Cellular Network Infrastructure » OsmoGSMTester

Core was generated by `/home/neels/trial-182/inst/osmo-bts-trx/bin/osmo-bts-trx -r 1 -c /home/neels/tr'.
Program terminated with signal SIGILL, Illegal instruction.
#0  _mm_shuffle_epi8 (__Y=..., __X=...) at /usr/lib/gcc/x86_64-linux-gnu/4.9/include/tmmintrin.h:138
138      return (__m128i) __builtin_ia32_pshufb128 ((__v16qi)__X, (__v16qi)__Y);
(gdb) bt
#0  _mm_shuffle_epi8 (__Y=..., __X=...) at /usr/lib/gcc/x86_64-linux-gnu/4.9/include/tmmintrin.h:138
#1  _sse_metrics_k5_n2 (norm=1, paths=0x1e26510, sums=0x1d79290, out=<optimized out>, val=<synthetic pointer>) at viterbi_sse.c:328
#2  osmo_conv_gen_metrics_k5_n2_sse (
    val=0x7fffc98bfb61 "xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, out=0x1deda50, sums=0x1d79290, paths=0x1e26510, norm=1) at viterbi_sse.c:571
#3  0x00007ff54f94f7a4 in forward_traverse (dec=dec@entry=0x1dbfab0, 
    seq=seq@entry=0x7fffc98bfb61 "xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>) at viterbi.c:606
#4  0x00007ff54f9501f8 in conv_decode (term=0, len=14, out=0x7fffc98bf940 "\217\371\213\311\377\177", punc=0x0, 
    seq=0x7fffc98bfb61 "xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, dec=0x1dbfab0) at viterbi.c:632
#5  osmo_conv_decode_acc (code=<optimized out>, 
    input=0x7fffc98bfb61 "xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, output=0x7fffc98bf940 "\217\371\213\311\377\177") at viterbi.c:688
#6  0x00007ff54f94cf75 in osmo_conv_decode (code=0x7ff54f9392c0 <gsm0503_rach>, 
    input=0x7fffc98bfb61 "xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, output=output@entry=0x7fffc98bf940 "\217\371\213\311\377\177") at conv.c:616
#7  0x00007ff54f2e14b0 in gsm0503_rach_decode (ra=0x7fffc98bf98f "", burst=<optimized out>, bsic=<optimized out>)
    at gsm0503_coding.c:2626
#8  0x000000000040a03d in rx_rach_fn (l1t=0x1dfef38, tn=80 'P', fn=547561, chan=31614224, bid=1 '\001', 
    bits=0x7fffc98bfb30 "\377\066\221\231\205t\207ki\211pp\204n\230\236p\212\211\206\201\201\201\221\225V\\\244\240Sc\247\226i\201\177\201s\213tud\223\212\213\220k\177\177xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, nbits=148, rssi=-42 '\326', toa=-1.44140625) at scheduler_trx.c:789
#9  0x0000000000427a4c in trx_sched_ul_burst (l1t=0x1dfef38, tn=122 'z', current_fn=547561, 
    bits=0x7fffc98bfb30 "\377\066\221\231\205t\207ki\211pp\204n\230\236p\212\211\206\201\201\201\221\225V\\\244\240Sc\247\226i\201\177\201s\213tud\223\212\213\220k\177\177xz\177\177s~\177~u~\177{t|\177~v|\177zx}\177\177w{\177{u\177\177\177w{\177\177yun\252\310", <incomplete sequence \360>, nbits=1, rssi=120 'x', toa=-1.44140625) at scheduler.c:1687
#10 0x00000000004052ed in trx_data_read_cb (ofd=0x7fffc98bfb61, what=31382096) at trx_if.c:479
#11 0x00007ff54f944ad2 in osmo_fd_disp_fds (_eset=0x7fffc98c0050, _wset=0x7fffc98bffd0, _rset=0x7fffc98bff50) at select.c:178
#12 osmo_select_main (polling=polling@entry=0) at select.c:218
#13 0x00000000004217d4 in bts_main (argc=<optimized out>, argv=<optimized out>) at main.c:359
#14 0x00007ff54e0c1b45 in __libc_start_main (main=0x404990 <main>, argc=7, argv=0x7fffc98c0228, init=<optimized out>, 
    fini=<optimized out>, rtld_fini=<optimized out>, stack_end=0x7fffc98c0218) at libc-start.c:287
#15 0x00000000004049be in _start ()
(gdb) 

The 'SIGILL' is introduced by commit 34e228a9bcf3ac37287bb5e684ace46818740f3b.

    core/conv: add x86 SSE support for Viterbi decoder

    Fast convolutional decoding is provided through x86 intrinsic based
    SSE operations. SSE3, found on virtually all modern x86 processors,
    is the minimal requirement. SSE4.1 and AVX2 are used if available.

    Also, the original code was extended with runtime SIMD detection,
    so only supported extensions will be used by target CPU. It makes
    the library more partable, what is very important for binary
    packages distribution. Runtime SIMD detection is currently
    implemented through the __builtin_cpu_supports call.

    Change-Id: I1da6d71ed0564f1d684f3a836e998d09de5f0351

Testing whether https://gerrit.osmocom.org/2760 fixes it...

https://gerrit.osmocom.org/2760 does indeed fix it!