From 2fe03e832adb11c50bcfc62679cf17779b284124 Mon Sep 17 00:00:00 2001 From: Rickard Green Date: Wed, 27 Jul 2016 19:45:41 +0200 Subject: Reclaim literal area after purge has completed --- lib/kernel/test/init_SUITE.erl | 48 +++++++++++++++++++++++++++++++++++++++--- 1 file changed, 45 insertions(+), 3 deletions(-) (limited to 'lib/kernel') diff --git a/lib/kernel/test/init_SUITE.erl b/lib/kernel/test/init_SUITE.erl index 1370e23195..b96de94427 100644 --- a/lib/kernel/test/init_SUITE.erl +++ b/lib/kernel/test/init_SUITE.erl @@ -27,7 +27,8 @@ -export([get_arguments/1, get_argument/1, boot_var/1, restart/1, many_restarts/0, many_restarts/1, get_plain_arguments/1, - reboot/1, stop_status/1, stop/1, get_status/1, script_id/1]). + reboot/1, stop_status/1, stop/1, get_status/1, script_id/1, + find_system_processes/0]). -export([boot1/1, boot2/1]). -export([init_per_testcase/2, end_per_testcase/2]). @@ -355,12 +356,16 @@ wait_for(N,Node,EHPid) -> restart(Config) when is_list(Config) -> Args = args(), + Pa = " -pa " ++ filename:dirname(code:which(?MODULE)), + %% Currently test_server:start_node cannot be used. The restarted %% node immediately halts due to the implementation of %% test_server:start_node. - {ok, Node} = loose_node:start(init_test, Args, ?DEFAULT_TIMEOUT_SEC), + {ok, Node} = loose_node:start(init_test, Args ++ Pa, ?DEFAULT_TIMEOUT_SEC), %% Ok, the node is up, now the real test test begins. erlang:monitor_node(Node, true), + SysProcs0 = rpc:call(Node, ?MODULE, find_system_processes, []), + [InitPid, PurgerPid, LitCollectorPid] = SysProcs0, InitPid = rpc:call(Node, erlang, whereis, [init]), PurgerPid = rpc:call(Node, erlang, whereis, [erts_code_purger]), Procs = rpc:call(Node, erlang, processes, []), @@ -375,6 +380,9 @@ restart(Config) when is_list(Config) -> end, ok = wait_restart(30, Node), + SysProcs1 = rpc:call(Node, ?MODULE, find_system_processes, []), + [InitPid1, PurgerPid1, LitCollectorPid1] = SysProcs1, + %% Still the same init process! InitPid1 = rpc:call(Node, erlang, whereis, [init]), InitP = pid_to_list(InitPid), @@ -385,8 +393,16 @@ restart(Config) when is_list(Config) -> PurgerP = pid_to_list(PurgerPid), PurgerP = pid_to_list(PurgerPid1), + %% and same literal area collector process! + case LitCollectorPid of + undefined -> undefined = LitCollectorPid1; + _ -> + LitCollectorP = pid_to_list(LitCollectorPid), + LitCollectorP = pid_to_list(LitCollectorPid1) + end, + NewProcs0 = rpc:call(Node, erlang, processes, []), - NewProcs = NewProcs0 -- [InitPid1, PurgerPid1], + NewProcs = NewProcs0 -- SysProcs1, case check_processes(NewProcs, MaxPid) of true -> ok; @@ -406,6 +422,32 @@ restart(Config) when is_list(Config) -> loose_node:stop(Node), ok. +-record(sys_procs, {init, + code_purger, + literal_collector}). + +find_system_processes() -> + find_system_procs(processes(), #sys_procs{}). + +find_system_procs([], SysProcs) -> + [SysProcs#sys_procs.init, + SysProcs#sys_procs.code_purger, + SysProcs#sys_procs.literal_collector]; +find_system_procs([P|Ps], SysProcs) -> + case process_info(P, initial_call) of + {initial_call,{otp_ring0,start,2}} -> + undefined = SysProcs#sys_procs.init, + find_system_procs(Ps, SysProcs#sys_procs{init = P}); + {initial_call,{erts_code_purger,start,0}} -> + undefined = SysProcs#sys_procs.code_purger, + find_system_procs(Ps, SysProcs#sys_procs{code_purger = P}); + {initial_call,{erts_literal_area_collector,start,0}} -> + undefined = SysProcs#sys_procs.literal_collector, + find_system_procs(Ps, SysProcs#sys_procs{literal_collector = P}); + _ -> + find_system_procs(Ps, SysProcs) + end. + wait_restart(0, _Node) -> ct:fail(not_restarted); wait_restart(N, Node) -> -- cgit v1.2.3 From 9d0638216d35ca0f21c1eea20f8daa3992ac4f71 Mon Sep 17 00:00:00 2001 From: Rickard Green Date: Tue, 2 Aug 2016 15:58:06 +0200 Subject: Fix purge of code Ensure that we cannot get any dangling pointers into code that has been purged. This is done by a two phase purge. At first phase all fun entries pointing into the code to purge are marked for purge. All processes trying to call these funs will be suspended and by this we avoid getting new direct references into the code. When all processes has been checked, these processes are resumed. The new purge strategy now also completely ignore the existence of indirect references to the code (funs). If such exist, they will cause bad fun exceptions to the caller, but will not prevent a soft purge or cause a kill of a process having such live references during a hard purge. This since it is impossible to give any guarantees that no processes in the system have such indirect references. Even when the system is completely clean from such references, new ones can appear via distribution and/or disk. --- lib/kernel/doc/src/code.xml | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'lib/kernel') diff --git a/lib/kernel/doc/src/code.xml b/lib/kernel/doc/src/code.xml index d3611d6a03..21f8a2d54a 100644 --- a/lib/kernel/doc/src/code.xml +++ b/lib/kernel/doc/src/code.xml @@ -651,6 +651,11 @@ ok = code:finish_loading(Prepared),

Purges the code for Module, that is, removes code marked as old. If some processes still linger in the old code, these processes are killed before the code is removed.

+

As of ERTS version 9.0, a process is only considered + to be lingering in the code if it has direct references to the code. + For more information see documentation of + erlang:check_process_code/3, + which is used in order to determine this.

Returns true if successful and any process is needed to be killed, otherwise false.

@@ -661,6 +666,11 @@ ok = code:finish_loading(Prepared),

Purges the code for Module, that is, removes code marked as old, but only if no processes linger in it.

+

As of ERTS version 9.0, a process is only considered + to be lingering in the code if it has direct references to the code. + For more information see documentation of + erlang:check_process_code/3, + which is used in order to determine this.

Returns false if the module cannot be purged because of processes lingering in old code, otherwise true.

 -- cgit v1.2.3