net/mlx5: Fix OOB access and stack information leak in PTP event handling
Subsystems (4)
PATCH v3 1/1 net/mlx5: Fix OOB access and stack information leak in PTP event handling
In mlx5_pps_event(), several critical issues were identified during review by Sashiko:
1. The 'pin' index from the hardware event was used without bounds checking to index 'pin_config' and 'pps_info->start', leading to potential out-of-bounds memory access.
2. 'ptp_event' was not zero-initialized. Since it contains a union, assigning a timestamp partially leaves the 'ts_raw' field with uninitialized stack memory, which can leak kernel data or corrupt time sync logic in hardpps().
3. A NULL 'pin_config' could be dereferenced if initialization failed.
4. 'clock->ptp' could be NULL if ptp_clock_register() failed.
Fix these by zero-initializing the event struct, adding a bounds check against n_pins, and adding appropriate NULL guards.
---
v3:
- Fix union corruption by using a local timestamp variable [Sashiko].
- Validate pin index against n_pins with WARN_ON_ONCE [Carolina].
- Remove redundant pin < 0 check and cleanup TODO comment.
v2:
- Zero-initialize ptp_event to prevent stack information leak [Sashiko].
- Add bounds check for hardware pin index to prevent OOB access [Sashiko].
- Add NULL guard for pin_config to handle initialization failures [Sashiko].
- Add NULL check for clock->ptp as originally intended.
.../net/ethernet/mellanox/mlx5/core/lib/clock.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-)
drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +12 -5
@@ -1164,16 +1164,22 @@ static int mlx5_pps_event(struct notifier_block *nb, pps_nb);struct mlx5_core_dev *mdev = clock_state->mdev;struct mlx5_clock *clock = mdev->clock; -struct ptp_clock_event ptp_event; +struct ptp_clock_event ptp_event = {};struct mlx5_eqe *eqe = data;int pin = eqe->data.pps.pin;unsigned long flags;u64 ns; +if (!clock->ptp_info.pin_config) +return NOTIFY_OK ; + +if (WARN_ON_ONCE (pin >= clock->ptp_info.n_pins)) +return NOTIFY_OK ; +switch (clock->ptp_info.pin_config[pin].func) {case PTP_PF_EXTTS : ptp_event.index = pin; - ptp_event.timestamp =mlx5_real_time_mode (mdev) ? + ns =mlx5_real_time_mode (mdev) ?mlx5_real_time_cyc2time (clock,be64_to_cpu (eqe->data.pps.time_stamp)) : mlx5_timecounter_cyc2time(clock,
@@ -1181,12 +1187,13 @@ static int mlx5_pps_event(struct notifier_block *nb,if (clock->pps_info.enabled) { ptp_event.type =PTP_CLOCK_PPSUSR ; ptp_event.pps_times.ts_real = -ns_to_timespec64 (ptp_event.timestamp); +ns_to_timespec64 (ns); }else { ptp_event.type =PTP_CLOCK_EXTTS ; + ptp_event.timestamp = ns; } -/* TODOL clock->ptp can be NULL if ptp_clock_register fails */ -ptp_clock_event (clock->ptp, &ptp_event); +if (clock->ptp) +ptp_clock_event (clock->ptp, &ptp_event);break ;case PTP_PF_PEROUT :if (clock->shared) {
Signature · 1 line
2.43.0
Quoted 43 lines · Prathamesh Deshpande
In
mlx5_pps_event(), several critical issues were identified during review by Sashiko:1. The 'pin' index from the hardware event was used without bounds checking to index '
pin_config' and 'pps_info->start', leading to potential out-of-bounds memory access.
2. 'ptp_event' was not zero-initialized. Since it contains a union, assigning a timestamp partially leaves the 'ts_raw' field with uninitialized stack memory, which can leak kernel data or corrupt time sync logic inhardpps().
3. A NULL 'pin_config' could be dereferenced if initialization failed.
4. 'clock->ptp' could be NULL ifptp_clock_register()failed.Fix these by zero-initializing the event struct, adding a bounds check against
n_pins, and adding appropriate NULL guards.Fixes 7c39afb394c7 ("net/mlx5: PTP code migration to driver core section")Suggested-by Carolina Jubran <[email protected]>Signed-off-by Prathamesh Deshpande <[email protected]>---
v3:
- Fix union corruption by using a local timestamp variable [Sashiko].
- Validate pin index againstn_pinswithWARN_ON_ONCE[Carolina].
- Remove redundant pin < 0 check and cleanup TODO comment.
v2:
- Zero-initializeptp_eventto prevent stack information leak [Sashiko].
- Add bounds check for hardware pin index to prevent OOB access [Sashiko].
- Add NULL guard forpin_configto handle initialization failures [Sashiko].
- Add NULL check forclock->ptpas originally intended..../net/ethernet/mellanox/mlx5/core/lib/clock.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-)drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +1
-1@@ -1164,16 +1164,22 @@ static int mlx5_pps_event(struct notifier_block *nb, pps_nb);struct mlx5_core_dev *mdev = clock_state->mdev;struct mlx5_clock *clock = mdev->clock; -struct ptp_clock_event ptp_event; +struct ptp_clock_event ptp_event = {};
drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c static int mlx5_pps_event(struct notifier_block *nb, · line 1177
Quoted 10 lines
drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +5
-0struct mlx5_eqe *eqe = data;int pin = eqe->data.pps.pin;unsigned long flags;u64 ns; +if (!clock->ptp_info.pin_config) +return NOTIFY_OK ; + +if (WARN_ON_ONCE (pin >= clock->ptp_info.n_pins)) +return NOTIFY_OK ;
Sorry if my previous comment wasn't clear enough.
The firmware will never report a pin higher than n_pins, thats not the concern
here. if future hardware reports n_pins > 8, checking against n_pins would still
allow OOB access on those arrays. The check should compare against MAX_PIN_NUM
instead, since thats the actual hard limit of the driver's data structures. and if a new
device supports more than 8 pins, the WARN_ON_ONCE would let us know we need
to update the driver.
Thanks,
Carolina
Hi Carolina,
Thanks for the feedback. I have just submitted v4 which addresses this by checking the pin index against MAX_PIN_NUM.
Thanks,
Prathamesh
Quoted 82 lines · Carolina Jubran
Quoted 43 lines · Prathamesh Deshpande
In
mlx5_pps_event(), several critical issues were identified during review by Sashiko:1. The 'pin' index from the hardware event was used without bounds checking to index '
pin_config' and 'pps_info->start', leading to potential out-of-bounds memory access.
2. 'ptp_event' was not zero-initialized. Since it contains a union, assigning a timestamp partially leaves the 'ts_raw' field with uninitialized stack memory, which can leak kernel data or corrupt time sync logic inhardpps().
3. A NULL 'pin_config' could be dereferenced if initialization failed.
4. 'clock->ptp' could be NULL ifptp_clock_register()failed.Fix these by zero-initializing the event struct, adding a bounds check against
n_pins, and adding appropriate NULL guards.Fixes 7c39afb394c7 ("net/mlx5: PTP code migration to driver core section")Suggested-by Carolina Jubran <[email protected]>Signed-off-by Prathamesh Deshpande <[email protected]>---
v3:
- Fix union corruption by using a local timestamp variable [Sashiko].
- Validate pin index againstn_pinswithWARN_ON_ONCE[Carolina].
- Remove redundant pin < 0 check and cleanup TODO comment.
v2:
- Zero-initializeptp_eventto prevent stack information leak [Sashiko].
- Add bounds check for hardware pin index to prevent OOB access [Sashiko].
- Add NULL guard forpin_configto handle initialization failures [Sashiko].
- Add NULL check forclock->ptpas originally intended..../net/ethernet/mellanox/mlx5/core/lib/clock.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-)drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +1
-1@@ -1164,16 +1164,22 @@ static int mlx5_pps_event(struct notifier_block *nb, pps_nb);struct mlx5_core_dev *mdev = clock_state->mdev;struct mlx5_clock *clock = mdev->clock; -struct ptp_clock_event ptp_event; +struct ptp_clock_event ptp_event = {};drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c static int mlx5_pps_event(struct notifier_block *nb, · line 1177
Quoted 10 lines
drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +5
-0struct mlx5_eqe *eqe = data;int pin = eqe->data.pps.pin;unsigned long flags;u64 ns; +if (!clock->ptp_info.pin_config) +return NOTIFY_OK ; + +if (WARN_ON_ONCE (pin >= clock->ptp_info.n_pins)) +return NOTIFY_OK ;Sorry if my previous comment wasn't clear enough.
The firmware will never report a pin higher than
n_pins, thats not the concernhere. if future hardware reports
n_pins> 8, checking againstn_pinswould stillallow OOB access on those arrays. The check should compare against
MAX_PIN_NUMinstead, since thats the actual hard limit of the driver's data structures. and if a new
device supports more than 8 pins, the
WARN_ON_ONCEwould let us know we needto update the driver.
Thanks,
Carolina
Signature · 2 lines
Thanks and Regards, Prathamesh Deshpande