Description of problem:

A customer hit a CPU consumption issue within bash, which happens when bash tries to reap a dead child.
Taking regularly coredumps, we could see that backtraces show it's spinning in bgp_delete():
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------
#0  0x000055b4ce1a64cf in bgp_delete (pid=pid@entry=2074684) at jobs.c:879
#1  0x000055b4ce1aa7d1 in make_child (command=0x55b4d0c77ed0 "cut -f 1 -d '|'", async_p=async_p@entry=0)
    at jobs.c:2104
#2  0x000055b4ce195e98 in execute_simple_command (simple_command=0x55b4d01725a0, pipe_in=pipe_in@entry=3, 
    pipe_out=pipe_out@entry=-1, async=async@entry=0, fds_to_close=fds_to_close@entry=0x55b4d0d17110)
    at /usr/include/bits/string_fortified.h:90
#3  0x000055b4ce1981a6 in execute_command_internal (command=0x55b4d0175140, asynchronous=0, pipe_in=3, pipe_out=-1, 
    fds_to_close=0x55b4d0d17110) at execute_cmd.c:819
#4  0x000055b4ce19b157 in execute_pipeline (command=command@entry=0x55b4d016c860, asynchronous=asynchronous@entry=0, 
    pipe_in=pipe_in@entry=-1, pipe_out=pipe_out@entry=-1, fds_to_close=fds_to_close@entry=0x55b4d0d17110)
    at execute_cmd.c:2502
[...]

(gdb) list
874	  if (bgpids.storage == 0 || bgpids.nalloc == 0 || bgpids.npid == 0)
875	    return 0;
876	
877	  /* Search chain using hash to find bucket in pidstat_table */
878	  for (psi = *(pshash_getbucket (pid)); psi != NO_PIDSTAT; psi = bgpids.storage[psi].bucket_next)
879	    if (bgpids.storage[psi].pid == pid)
880	      break;
881	
882	  if (psi == NO_PIDSTAT)
883	    return 0;		/* not found */
-------- 8< ---------------- 8< ---------------- 8< ---------------- 8< --------

The issues started appearing after fixing BZ #1890888, but since we don't have a reproducer yet, it's hard to verify this.

Anyway we found out that the issue was reported upstream and on ubuntu:

  https://bugs.launchpad.net/ubuntu/+source/bash/+bug/1822776
  https://lists.gnu.org/archive/html/bug-bash/2017-02/msg00025.html

And patched by https://ftp.gnu.org/gnu/bash/bash-4.4-patches/bash44-020

Please backport this ASAP, since we cannot know if other customers will hit this and the consequence is severe:
 - zombies accumulating
 - high CPU consumption

Version-Release number of selected component (if applicable):

bash-4.4.19-12.el8.x86_64 (RHEL8.3)
bash-4.4.19-11.el8_1.x86_64 (RHEL8.1)


How reproducible:

Don't know, upstream reproducer seems to take days / weeks.