Message ID | 20200418161729.14422-1-ap420073@gmail.com |
---|---|
State | Changes Requested |
Delegated to: | David Miller |
Headers | show |
Series | [net] team: fix hang in team_mode_get() | expand |
Sat, Apr 18, 2020 at 06:17:29PM CEST, ap420073@gmail.com wrote: >When team mode is changed or set, the team_mode_get() is called to check >whether the mode module is inserted or not. If the mode module is not >inserted, it calls the request_module(). >In the request_module(), it creates a child process, which is >the "modprobe" process and waits for the done of the child process. >At this point, the following locks were used. >down_read(&cb_lock()); by genl_rcv() > genl_lock(); by genl_rcv_msc() > rtnl_lock(); by team_nl_cmd_options_set() > mutex_lock(&team->lock); by team_nl_team_get() > >Concurrently, the team module could be removed by rmmod or "modprobe -r" >The __exit function of team module is team_module_exit(), which calls >team_nl_fini() and it tries to acquire following locks. >down_write(&cb_lock); > genl_lock(); >Because of the genl_lock() and cb_lock, this process can't be finished >earlier than request_module() routine. > >The problem secenario. >CPU0 CPU1 >team_mode_get > request_module() > modprobe -r team_mode_roundrobin > team <--(B) > modprobe team <--(A) > team_mode_roundrobin > >By request_module(), the "modprobe team_mode_roundrobin" command >will be executed. At this point, the modprobe process will decide >that the team module should be inserted before team_mode_roundrobin. >Because the team module is being removed. > >By the module infrastructure, the same module insert/remove operations >can't be executed concurrently. >So, (A) waits for (B) but (B) also waits for (A) because of locks. >So that the hang occurs at this point. > >Test commands: > while : > do > teamd -d & > killall teamd & > modprobe -rv team_mode_roundrobin & > done > >The approach of this patch is to hold the reference count of the team >module if the team module is compiled as a module. If the reference count >of the team module is not zero while request_module() is being called, >the team module will not be removed at that moment. >So that the above scenario could not occur. > >Fixes: 3d249d4ca7d0 ("net: introduce ethernet teaming device") >Signed-off-by: Taehee Yoo <ap420073@gmail.com> >--- > drivers/net/team/team.c | 10 +++++++++- > 1 file changed, 9 insertions(+), 1 deletion(-) > >diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c >index 4004f98e50d9..21702bc23705 100644 >--- a/drivers/net/team/team.c >+++ b/drivers/net/team/team.c >@@ -465,9 +465,15 @@ EXPORT_SYMBOL(team_mode_unregister); > > static const struct team_mode *team_mode_get(const char *kind) > { >- struct team_mode_item *mitem; > const struct team_mode *mode = NULL; >+ struct team_mode_item *mitem; >+ bool put = false; > >+#if IS_MODULE(CONFIG_NET_TEAM) >+ if (!try_module_get(THIS_MODULE)) Can't you call this in case this is not a module? Wouldn't THIS_MODULE be NULL then? try_module_get() handles that correctly. >+ return NULL; >+ put = true; >+#endif > spin_lock(&mode_list_lock); > mitem = __find_mode(kind); > if (!mitem) { >@@ -483,6 +489,8 @@ static const struct team_mode *team_mode_get(const char *kind) > } > > spin_unlock(&mode_list_lock); >+ if (put) >+ module_put(THIS_MODULE); Can't you just put this under the same "if IS_MODULE" statement and avoid the "put" variable? Or in case the statement is not needed, just do plain module_put call. Otherwise, the patch looks fine. > return mode; > } > >-- >2.17.1 >
On Mon, 20 Apr 2020 at 19:21, Jiri Pirko <jiri@resnulli.us> wrote: > Hi Jiri, Thank you for your review! > Sat, Apr 18, 2020 at 06:17:29PM CEST, ap420073@gmail.com wrote: > >When team mode is changed or set, the team_mode_get() is called to check > >whether the mode module is inserted or not. If the mode module is not > >inserted, it calls the request_module(). > >In the request_module(), it creates a child process, which is > >the "modprobe" process and waits for the done of the child process. > >At this point, the following locks were used. > >down_read(&cb_lock()); by genl_rcv() > > genl_lock(); by genl_rcv_msc() > > rtnl_lock(); by team_nl_cmd_options_set() > > mutex_lock(&team->lock); by team_nl_team_get() > > > >Concurrently, the team module could be removed by rmmod or "modprobe -r" > >The __exit function of team module is team_module_exit(), which calls > >team_nl_fini() and it tries to acquire following locks. > >down_write(&cb_lock); > > genl_lock(); > >Because of the genl_lock() and cb_lock, this process can't be finished > >earlier than request_module() routine. > > > >The problem secenario. > >CPU0 CPU1 > >team_mode_get > > request_module() > > modprobe -r team_mode_roundrobin > > team <--(B) > > modprobe team <--(A) > > team_mode_roundrobin > > > >By request_module(), the "modprobe team_mode_roundrobin" command > >will be executed. At this point, the modprobe process will decide > >that the team module should be inserted before team_mode_roundrobin. > >Because the team module is being removed. > > > >By the module infrastructure, the same module insert/remove operations > >can't be executed concurrently. > >So, (A) waits for (B) but (B) also waits for (A) because of locks. > >So that the hang occurs at this point. > > > >Test commands: > > while : > > do > > teamd -d & > > killall teamd & > > modprobe -rv team_mode_roundrobin & > > done > > > >The approach of this patch is to hold the reference count of the team > >module if the team module is compiled as a module. If the reference count > >of the team module is not zero while request_module() is being called, > >the team module will not be removed at that moment. > >So that the above scenario could not occur. > > > >Fixes: 3d249d4ca7d0 ("net: introduce ethernet teaming device") > >Signed-off-by: Taehee Yoo <ap420073@gmail.com> > >--- > > drivers/net/team/team.c | 10 +++++++++- > > 1 file changed, 9 insertions(+), 1 deletion(-) > > > >diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c > >index 4004f98e50d9..21702bc23705 100644 > >--- a/drivers/net/team/team.c > >+++ b/drivers/net/team/team.c > >@@ -465,9 +465,15 @@ EXPORT_SYMBOL(team_mode_unregister); > > > > static const struct team_mode *team_mode_get(const char *kind) > > { > >- struct team_mode_item *mitem; > > const struct team_mode *mode = NULL; > >+ struct team_mode_item *mitem; > >+ bool put = false; > > > >+#if IS_MODULE(CONFIG_NET_TEAM) > >+ if (!try_module_get(THIS_MODULE)) > > Can't you call this in case this is not a module? Wouldn't THIS_MODULE > be NULL then? try_module_get() handles that correctly. > > > >+ return NULL; > >+ put = true; > >+#endif > > spin_lock(&mode_list_lock); > > mitem = __find_mode(kind); > > if (!mitem) { > >@@ -483,6 +489,8 @@ static const struct team_mode *team_mode_get(const char *kind) > > } > > > > spin_unlock(&mode_list_lock); > >+ if (put) > >+ module_put(THIS_MODULE); > > Can't you just put this under the same "if IS_MODULE" statement and > avoid the "put" variable? Or in case the statement is not needed, just > do plain module_put call. > > Otherwise, the patch looks fine. > You're right, If the team is not a module, THIS_MODULE will be null. try_module_get() and module_put() deal with NULL well. So, I will send a v2 patch that will remove the #if statement and the put variable. Thanks a lot! Taehee Yoo
diff --git a/drivers/net/team/team.c b/drivers/net/team/team.c index 4004f98e50d9..21702bc23705 100644 --- a/drivers/net/team/team.c +++ b/drivers/net/team/team.c @@ -465,9 +465,15 @@ EXPORT_SYMBOL(team_mode_unregister); static const struct team_mode *team_mode_get(const char *kind) { - struct team_mode_item *mitem; const struct team_mode *mode = NULL; + struct team_mode_item *mitem; + bool put = false; +#if IS_MODULE(CONFIG_NET_TEAM) + if (!try_module_get(THIS_MODULE)) + return NULL; + put = true; +#endif spin_lock(&mode_list_lock); mitem = __find_mode(kind); if (!mitem) { @@ -483,6 +489,8 @@ static const struct team_mode *team_mode_get(const char *kind) } spin_unlock(&mode_list_lock); + if (put) + module_put(THIS_MODULE); return mode; }
When team mode is changed or set, the team_mode_get() is called to check whether the mode module is inserted or not. If the mode module is not inserted, it calls the request_module(). In the request_module(), it creates a child process, which is the "modprobe" process and waits for the done of the child process. At this point, the following locks were used. down_read(&cb_lock()); by genl_rcv() genl_lock(); by genl_rcv_msc() rtnl_lock(); by team_nl_cmd_options_set() mutex_lock(&team->lock); by team_nl_team_get() Concurrently, the team module could be removed by rmmod or "modprobe -r" The __exit function of team module is team_module_exit(), which calls team_nl_fini() and it tries to acquire following locks. down_write(&cb_lock); genl_lock(); Because of the genl_lock() and cb_lock, this process can't be finished earlier than request_module() routine. The problem secenario. CPU0 CPU1 team_mode_get request_module() modprobe -r team_mode_roundrobin team <--(B) modprobe team <--(A) team_mode_roundrobin By request_module(), the "modprobe team_mode_roundrobin" command will be executed. At this point, the modprobe process will decide that the team module should be inserted before team_mode_roundrobin. Because the team module is being removed. By the module infrastructure, the same module insert/remove operations can't be executed concurrently. So, (A) waits for (B) but (B) also waits for (A) because of locks. So that the hang occurs at this point. Test commands: while : do teamd -d & killall teamd & modprobe -rv team_mode_roundrobin & done The approach of this patch is to hold the reference count of the team module if the team module is compiled as a module. If the reference count of the team module is not zero while request_module() is being called, the team module will not be removed at that moment. So that the above scenario could not occur. Fixes: 3d249d4ca7d0 ("net: introduce ethernet teaming device") Signed-off-by: Taehee Yoo <ap420073@gmail.com> --- drivers/net/team/team.c | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-)