[0.11.22] [Martin] Crash depend on the system encoding (OSX)

This subforum contains all the issues which we already resolved.
Post Reply
roentgen
Manual Inserter
Manual Inserter
Posts: 2
Joined: Thu May 14, 2015 4:15 pm
Contact:

[0.11.22] [Martin] Crash depend on the system encoding (OSX)

Post by roentgen »

Hi,

factorio 0.11.22 shall crash while it boots up before producing its log file.
This problem will occur when the system encoding is kCFStringEncodingMacJapanese. (also may occur at other encoding except kCFStringEncodingMacRoman. maybe)

callstack:

#0 0x00007fff89d17732 in strlen ()
#1 0x00000001002a33e9 in Paths::getSystemReadData ()
#2 0x0000000100356c86 in PathMacroReplacer::apply ()
#3 0x0000000100356594 in MacroReplacer::operator() ()
#4 0x0000000100465c95 in boost::re_detail::format_functor1<ReplacerWrapper, boost::match_results<std::__1::__wrap_iter<char const*>, std::__1::allocator<boost::sub_match<std::__1::__wrap_iter<char const*> > > > >::operator()<boost::re_detail::string_out_iterator<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > > > ()
#5 0x000000010045f1db in boost::regex_replace<boost::re_detail::string_out_iterator<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > >, std::__1::__wrap_iter<char const*>, boost::regex_traits<char, boost::cpp_regex_traits<char> >, char, ReplacerWrapper> ()
#6 0x000000010019123a in MacroReplacer::replace ()
#7 0x00000001002a3516 in Paths::load ()
#8 0x00000001002a23e0 in Paths::Paths ()
#9 0x000000010018fc52 in GlobalContext::init ()
#10 0x000000010027d7f6 in MainLoop::run ()
#11 0x000000010027cc02 in _al_mangled_main ()
#12 0x00000001006af4a7 in +[AllegroAppDelegate app_main:] ()
#13 0x00007fff8522376b in __NSThread__main__ ()
#14 0x00007fff8be89899 in _pthread_body ()
#15 0x00007fff8be8972a in _pthread_start ()
#16 0x00007fff8be8dfc9 in thread_start ()

disassembled and comments:

Dump of assembler code for function _ZN5Paths17getSystemReadDataEv:
0x00000001002a33a0 <_ZN5Paths17getSystemReadDataEv+0>: push %rbp
0x00000001002a33a1 <_ZN5Paths17getSystemReadDataEv+1>: mov %rsp,%rbp
0x00000001002a33a4 <_ZN5Paths17getSystemReadDataEv+4>: push %r15
0x00000001002a33a6 <_ZN5Paths17getSystemReadDataEv+6>: push %r14
0x00000001002a33a8 <_ZN5Paths17getSystemReadDataEv+8>: push %r12
0x00000001002a33aa <_ZN5Paths17getSystemReadDataEv+10>: push %rbx
0x00000001002a33ab <_ZN5Paths17getSystemReadDataEv+11>: sub $0x50,%rsp
0x00000001002a33af <_ZN5Paths17getSystemReadDataEv+15>: mov %rdi,%r14
0x00000001002a33b2 <_ZN5Paths17getSystemReadDataEv+18>: callq 0x1007a094c <dyld_stub_CFBundleGetMainBundle>
0x00000001002a33b7 <_ZN5Paths17getSystemReadDataEv+23>: mov %rax,%rdi
0x00000001002a33ba <_ZN5Paths17getSystemReadDataEv+26>: callq 0x1007a093a <dyld_stub_CFBundleCopyBundleURL>
0x00000001002a33bf <_ZN5Paths17getSystemReadDataEv+31>: mov %rax,%r12
0x00000001002a33c2 <_ZN5Paths17getSystemReadDataEv+34>: xor %esi,%esi
0x00000001002a33c4 <_ZN5Paths17getSystemReadDataEv+36>: mov %r12,%rdi
0x00000001002a33c7 <_ZN5Paths17getSystemReadDataEv+39>: callq 0x1007a09be <dyld_stub_CFURLCopyFileSystemPath>
0x00000001002a33cc <_ZN5Paths17getSystemReadDataEv+44>: mov %rax,%r15
0x00000001002a33cf <_ZN5Paths17getSystemReadDataEv+47>: callq 0x1007a09b8 <dyld_stub_CFStringGetSystemEncoding>
;
; CFStringGetSystemEncoding returns '1' (kCFStringEncodingMacJapanese) on my env
;
0x00000001002a33d4 <_ZN5Paths17getSystemReadDataEv+52>: mov %r15,%rdi
;
; eax has a return value of CFStringGetSystemEncoding: 1
; (Here I break it at here and did 'set $rax=0' on gdb, then factorio runs finely)
;
0x00000001002a33d7 <_ZN5Paths17getSystemReadDataEv+55>: mov %eax,%esi
0x00000001002a33d9 <_ZN5Paths17getSystemReadDataEv+57>: callq 0x1007a09b2 <dyld_stub_CFStringGetCStringPtr>
;
; CFStringGetCStringPtr returns NULL if encoding argument is non-8bits-encoding(like kCGStringEncofingMacJapanese).
;
0x00000001002a33de <_ZN5Paths17getSystemReadDataEv+62>: mov %rax,%rbx
0x00000001002a33e1 <_ZN5Paths17getSystemReadDataEv+65>: mov %rbx,%rdi
;
; It crashes in strlem(NULL).
;
0x00000001002a33e4 <_ZN5Paths17getSystemReadDataEv+68>: callq 0x1007a01d2 <dyld_stub_strlen>
0x00000001002a33e9 <_ZN5Paths17getSystemReadDataEv+73>: lea -0x50(%rbp),%rdi

I think the problem is easy to fix.
HTH,

Mac OSX 10.9.4

MF-
Smart Inserter
Smart Inserter
Posts: 1235
Joined: Sun Feb 24, 2013 12:07 am
Contact:

Re: [0.11.22] Crash depend on the system encoding (OSX)

Post by MF- »

Any idea what does "system encoding" mean?

If it is an encoding throughout the entire system, it should take several hours to change.
If it doesn't do that, what the hell is it an encoding of?

roentgen
Manual Inserter
Manual Inserter
Posts: 2
Joined: Thu May 14, 2015 4:15 pm
Contact:

Re: [0.11.22] Crash depend on the system encoding (OSX)

Post by roentgen »

MF- wrote:Any idea what does "system encoding" mean?

If it is an encoding throughout the entire system, it should take several hours to change.
If it doesn't do that, what the hell is it an encoding of?
An encoding throughout the entire system, that's right..
Wow you called the API, didn't you? ;-)

User avatar
HanziQ
Former Staff
Former Staff
Posts: 630
Joined: Fri Mar 27, 2015 7:07 am
Contact:

Re: [0.11.22] [Martin] Crash depend on the system encoding (OSX)

Post by HanziQ »

I forced CFStringGetCStringPtr to use kCFStringEncodingUTF8 and then told boost to use utf8_codecvt_facet to represent it properly, so that should work, if the CFString manages to convert itself to UTF-8. I unfortunately have no way of testing this and getting a Japanese OS X is no easy task, even having it as a VM would probably be difficult. I'll mark this as resolved and we'll have to see, if this actually works when 0.12 is released.

Post Reply

Return to “Resolved Problems and Bugs”