strace segfaults on 3.4 kernels

Tested on htc-memul and fairphone-fp2, both with a 3.4 kernel.

straceing some processes (tested: sshd, qmicli, qcom_rmtfs, "libinput debug-events") produces a segfault instead of the strace to work.

"Simpler"(?) processes like busybox applets or others seem to work fine.

I've got a backtrace from running gdb strace and then run qmicli:

Program received signal SIGSEGV, Segmentation fault.
0xb6fb58a6 in sigsetjmp () from /lib/ld-musl-armhf.so.1
(gdb) bt
#0  0xb6fb58a6 in sigsetjmp () from /lib/ld-musl-armhf.so.1
#1  0xb6fb5918 in raise () from /lib/ld-musl-armhf.so.1
#2  0x2a018190 in ?? ()

All of the processes listed above (except qmicli) fail with this line as last line in strace:

rt_sigprocmask(SIG_UNBLOCK, [RT_1 RT_2], NULL, 8) = 0
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x1} ---

qmicli last line:

futex(0xb6b0b950, FUTEX_WAKE_PRIVATE, 2147483647) = 0
--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=0x215} ---

I've heard from insep on Matrix, that strace'ing qcom_rmtfs works on kernel 3.18