Blob Blame History Raw
From 10f1730073b9fb02d2ed7f7de855afd6df0e5202 Mon Sep 17 00:00:00 2001
From: Xavi Hernandez <xhernandez@redhat.com>
Date: Wed, 19 Feb 2020 12:24:15 +0100
Subject: [PATCH 355/355] core: Prevent crash on process termination

A previous patch (ce61da816a) has fixed a use-after-free issue,
but it doesn't work well when the final cleanup is done at process
termination because gluster doesn't stop other threads before
calling exit().

For this reason, the final cleanup is removed to avoid the crash,
at least until the termination sequence properly stops all gluster
threads before exiting the program.

Upstream patch:
> Upstream patch link: https://review.gluster.org/c/glusterfs/+/24138
> Change-Id: Id7cfb4407fcf208e28f03a7c3cdc3ef9c1f3bf9b
> Fixes: bz#1801684
> Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>

Change-Id: Id7cfb4407fcf208e28f03a7c3cdc3ef9c1f3bf9b
BUG: 1800703
Signed-off-by: Xavi Hernandez <xhernandez@redhat.com>
Reviewed-on: https://code.engineering.redhat.com/gerrit/192344
Tested-by: RHGS Build Bot <nigelb@redhat.com>
Reviewed-by: Sunil Kumar Heggodu Gopala Acharya <sheggodu@redhat.com>
---
 libglusterfs/src/mem-pool.c | 30 +++++++++++-------------------
 1 file changed, 11 insertions(+), 19 deletions(-)

diff --git a/libglusterfs/src/mem-pool.c b/libglusterfs/src/mem-pool.c
index 2b41c01..73503e0 100644
--- a/libglusterfs/src/mem-pool.c
+++ b/libglusterfs/src/mem-pool.c
@@ -541,25 +541,17 @@ mem_pools_preinit(void)
 static __attribute__((destructor)) void
 mem_pools_postfini(void)
 {
-    per_thread_pool_list_t *pool_list, *next;
-
-    /* This is part of a process shutdown (or dlclose()) which means that
-     * most probably all threads should be stopped. However this is not the
-     * case for gluster and there are even legitimate situations in which we
-     * could have some threads alive. What is sure is that none of those
-     * threads should be using anything from this library, so destroying
-     * everything here should be fine and safe. */
-
-    list_for_each_entry_safe(pool_list, next, &pool_threads, thr_list)
-    {
-        mem_pool_thread_destructor(pool_list);
-    }
-
-    list_for_each_entry_safe(pool_list, next, &pool_free_threads, thr_list)
-    {
-        list_del(&pool_list->thr_list);
-        FREE(pool_list);
-    }
+    /* TODO: This function should destroy all per thread memory pools that
+     *       are still alive, but this is not possible right now because glibc
+     *       starts calling destructors as soon as exit() is called, and
+     *       gluster doesn't ensure that all threads have been stopped before
+     *       calling exit(). Existing threads would crash when they try to use
+     *       memory or they terminate if we destroy things here.
+     *
+     *       When we propertly terminate all threads, we can add the needed
+     *       code here. Till then we need to leave the memory allocated. Most
+     *       probably this function will be executed on process termination,
+     *       so the memory will be released anyway by the system. */
 }
 
 /* Call mem_pools_init() once threading has been configured completely. This
-- 
1.8.3.1