Segmentation fault (core dumped) - reconnect using dns and multithread

[yamoe]

Segmentation fault occurs when reconnect occurs in multi-thread.

gdb> bt
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007f6540976448 in internal_getent (stream=stream@entry=0x7f6540258490, result=result@entry=0x7f6540255b90,
    buffer=buffer@entry=0x7f6540255e00 "\377\377\377\377", buflen=buflen@entry=1024, errnop=errnop@entry=0x7f65402576b0,
    herrnop=herrnop@entry=0x7f65402576e8, af=2, flags=0) at nss_files/files-XXX.c:216
216     nss_files/files-XXX.c: No such file or directory.
[Current thread is 1 (Thread 0x7f6540257700 (LWP 24958))]
(gdb) bt
#0  0x00007f6540976448 in internal_getent (stream=stream@entry=0x7f6540258490, result=result@entry=0x7f6540255b90,
    buffer=buffer@entry=0x7f6540255e00 "\377\377\377\377", buflen=buflen@entry=1024, errnop=errnop@entry=0x7f65402576b0,
    herrnop=herrnop@entry=0x7f65402576e8, af=2, flags=0) at nss_files/files-XXX.c:216
#1  0x00007f65409776af in _nss_files_gethostbyname3_r (name=0x7f6538000b60 "localhost", af=2, result=0x7f6540255b90, buffer=<optimized out>,
    buflen=1024, errnop=0x7f65402576b0, herrnop=0x7f65402576e8, ttlp=0x0, canonp=0x0) at nss_files/files-hosts.c:352
#2  0x00007f6540977801 in _nss_files_gethostbyname2_r (name=<optimized out>, af=<optimized out>, result=<optimized out>,
    buffer=<optimized out>, buflen=<optimized out>, errnop=<optimized out>, herrnop=0x7f65402576e8) at nss_files/files-hosts.c:389
#3  0x00000000006125dd in gethostbyname2_r ()
#4  0x0000000000609994 in gaih_inet.constprop ()
#5  0x000000000060b624 in getaddrinfo ()
#6  0x0000000000499122 in _redisContextConnectTcp (c=c@entry=0x7f6538017d70, addr=addr@entry=0x7f6540256780 "localhost",
    port=port@entry=6001, timeout=timeout@entry=0x0, source_addr=source_addr@entry=0x0) at net.c:328
#7  0x0000000000499808 in redisContextConnectTcp (c=c@entry=0x7f6538017d70, addr=addr@entry=0x7f6540256780 "localhost", port=port@entry=6001,
    timeout=timeout@entry=0x0) at net.c:422
#8  0x000000000049384f in redisConnect (ip=0x7f6540256780 "localhost", port=6001) at hiredis.c:664
#9  0x000000000049b868 in sw::redis::Connection::Connector::_connect() const ()
#10 0x000000000049ba63 in sw::redis::Connection::Connector::connect() const ()
#11 0x000000000049cf90 in sw::redis::Connection::Connection(sw::redis::ConnectionOptions const&) ()
#12 0x000000000049d38e in sw::redis::Connection::reconnect() ()
#13 0x00000000004a2ad2 in sw::redis::ConnectionPool::fetch() ()
#14 0x00000000004b3cfc in std::enable_if<!std::is_convertible<void (*)(sw::redis::Connection&), sw::redis::StringView>::value, std::unique_ptr<redisReply, sw::redis::ReplyDeleter> >::type sw::redis::Redis::command<void (*)(sw::redis::Connection&)>(void (*)(sw::redis::Connection&)) ()
#15 0x00000000004ad6d7 in sw::redis::Redis::ping[abi:cxx11]() ()

I am using hiredis c0.14.0 and v1.0.0 is the same.

The reason is as follows.
https://sourceware.org/bugzilla/show_bug.cgi?id=10652

Currently, it is solved by one of the methods below.

• use single thread (not use multi thread)
• use dynamic link
• use ip address (not dns)

If you have any other solution, please share.

thank you.

[michael-grunder]

This is interesting although I don't see how it's a bug with hiredis. This appears to be an issue with trying to statically link libpthread.a, which I suspect will cause all kinds of problems unless done very carefully.

[yamoe]

thank you for your interest.
i will close this issue.