Andreas Metzler wrote:
> On AMD64 I also see "eatmydata gnutls-cli" breaking, but not in ix86.
> Sprinkling GnuTLS with fprintf() like this
>
> device_fd = open("/dev/urandom", O_RDONLY);
> perror("AMDEBUG perror");
>...
> AMDEBUG perror: Bad address
> AMDEBUG device_fd [-1]
Good debugging! That is the code decision point that causes the
problem. Due to device_fd being -1 the gnutls
_rnd_system_entropy_init() routine decides to give up on /dev/urandom
and then to fall back and try connecting to an egd and then the
failure report comes from the egd portion of the code.
> EFAULT does not really make sense here, does it?
Not to me either. How can open("/dev/urandom", O_RDONLY); return an
EFAULT? Does not make sense to me.
Bob
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
