While by default max_autoclose equals to INT_MAX / HZ, one may set net.sctp.max_autoclose to UINT_MAX. There is code in sctp_association_init() that can consequently trigger overflow.
Cc: stable@vger.kernel.org Fixes: 9f70f46bd4c7 ("sctp: properly latch and use autoclose value from sock to association") Signed-off-by: Nikolay Kuratov kniv@yandex-team.ru --- net/sctp/associola.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/sctp/associola.c b/net/sctp/associola.c index c45c192b7878..0b0794f164cf 100644 --- a/net/sctp/associola.c +++ b/net/sctp/associola.c @@ -137,7 +137,8 @@ static struct sctp_association *sctp_association_init( = 5 * asoc->rto_max;
asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] = asoc->sackdelay; - asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] = sp->autoclose * HZ; + asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] = + (unsigned long)sp->autoclose * HZ;
/* Initializes the timers */ for (i = SCTP_EVENT_TIMEOUT_NONE; i < SCTP_NUM_TIMEOUT_TYPES; ++i)
On Thu, Dec 19, 2024 at 11:21 AM Nikolay Kuratov kniv@yandex-team.ru wrote:
While by default max_autoclose equals to INT_MAX / HZ, one may set net.sctp.max_autoclose to UINT_MAX. There is code in sctp_association_init() that can consequently trigger overflow.
Cc: stable@vger.kernel.org Fixes: 9f70f46bd4c7 ("sctp: properly latch and use autoclose value from sock to association") Signed-off-by: Nikolay Kuratov kniv@yandex-team.ru
net/sctp/associola.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/sctp/associola.c b/net/sctp/associola.c index c45c192b7878..0b0794f164cf 100644 --- a/net/sctp/associola.c +++ b/net/sctp/associola.c @@ -137,7 +137,8 @@ static struct sctp_association *sctp_association_init( = 5 * asoc->rto_max;
asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] = asoc->sackdelay;
asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] = sp->autoclose * HZ;
asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] =(unsigned long)sp->autoclose * HZ; /* Initializes the timers */ for (i = SCTP_EVENT_TIMEOUT_NONE; i < SCTP_NUM_TIMEOUT_TYPES; ++i)-- 2.34.1
Acked-by: Xin Long lucien.xin@gmail.com
Hello:
This patch was applied to netdev/net.git (main) by Jakub Kicinski kuba@kernel.org:
On Thu, 19 Dec 2024 19:21:14 +0300 you wrote:
While by default max_autoclose equals to INT_MAX / HZ, one may set net.sctp.max_autoclose to UINT_MAX. There is code in sctp_association_init() that can consequently trigger overflow.
Cc: stable@vger.kernel.org Fixes: 9f70f46bd4c7 ("sctp: properly latch and use autoclose value from sock to association") Signed-off-by: Nikolay Kuratov kniv@yandex-team.ru
[...]
Here is the summary with links: - net/sctp: Prevent autoclose integer overflow in sctp_association_init() https://git.kernel.org/netdev/net/c/4e86729d1ff3
You are awesome, thank you!
On Thu, 19 Dec 2024 19:21:14 +0300 Nikolay Kuratov kniv@yandex-team.ru wrote:
While by default max_autoclose equals to INT_MAX / HZ, one may set net.sctp.max_autoclose to UINT_MAX. There is code in sctp_association_init() that can consequently trigger overflow.
Cc: stable@vger.kernel.org Fixes: 9f70f46bd4c7 ("sctp: properly latch and use autoclose value from sock to association") Signed-off-by: Nikolay Kuratov kniv@yandex-team.ru
net/sctp/associola.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/sctp/associola.c b/net/sctp/associola.c index c45c192b7878..0b0794f164cf 100644 --- a/net/sctp/associola.c +++ b/net/sctp/associola.c @@ -137,7 +137,8 @@ static struct sctp_association *sctp_association_init( = 5 * asoc->rto_max; asoc->timeouts[SCTP_EVENT_TIMEOUT_SACK] = asoc->sackdelay;
- asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] = sp->autoclose * HZ;
- asoc->timeouts[SCTP_EVENT_TIMEOUT_AUTOCLOSE] =
(unsigned long)sp->autoclose * HZ;
That doesn't fix 32bit systems.
Looking through sctp/structs.h there are a lot of 'long' used for timeouts. it can't be right that any of these change size between 32bit and 64bit. So they should either be __u32 or __u64 (or similar).
David
/* Initializes the timers */ for (i = SCTP_EVENT_TIMEOUT_NONE; i < SCTP_NUM_TIMEOUT_TYPES; ++i)
David Laight wrote:
That doesn't fix 32bit systems.
Looking through sctp/structs.h there are a lot of 'long' used for timeouts. it can't be right that any of these change size between 32bit and 64bit. So they should either be __u32 or __u64 (or similar).
I guess all the 'long' timeout values - not only in sctp, this concerns other modules in general - follow the jiffies thing itself which is declared as 'unsigned long'.
linux-stable-mirror@lists.linaro.org
-
David Laight -
Fedor Pchelkin -
Nikolay Kuratov -
patchwork-bot+netdevbpf＠kernel.org -
Xin Long