Message ID | 20220329161213.93576-1-dossche.niels@gmail.com (mailing list archive) |
---|---|
State | Changes Requested |
Delegated to: | Netdev Maintainers |
Headers | show |
Series | [net] tipc: use a write lock for keepalive_intv instead of a read lock | expand |
Hi Niels, I did consider this function however I guess it is safe to use tipc_node_read_lock()/unlock() since this value is being apply in this callback function. BTW, you must be using tipc_node_write_unlock_fast() instead of tipc_node_write_unlock(). Regards, Hoang > -----Original Message----- > From: Niels Dossche <dossche.niels@gmail.com> > Sent: Tuesday, March 29, 2022 11:12 PM > To: tipc-discussion@lists.sourceforge.net > Cc: netdev@vger.kernel.org; Jon Maloy <jmaloy@redhat.com>; Ying Xue <ying.xue@windriver.com>; David S. Miller > <davem@davemloft.net>; Jakub Kicinski <kuba@kernel.org>; Paolo Abeni <pabeni@redhat.com>; Hoang Huu Le > <hoang.h.le@dektech.com.au>; Niels Dossche <dossche.niels@gmail.com> > Subject: [PATCH net] tipc: use a write lock for keepalive_intv instead of a read lock > > Currently, n->keepalive_intv is written to while n is locked by a read > lock instead of a write lock. This seems to me to break the atomicity > against other readers. > Change this to a write lock instead to solve the issue. > > Note: > I am currently working on a static analyser to detect missing locks > using type-based static analysis as my master's thesis > in order to obtain my master's degree. > If you would like to have more details, please let me know. > This was a reported case. I manually verified the report by looking > at the code, so that I do not send wrong information or patches. > After concluding that this seems to be a true positive, I created > this patch. I have both compile-tested this patch and runtime-tested > this patch on x86_64. The effect on a running system could be a > potential race condition in exceptional cases. > This issue was found on Linux v5.17. > > Fixes: f5d6c3e5a359 ("tipc: fix node keep alive interval calculation") > Signed-off-by: Niels Dossche <dossche.niels@gmail.com> > --- > net/tipc/node.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/net/tipc/node.c b/net/tipc/node.c > index 6ef95ce565bd..da867ddb93f5 100644 > --- a/net/tipc/node.c > +++ b/net/tipc/node.c > @@ -806,9 +806,9 @@ static void tipc_node_timeout(struct timer_list *t) > /* Initial node interval to value larger (10 seconds), then it will be > * recalculated with link lowest tolerance > */ > - tipc_node_read_lock(n); > + tipc_node_write_lock(n); > n->keepalive_intv = 10000; > - tipc_node_read_unlock(n); > + tipc_node_write_unlock(n); > for (bearer_id = 0; remains && (bearer_id < MAX_BEARERS); bearer_id++) { > tipc_node_read_lock(n); > le = &n->links[bearer_id]; > -- > 2.35.1
On Tue, 29 Mar 2022 18:12:14 +0200 Niels Dossche wrote: > Currently, n->keepalive_intv is written to while n is locked by a read > lock instead of a write lock. This seems to me to break the atomicity > against other readers. > Change this to a write lock instead to solve the issue. > > Note: > I am currently working on a static analyser to detect missing locks > using type-based static analysis as my master's thesis > in order to obtain my master's degree. > If you would like to have more details, please let me know. > This was a reported case. I manually verified the report by looking > at the code, so that I do not send wrong information or patches. > After concluding that this seems to be a true positive, I created > this patch. I have both compile-tested this patch and runtime-tested > this patch on x86_64. The effect on a running system could be a > potential race condition in exceptional cases. > This issue was found on Linux v5.17. > > Fixes: f5d6c3e5a359 ("tipc: fix node keep alive interval calculation") > Signed-off-by: Niels Dossche <dossche.niels@gmail.com> Looks good, Jon?
On Tue, 2022-03-29 at 18:12 +0200, Niels Dossche wrote: > Currently, n->keepalive_intv is written to while n is locked by a read > lock instead of a write lock. This seems to me to break the atomicity > against other readers. > Change this to a write lock instead to solve the issue. > > Note: > I am currently working on a static analyser to detect missing locks > using type-based static analysis as my master's thesis > in order to obtain my master's degree. > If you would like to have more details, please let me know. > This was a reported case. I manually verified the report by looking > at the code, so that I do not send wrong information or patches. > After concluding that this seems to be a true positive, I created > this patch. I have both compile-tested this patch and runtime-tested > this patch on x86_64. The effect on a running system could be a > potential race condition in exceptional cases. > This issue was found on Linux v5.17. > > Fixes: f5d6c3e5a359 ("tipc: fix node keep alive interval calculation") > Signed-off-by: Niels Dossche <dossche.niels@gmail.com> > --- > net/tipc/node.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/net/tipc/node.c b/net/tipc/node.c > index 6ef95ce565bd..da867ddb93f5 100644 > --- a/net/tipc/node.c > +++ b/net/tipc/node.c > @@ -806,9 +806,9 @@ static void tipc_node_timeout(struct timer_list *t) > /* Initial node interval to value larger (10 seconds), then it will be > * recalculated with link lowest tolerance > */ > - tipc_node_read_lock(n); > + tipc_node_write_lock(n); I agree with Hoang, this should be safe even without write lock, as tipc_node_timeout() is the only function modifying keepalive_intv, and such function is invoked only by a timer, so we are guaranteeded there are no possible concurrent updates... > n->keepalive_intv = 10000; > - tipc_node_read_unlock(n); > + tipc_node_write_unlock(n); > for (bearer_id = 0; remains && (bearer_id < MAX_BEARERS); bearer_id++) { > tipc_node_read_lock(n); ...otherwise we have a similar issue here: a few line below keepalive_intv is updated via tipc_node_calculate_timer(), still under the read lock Thanks! Paolo
On 3/31/22 10:28, Paolo Abeni wrote: > On Tue, 2022-03-29 at 18:12 +0200, Niels Dossche wrote: >> Currently, n->keepalive_intv is written to while n is locked by a read >> lock instead of a write lock. This seems to me to break the atomicity >> against other readers. >> Change this to a write lock instead to solve the issue. >> >> Note: >> I am currently working on a static analyser to detect missing locks >> using type-based static analysis as my master's thesis >> in order to obtain my master's degree. >> If you would like to have more details, please let me know. >> This was a reported case. I manually verified the report by looking >> at the code, so that I do not send wrong information or patches. >> After concluding that this seems to be a true positive, I created >> this patch. I have both compile-tested this patch and runtime-tested >> this patch on x86_64. The effect on a running system could be a >> potential race condition in exceptional cases. >> This issue was found on Linux v5.17. >> >> Fixes: f5d6c3e5a359 ("tipc: fix node keep alive interval calculation") >> Signed-off-by: Niels Dossche <dossche.niels@gmail.com> >> --- >> net/tipc/node.c | 4 ++-- >> 1 file changed, 2 insertions(+), 2 deletions(-) >> >> diff --git a/net/tipc/node.c b/net/tipc/node.c >> index 6ef95ce565bd..da867ddb93f5 100644 >> --- a/net/tipc/node.c >> +++ b/net/tipc/node.c >> @@ -806,9 +806,9 @@ static void tipc_node_timeout(struct timer_list *t) >> /* Initial node interval to value larger (10 seconds), then it will be >> * recalculated with link lowest tolerance >> */ >> - tipc_node_read_lock(n); >> + tipc_node_write_lock(n); > I agree with Hoang, this should be safe even without write lock, as > tipc_node_timeout() is the only function modifying keepalive_intv, and > such function is invoked only by a timer, so we are guaranteeded there > are no possible concurrent updates... > >> n->keepalive_intv = 10000; >> - tipc_node_read_unlock(n); >> + tipc_node_write_unlock(n); >> for (bearer_id = 0; remains && (bearer_id < MAX_BEARERS); bearer_id++) { >> tipc_node_read_lock(n); > ...otherwise we have a similar issue here: a few line below > keepalive_intv is updated via tipc_node_calculate_timer(), still under > the read lock > > Thanks! > > Paolo > Hoang's and Paolo's conclusion is correct. The patch is not needed. ///jon
On 31/03/2022 18:54, Jon Maloy wrote: > Hoang's and Paolo's conclusion is correct. > The patch is not needed. > ///jon > Thank you everyone for commenting on this. Kind regards Niels
diff --git a/net/tipc/node.c b/net/tipc/node.c index 6ef95ce565bd..da867ddb93f5 100644 --- a/net/tipc/node.c +++ b/net/tipc/node.c @@ -806,9 +806,9 @@ static void tipc_node_timeout(struct timer_list *t) /* Initial node interval to value larger (10 seconds), then it will be * recalculated with link lowest tolerance */ - tipc_node_read_lock(n); + tipc_node_write_lock(n); n->keepalive_intv = 10000; - tipc_node_read_unlock(n); + tipc_node_write_unlock(n); for (bearer_id = 0; remains && (bearer_id < MAX_BEARERS); bearer_id++) { tipc_node_read_lock(n); le = &n->links[bearer_id];
Currently, n->keepalive_intv is written to while n is locked by a read lock instead of a write lock. This seems to me to break the atomicity against other readers. Change this to a write lock instead to solve the issue. Note: I am currently working on a static analyser to detect missing locks using type-based static analysis as my master's thesis in order to obtain my master's degree. If you would like to have more details, please let me know. This was a reported case. I manually verified the report by looking at the code, so that I do not send wrong information or patches. After concluding that this seems to be a true positive, I created this patch. I have both compile-tested this patch and runtime-tested this patch on x86_64. The effect on a running system could be a potential race condition in exceptional cases. This issue was found on Linux v5.17. Fixes: f5d6c3e5a359 ("tipc: fix node keep alive interval calculation") Signed-off-by: Niels Dossche <dossche.niels@gmail.com> --- net/tipc/node.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)