Jeffrey
>> 
>> On Wed, Apr 6, 2016 at 11:49 AM, Enrico Granata <egranata at apple.com> wrote:
>> 
>>> On Apr 5, 2016, at 2:42 PM, Jeffrey Tan <jeffrey.fudan at gmail.com> wrote:
>>> 
>>> Hi Enrico,
>>> 
>>> Any suggestion/example how to add a data formatter for our own STL string? From the output below I can see we are using our own "fbstring_core" which I assume I need to write a type summary for this type:
>>> 
>>> frame variable corpus -T
>>> (const string &const) corpus = error: summary string parsing error: {
>>>  (std::fbstring_core<char>) store_ = {
>>>    (std::fbstring_core<char>::(anonymous union))  = {
>>>      (char [24]) small_ = "www"
>>>      (std::fbstring_core<char>::MediumLarge) ml_ = {
>>>        (char *) data_ = 0x0000000000777777 "H\x89U\xa8H\x89M\xa0L\x89E\x98H\x8bE\xa8H\x89��_U��D\x88e�H\x8bE\xa0H\x89��]U��H\x89�H\x8dE�H\x89�H\x89��� ��L\x8dm�H\x8bE\x98H\x89��IU��\x88]�L\x8be\xb0L\x89��
>>>        (std::size_t) size_ = 0
>>>        (std::size_t) capacity_ = 1441151880758558720
>>>      }
>>>    }
>>>  }
>>> }
>>> 
>> 
>> Admittedly, this is going to be a little vague since I haven’t really seen your code and I am only working off of one sample
>> 
>> There’s going to be two parts to getting this to work:
>> 
>> Part 1 - Formatting fbstring_core
>> 
>> At a glance, an fbstring_core<char> can be backed by two representations. A “small” representation (a char array), and a “medium/large" representation (a char* + a size)
>> I assume that the way you tell one from the other is
>> 
>> if (size == 0) small
>> else medium-large
>> 
>> If my assumption is not correct, you’ll need to discover what the correct discriminator logic is - the class has to know, and so do you :-)
>> 
>> Armed with that knowledge, look in lldb source/Plugins/Language/CPlusPlus/Formatters/LibCxx.cpp
>> There’s a bunch of code that deals with formatting llvm’s libc++ std::string - which follows a very similar logic to your class
>> 
>> ExtractLibcxxStringInfo() is the function that handles discovering which layout the string uses - where the data lives - and how much data there is
>> 
>> Once you have told yourself how much data there is (the size) and where it lives (array or pointer), LibcxxStringSummaryProvider() has the easy task - it sets up a StringPrinter, tells it how much data to print, where to get it from, and then delegates the StringPrinter to do the grunt work
>> StringPrinter is a nifty little tool - it can handle generating summaries for different kinds of strings (UTF8? UTF16? we got it - is a \0 a terminator? what quote character would you like? …) - you point it at some data, set up a few options, and it will generate a printable representation for you - if your string type is doing anything out of the ordinary, let’s talk - I am definitely open to extending StringPrinter to handle even more magic
>> 
>> Part 2 - Teaching std::string that it can be backed by an fbstring_core
>> 
>> At the end of part 1, you’ll probably end up with a FBStringCoreSummaryProvider() - now you need to teach LLDB about it
>> The obvious thing you could do would be to go in CPlusPlusLanguage::GetFormatters() add a LoadFBStringFormatter(g_category) to it - and then imitate - say - LoadLibCxxFormatters()
>> 
>>    AddCXXSummary(cpp_category_sp, lldb_private::formatters::FBStringCoreSummaryProvider, “fbstringcore summary provider", ConstString(“std::fbstring_core<.+>"), stl_summary_flags, true);
>> 
>> That will work - but what you would see is:
>> 
>>> (const string &const) corpus = error: summary string parsing error: {
>>>  (std::fbstring_core<char>) store_ = “www"
>> 
>> You wanna do
>> 
>> (lldb) log enable lldb formatters
>> (lldb) frame variable -T corpus
>> 
>> It will list one or more typenames - the most specific one is the one you like (e.g. for libc++ we get std::__1::string - this is how we tell ourselves this is the std::string from libc++)
>> Once you find that typename, you’ll make a new formatter - FBStringSummaryProvider() - and register that formatter with that very specific typename
>> 
>> All that FBStringSummaryProvider() has to do is get the “store_” member (ValueObject::GetChildMemberWithName() is your friend) - and pass it down to FBStringCoreSummaryProvider()
>> 
>> 
>> I understand this may seem a little convoluted and arcane at first - but feel free to ask more questions, and I’ll try to help out!
>> 
>>> Thanks.
>>> Jeffrey
>>> 
>>> On Mon, Mar 28, 2016 at 11:38 AM, Enrico Granata <egranata at apple.com> wrote:
>>> This is kind of orthogonal to your problem, but the reason why you are not seeing the kind of simplified printing Greg is suggesting, is because your std::string doesn’t look like any of the kinds we recognize
>>> 
>>> Specifically, LLDB data formatters work by matching against type names, and once they recognize a typename, then they try to inspect the variable in order to grab a summary
>>> In your example, your std::string exposes a layout that we are not handling - hence we bail out of the formatter and we fall back to the raw view
>>> 
>>> If you want pretty printing to work, you’ll need to write a data formatter
>>> 
>>> There are a few avenues. The obvious easy one is to extend the existing std::string formatter to recognize your type’s internal layout.
>>> If one were signing up for more infrastructure work, they could decide to try and detect shared library loads and load formatters that match with whatever libraries are being loaded.
>>> 
>>>> On Mar 28, 2016, at 9:47 AM, Greg Clayton via lldb-dev <lldb-dev at lists.llvm.org> wrote:
>>>> 
>>>> So you need to be prepared to escape any text that can have special characters. A "std::string" or any container can contain special characters. If you are encoding stuff into JSON, you will either need to escape any special characters, or hex encode the string into ASCII hex bytes. 
>>>> 
>>>> In debuggers we often get bogus data because variables are not initialized, but the compiler tells us that a variable is valid in address range [0x1000-0x2000), but it actually is [0x1200-0x2000). If we read a variable in this case, a std::string might contain bogus data and the bytes might not make sense. So you always have to be prepared for bad data.
>>>> 
>>>> If we look at:
>>>> 
>>>> store_ = {
>>>>    = {
>>>>     small_ = "www"
>>>>     ml_ = (data_ =
>>>> "��UH\x89�H�}�H\x8bE�]ÐUH\x89�H��H\x89}�H\x8bE�H\x89��~\xb4��\x90��UH\x89�SH\x83�H\x89}�H�u�H�E�H���\x9e���H\x8b\x18H\x8bE�H���O\xb4��H\x89ƿ\b",
>>>> size_ = 0, capacity_ = 1441151880758558720)
>>>>   }
>>>> }
>>>> }
>>>> 
>>>> We can see the "size_" is zero, and capacity_ is 1441151880758558720 (which is 0x1400000000000000). "data_" seems to be some random pointer. 
>>>> 
>>>> On MacOSX, we have a special formatting code that displays std::string in CPlusPlusLanguage.cpp that gets installed in the LoadLibCxxFormatters() or LoadLibStdcppFormatters() functions with code like:
>>>> 
>>>>   lldb::TypeSummaryImplSP std_string_summary_sp(new CXXFunctionSummaryFormat(stl_summary_flags, lldb_private::formatters::LibcxxStringSummaryProvider, "std::string summary provider"));
>>>>   cpp_category_sp->GetTypeSummariesContainer()->Add(ConstString("std::__1::string"), std_string_summary_sp);
>>>> 
>>>> Special flags are set on std::string to say "don't show children of this and just show a summary" So if a std::string contained "hello". So for the following code:
>>>> 
>>>> std::string h ("hello");
>>>> 
>>>> You should just see:
>>>> 
>>>> (lldb) fr var h
>>>> (std::__1::string) h = "hello"
>>>> 
>>>> If you take a look at the normal value in the raw we see:
>>>> 
>>>> (lldb) fr var --raw h
>>>> (std::__1::string) h = {
>>>> __r_ = {
>>>>   std::__1::__libcpp_compressed_pair_imp<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >::__rep, std::__1::allocator<char>, 2> = {
>>>>     __first_ = {
>>>>        = {
>>>>         __l = {
>>>>           __cap_ = 122511465736202
>>>>           __size_ = 0
>>>>           __data_ = 0x0000000000000000
>>>>         }
>>>>         __s = {
>>>>            = {
>>>>             __size_ = '\n'
>>>>             __lx = '\n'
>>>>           }
>>>>           __data_ = {
>>>>             [0] = 'h'
>>>>             [1] = 'e'
>>>>             [2] = 'l'
>>>>             [3] = 'l'
>>>>             [4] = 'o'
>>>>             [5] = '\0'
>>>>             [6] = '\0'
>>>>             [7] = '\0'
>>>>             [8] = '\0'
>>>>             [9] = '\0'
>>>>             [10] = '\0'
>>>>             [11] = '\0'
>>>>             [12] = '\0'
>>>>             [13] = '\0'
>>>>             [14] = '\0'
>>>>             [15] = '\0'
>>>>             [16] = '\0'
>>>>             [17] = '\0'
>>>>             [18] = '\0'
>>>>             [19] = '\0'
>>>>             [20] = '\0'
>>>>             [21] = '\0'
>>>>             [22] = '\0'
>>>>           }
>>>>         }
>>>>         __r = {
>>>>           __words = {
>>>>             [0] = 122511465736202
>>>>             [1] = 0
>>>>             [2] = 0
>>>>           }
>>>>         }
>>>>       }
>>>>     }
>>>>   }
>>>> }
>>>> }
>>>> 
>>>> So the main question is why are our "std::string" formatters not kicking in for you. That comes down to a typename match, or the format of the string isn't what the formatter is expecting.
>>>> 
>>>> But again, since you std::string can contain anything, you will need to escape any and all text that is encoded into JSON to ensure it doesn't contain anything JSON can't deal with.
>>>> 
>>>>> On Mar 27, 2016, at 9:20 PM, Jeffrey Tan via lldb-dev <lldb-dev at lists.llvm.org> wrote:
>>>>> 
>>>>> Thanks Siva. All the DW_TAG_member related errors seems to go away after patching with your fix. The current problem is handling the decoding. 
>>>>> 
>>>>> Here is the correct decoding from gdb whic might be useful:
>>>>> (gdb) p corpus
>>>>> $3 = (const std::string &) @0x7fd133cfb888: {
>>>>> static npos = 18446744073709551615, store_ = {
>>>>>   static kIsLittleEndian = <optimized out>,
>>>>>   static kIsBigEndian = <optimized out>, {
>>>>>     small_ = "www", '\000' <repeats 20 times>, "\024", ml_ = {
>>>>>       data_ = 0x777777 <std::_Any_data::_M_access<void folly::fibers::Baton::waitFiber<folly::fibers::FirstArgOf<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::{lambda(folly::fibers::Promise<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::SelectionResult>)#1}, void>::type::value_type folly::fibers::await<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::{lambda(folly::fibers::Promise<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::SelectionResult>)#1}>(folly::fibers::FirstArgOf&&)::{lambda()#1}>(folly::fibers::FiberManager&, folly::fibers::FirstArgOf<folly::fibers::FirstArgOf<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::{lambda(folly::fibers::Promise<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::SelectionResult>)#1}, void>::type::value_type folly::fibers::await<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::{lambda(folly::fibers::Promise<facebook::servicerouter::RequestDispatcherBase<facebook::servicerouter::ThriftDispatcher>::prepareForSelection(facebook::servicerouter::DispatchContext&)::SelectionResult>)#1}>(folly::fibers::FirstArgOf&&)::{lambda()#1}, void>::type::value_type)::{lambda(folly::fibers::Fiber&)#1}*>() const+25> "\311\303UH\211\345H\211}\370H\213E\370]ÐUH\211\345H\203\354\020H\211}\370H\213E\370H\211\307\350~\264\312\377\220\311\303UH\211\345SH\203\354\030H\211}\350H\211u\340H\213E\340H\211\307\350\236\377\377\377H\213\030H\213E\350H\211\307\350O\264\312\377H\211ƿ\b", size_ = 0,
>>>>>       capacity_ = 1441151880758558720}}}}
>>>>> 
>>>>> Utf-16 does not seem to decode it, while 'latin-1' does:
>>>>>>>> '\xc9'.decode('utf-16')
>>>>> Traceback (most recent call last):
>>>>> File "<stdin>", line 1, in <module>
>>>>> File "/mnt/gvfs/third-party2/python/55c1fd79d91c77c95932db31a4769919611c12bb/2.7.8/centos6-native/da39a3e/lib/python2.7/encodings/utf_16.py", line 16, in decode
>>>>>   return codecs.utf_16_decode(input, errors, True)
>>>>> UnicodeDecodeError: 'utf16' codec can't decode byte 0xc9 in position 0: truncated data
>>>>>>>> '\xc9'.decode('latin-1')
>>>>> u'\xc9'
>>>>> 
>>>>> Instead of guessing what kind of decoding I should use, I would use 'ensure_ascii=False' to prevent the crash for now.
>>>>> 
>>>>> I tried to reproduce this crash, but it seems that the crash might be related with some internal stl implementation we are using. I will see if I can narrow down to a small repro later. 
>>>>> 
>>>>> Thanks
>>>>> Jeffrey
>>>>> 
>>>>> On Sun, Mar 27, 2016 at 2:49 PM, Siva Chandra <sivachandra at gmail.com> wrote:
>>>>> On Sat, Mar 26, 2016 at 11:58 PM, Jeffrey Tan <jeffrey.fudan at gmail.com> wrote:
>>>>>> Btw: after patching with Siva's fix http://reviews.llvm.org/D18008, the
>>>>>> first field 'small_' is fixed, however the second field 'ml_' still emits
>>>>>> garbage:
>>>>>> 
>>>>>> (lldb) fr v corpus
>>>>>> (const string &const) corpus = error: summary string parsing error: {
>>>>>> store_ = {
>>>>>>    = {
>>>>>>     small_ = "www"
>>>>>>     ml_ = (data_ =
>>>>>> "��UH\x89�H�}�H\x8bE�]ÐUH\x89�H��H\x89}�H\x8bE�H\x89��~\xb4��\x90��UH\x89�SH\x83�H\x89}�H�u�H�E�H���\x9e���H\x8b\x18H\x8bE�H���O\xb4��H\x89ƿ\b",
>>>>>> size_ = 0, capacity_ = 1441151880758558720)
>>>>>>   }
>>>>>> }
>>>>>> }
>>>>> 
>>>>> Do you still see the DW_TAG_member related error?
>>>>> 
>>>>> A wild (and really wild at that) guess: Is it utf16 data that is being
>>>>> decoded as utf8?
>>>>> 
>>>>> As David Blaikie mentioned on the other thread, it would really help
>>>>> if you provide us with a minimal example to repro this. Atleast, repro
>>>>> instructions.
>>>>> 
>>>>> _______________________________________________
>>>>> lldb-dev mailing list
>>>>> lldb-dev at lists.llvm.org
>>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
>>>> 
>>>> _______________________________________________
>>>> lldb-dev mailing list
>>>> lldb-dev at lists.llvm.org
>>>> http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
>>> 
>>> 
>>> Thanks,
>>> - Enrico
>>> 📩 egranata@.com ☎️ 27683
>>> 
>>> 
>> 
>> 
>> Thanks,
>> - Enrico
>> 📩 egranata@.com ☎️ 27683
>> 
>> 
>> _______________________________________________
>> lldb-dev mailing list
>> lldb-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
>> _______________________________________________
>> lldb-dev mailing list
>> lldb-dev at lists.llvm.org
>> http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-dev
> 


Thanks,
- Enrico
📩 egranata@.com ☎️ 27683

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-dev/attachments/20160407/f787b687/attachment-0001.html>


More information about the lldb-dev mailing list