Natanael Copa wrote in <20190119004724.78b1b0c9_at_ncopa-macbook.copa.dup.pw>:
|On Thu, 17 Jan 2019 15:02:07 +0100
|Steffen Nurpmeso <steffen_at_sdaoden.eu> wrote:
|> After upgrading to [edge] i see a possibly endless hang upon
|> startup, which seems to be caused by PRNG init.
|Try add random.trust_cpu=1 as boot option.
Will do when i finally can upgrade to 4.19.y series. Merci.
I am back at 4.14.89 since the nf_conntrack crashes persist, and
even 4.19.16 does not seem to have anything related. (4.19.15
had, but crashes, ehm, persist.) That is: i have to wait.
But ok i mean all that seems to be problematic, for example with
the program i have posted:
#?64|xhost:/etc/init.d# /sbin/entropy-saver save
- 1588 bits of entropy available at /dev/random
- 2422 bits of entropy remain at /dev/random
Insufficient entropy to save from /dev/random (-834 bits)
Two adjacent ioctl() asking for entropy_avail with just one read
(!) of /dev/random in between causes the entropy to explode!
This _cannot_ be right, can it?
Thereafter it works for the second try:
#?0|xhost:/etc/init.d# /sbin/entropy-saver save
- 2435 bits of entropy available at /dev/random
- 1411 bits of entropy remain at /dev/random
- 1024 bits / 80 bytes of entropy saved to /var/lib/misc/random.dat
Very weakly rated, this stuff.
Say, isn't that flaky? This VM really crawls along, with
jitterentropy, with haveged, and with the above. While i am
writing, the entropy went down to 1343, with only postfix
(urandom), mailman (1000 x urandom), dnsmasq (using internal surf
rand with one-time seeding), lighttpd (uses getrandom/urandom)
and cgit running. Whereas my laptop is stable at 38xy.
So sorry to bother you all with this!
I will turn to use linux-virt on the VM. Sorry for the noise.
A nice weekend i wish.
|Der Kragenbaer, The moon bear,
|der holt sich munter he cheerfully and one by one
|einen nach dem anderen runter wa.ks himself off
|(By Robert Gernhardt)
Received on Sat Jan 19 2019 - 18:15:35 UTC