search for: open_urandom

...ystem (multiple user reports[3]). SELinux is disabled and /dev/urandom is chmod' 666. Any pointers? Backtrace from samba-client-libs-4.9.1-2.fc29.x86_64: #0  0x00007fe785c0b53f in raise () from /lib64/libc.so.6 #1  0x00007fe785bf5895 in abort () from /lib64/libc.so.6 #2  0x00007fe782ad97d5 in open_urandom () at ../lib/util/genrand.c:37 #3  generate_random_buffer (out=out at entry=0x7ffeb252b536 ".\025\260\230\027\207\260U",     len=len at entry=2) at ../lib/util/genrand.c:46 #4  0x00007fe78501fc6e in generate_trn_id () at ../source3/libsmb/namequery.c:1276 #5  name_query_send (mem_ctx=&lt...

...x is disabled and > /dev/urandom is chmod' 666. > > Any pointers? > > Backtrace from samba-client-libs-4.9.1-2.fc29.x86_64: > #0  0x00007fe785c0b53f in raise () from /lib64/libc.so.6 > #1  0x00007fe785bf5895 in abort () from /lib64/libc.so.6 > #2  0x00007fe782ad97d5 in open_urandom () at ../lib/util/genrand.c:37 > #3  generate_random_buffer (out=out at entry=0x7ffeb252b536 ".\025\260\230\027\207\260U", >     len=len at entry=2) at ../lib/util/genrand.c:46 The above is deliberate. open of /dev/urandom should never fail. If it does we have worse problems :-)....

On 11/2/18 1:32 PM, Jeremy Allison wrote: > What errno is being returned on open of /dev/urandom ? I think the backtrace is being munged or the stack is being blown. When I debug during the crash here is what I find: (gdb) p urand_fd $4 = 56 OK... so the fd is set? and 56? I can confirm that fd 56 is /dev/urandom: $ ls -l /proc/14777/fd/56 lr-x------. 1 michael michael 64 Nov  2 14:17

On 11/2/18 2:46 PM, Jeremy Allison via samba wrote: > the abort(), which gets triggered if urand_fd == -1 > after open. So you'll need to debug that. Correct. The gdb output I pasted was during the abort. Gdb is showing urand_fd is not -1.