Unbound 1.7.1 failing on some kvm servers

[James Cloos]

>>>>> "WW" == W C A Wijngaards <wouter at nlnetlabs.nl> writes:

>> Once the call to glibc's genentropy(3) fails, it immediately sends a
>> SIGKILL.

WW> But in your strace you have a call to getrandom() that fails.

compat/getentropy_linux.c calls genentropy(3) which calls getrandom(2).

>> Indeed, compat/arc4random.c:_rs_stir calls genentropy and does not fall back.

WW> Yes that does that.  fall back for some glibc implementation?

It is not really glibc-specific.  Rather everything is compiled
expecting one kernel version and running on a different kernel.

WW> but can you run ltrace, it should intercept
WW> the library call and tell us if unbound is calling a function from glibc

The ltrace(8) output ends with:

21004 14:53:22.120825 [0x7f91c82d638e] getpid( <unfinished ...>
21004 14:53:22.120981 [0x7f91c641d257] SYS_getpid()                              = 21004 <0.000053>
21004 14:53:22.121072 [0x7f91c82d638e] <... getpid resumed> )                    = 21004 <0.000244>
21004 14:53:22.121142 [0x7f91c82d613a] getentropy(0x7ffcfab88650, 40, 0, -1 <unfinished ...>
21004 14:53:22.121341 [0x7f91c6390f7a] SYS_318(0x7ffcfab88650, 40, 0, 318)       = -38 <0.000087>
21004 14:53:22.121480 [0x7f91c82d613a] <... getentropy resumed> )                = 0xffffffff <0.000331>
21004 14:53:22.121555 [0x7f91c82d627a] raise(9, 40, -264, -1 <unfinished ...>
21004 14:53:22.121729 [0x7f91c6723dab] SYS_rt_sigprocmask(0, 0x7ffcfab885b0, 0x7ffcfab88530, 8) = 0 <0.000071>
21004 14:53:22.121843 [0x7f91c6723db4] SYS_getpid()                              = 21004 <0.000044>
21004 14:53:22.121929 [0x7f91c6723dbe] SYS_gettid(0x520c, 0x7ffcfab885b0, 0x7ffcfab88530, 8) = 0x520c <0.000058>
21004 14:53:22.122039 [0x7f91c6723dc9] SYS_tgkill(0x520c, 0x520c, 9, 8 <no return ...>
21004 14:53:22.123306 [0xffffffffffffffff] +++ killed by SIGKILL +++

-JimC
-- 
James Cloos <cloos at jhcloos.com>         OpenPGP: 0x997A9F17ED7DAEA6