Skip to contentSkip to conversation

Venus - Handle race conditions in concurrency

PATCH RESEND Superseded Mansur Alisha Shaik 2 participants 10 messages Aug 7, 2020 – Sep 10, 2020

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover Venus - Handle race conditions in concurrency

The intention of this patchset is to handle race conditions during concurrency usecases like Multiple YouTube browser tabs(approx 50 plus tabs), graphics_Stress, WiFi ON/OFF, Bluetooth ON/OF, and reboot in parallel.

---
Resending the fixes by describing more about the issue and correcting typo errors.

Mansur Alisha Shaik (3):
venus: core: handle race condititon for core ops
venus: core: cancel pending work items in workqueue
venus: handle use after free for iommu_map/iommu_unmap

 drivers/media/platform/qcom/venus/core.c     |  6 +++++-
 drivers/media/platform/qcom/venus/firmware.c | 17 +++++++++++++----
 drivers/media/platform/qcom/venus/hfi.c      |  5 ++++-
 3 files changed, 22 insertions(+), 6 deletions(-)
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member 
of Code Aurora Forum, hosted by The Linux Foundation
#1 lore raw

PATCH RESEND 1/3 venus: core: handle race condititon for core ops

For core ops we are having only write protect but there is no read protect, because of this in multthreading and concurrency, one CPU core is reading without wait which is causing the NULL pointer dereferece crash.

one such scenario is as show below, where in one
core core->ops becoming NULL and in another core
calling core->ops->session_init().

CPU: core-7:

Call trace:
 hfi_session_init+0x180/0x1dc [venus_core]
 vdec_queue_setup+0x9c/0x364 [venus_dec]
 vb2_core_reqbufs+0x1e4/0x368 [videobuf2_common]
 vb2_reqbufs+0x4c/0x64 [videobuf2_v4l2]
 v4l2_m2m_reqbufs+0x50/0x84 [v4l2_mem2mem]
 v4l2_m2m_ioctl_reqbufs+0x2c/0x38 [v4l2_mem2mem]
 v4l_reqbufs+0x4c/0x5c
__video_do_ioctl+0x2b0/0x39c

CPU: core-0:

Call trace:
 venus_shutdown+0x98/0xfc [venus_core]
 venus_sys_error_handler+0x64/0x148 [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
---
 drivers/media/platform/qcom/venus/core.c | 2 +-
 drivers/media/platform/qcom/venus/hfi.c  | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)
drivers/media/platform/qcom/venus/core.c +1 -1
index 203c653..fe99c83 100644
@@ -64,8 +64,8 @@ static void venus_sys_error_handler(struct work_struct *work)
 	pm_runtime_get_sync(core->dev);
 
 	hfi_core_deinit(core, true);
-	hfi_destroy(core);
 	mutex_lock(&core->lock);
+	hfi_destroy(core);
 	venus_shutdown(core);
 
 	pm_runtime_put_sync(core->dev);
drivers/media/platform/qcom/venus/hfi.c +4 -1
index a211eb9..2eeb31f 100644
@@ -195,7 +195,7 @@ EXPORT_SYMBOL_GPL(hfi_session_create);
 int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 {
 	struct venus_core *core = inst->core;
-	const struct hfi_ops *ops = core->ops;
+	const struct hfi_ops *ops;
 	int ret;
 
 	if (inst->state != INST_UNINIT)
@@ -204,10 +204,13 @@ int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 	inst->hfi_codec = to_codec_type(pixfmt);
 	reinit_completion(&inst->done);
 
+	mutex_lock(&core->lock);
+	ops = core->ops;
 	ret = ops->session_init(inst, inst->session_type, inst->hfi_codec);
 	if (ret)
 		return ret;
 
+	mutex_unlock(&core->lock);
 	ret = wait_session_msg(inst);
 	if (ret)
 		return ret;
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member 
of Code Aurora Forum, hosted by The Linux Foundation
#2 lore raw

Re: [RESEND 1/3] venus: core: handle race condititon for core ops

Hi Mansur,

Thanks for the patches!

Quoted 39 lines · Mansur Alisha Shaik

For core ops we are having only write protect but there is no read protect, because of this in multthreading and concurrency, one CPU core is reading without wait which is causing the NULL pointer dereferece crash.

one such scenario is as show below, where in one core core->ops becoming NULL and in another core calling core->ops->session_init().

CPU: core-7:

Call trace:
 hfi_session_init+0x180/0x1dc [venus_core]
 vdec_queue_setup+0x9c/0x364 [venus_dec]
 vb2_core_reqbufs+0x1e4/0x368 [videobuf2_common]
 vb2_reqbufs+0x4c/0x64 [videobuf2_v4l2]
 v4l2_m2m_reqbufs+0x50/0x84 [v4l2_mem2mem]
 v4l2_m2m_ioctl_reqbufs+0x2c/0x38 [v4l2_mem2mem]
 v4l_reqbufs+0x4c/0x5c
__video_do_ioctl+0x2b0/0x39c

CPU: core-0:

Call trace:
 venus_shutdown+0x98/0xfc [venus_core]
 venus_sys_error_handler+0x64/0x148 [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
---
 drivers/media/platform/qcom/venus/core.c | 2 +-
 drivers/media/platform/qcom/venus/hfi.c  | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)
drivers/media/platform/qcom/venus/core.c +0 -0
index 203c653..fe99c83 100644

drivers/media/platform/qcom/venus/core.c static void venus_sys_error_handler(struct work_struct *work) · line 68

Quoted 6 lines
drivers/media/platform/qcom/venus/core.c +1 -1
@@ -64,8 +64,8 @@ static void venus_sys_error_handler(struct work_struct *work)
 	pm_runtime_get_sync(core->dev);
 
 	hfi_core_deinit(core, true);
-	hfi_destroy(core);
 	mutex_lock(&core->lock);
+	hfi_destroy(core);

As my recovery fixes [1] touches this part also, could you please apply them on top of yours and re-test?

Otherwise this patch looks good to me.

[1] https://www.spinics.net/lists/linux-arm-msm/msg70092.html

Quoted 31 lines
drivers/media/platform/qcom/venus/core.c +0 -0
 	venus_shutdown(core);
 
 	pm_runtime_put_sync(core->dev);
drivers/media/platform/qcom/venus/hfi.c +4 -1
index a211eb9..2eeb31f 100644
@@ -195,7 +195,7 @@ EXPORT_SYMBOL_GPL(hfi_session_create);
 int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 {
 	struct venus_core *core = inst->core;
-	const struct hfi_ops *ops = core->ops;
+	const struct hfi_ops *ops;
 	int ret;
 
 	if (inst->state != INST_UNINIT)
@@ -204,10 +204,13 @@ int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 	inst->hfi_codec = to_codec_type(pixfmt);
 	reinit_completion(&inst->done);
 
+	mutex_lock(&core->lock);
+	ops = core->ops;
 	ret = ops->session_init(inst, inst->session_type, inst->hfi_codec);
 	if (ret)
 		return ret;
 
+	mutex_unlock(&core->lock);
 	ret = wait_session_msg(inst);
 	if (ret)
 		return ret;
Signature · 2 lines
regards,
Stan
#3 lore raw

Re: [RESEND 1/3] venus: core: handle race condititon for core ops

Hi Mansur,

Quoted 38 lines · Stanimir Varbanov

Hi Mansur,

Thanks for the patches!

Quoted 33 lines · Mansur Alisha Shaik

For core ops we are having only write protect but there is no read protect, because of this in multthreading and concurrency, one CPU core is reading without wait which is causing the NULL pointer dereferece crash.

one such scenario is as show below, where in one core core->ops becoming NULL and in another core calling core->ops->session_init().

CPU: core-7:

Call trace:
 hfi_session_init+0x180/0x1dc [venus_core]
 vdec_queue_setup+0x9c/0x364 [venus_dec]
 vb2_core_reqbufs+0x1e4/0x368 [videobuf2_common]
 vb2_reqbufs+0x4c/0x64 [videobuf2_v4l2]
 v4l2_m2m_reqbufs+0x50/0x84 [v4l2_mem2mem]
 v4l2_m2m_ioctl_reqbufs+0x2c/0x38 [v4l2_mem2mem]
 v4l_reqbufs+0x4c/0x5c
__video_do_ioctl+0x2b0/0x39c

CPU: core-0:

Call trace:
 venus_shutdown+0x98/0xfc [venus_core]
 venus_sys_error_handler+0x64/0x148 [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
---
 drivers/media/platform/qcom/venus/core.c | 2 +-
 drivers/media/platform/qcom/venus/hfi.c  | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)

See below comment, otherwise:

Quoted 8 lines
Quoted 8 lines
drivers/media/platform/qcom/venus/core.c +0 -0
index 203c653..fe99c83 100644
@@ -64,8 +64,8 @@ static void venus_sys_error_handler(struct work_struct *work)
 	pm_runtime_get_sync(core->dev);
 
Quoted 7 lines

drivers/media/platform/qcom/venus/core.c static void venus_sys_error_handler(struct work_struct *work) · line 68

Quoted 4 lines
drivers/media/platform/qcom/venus/core.c +1 -1
 	hfi_core_deinit(core, true);
-	hfi_destroy(core);
 	mutex_lock(&core->lock);
+	hfi_destroy(core);

As my recovery fixes [1] touches this part also, could you please apply them on top of yours and re-test?

I'll drop above chunk from the patch because it is already taken into account in my recovery fixes series and queue up the patch for v5.10.

Quoted 37 lines

Otherwise this patch looks good to me.

[1] https://www.spinics.net/lists/linux-arm-msm/msg70092.html

Quoted 31 lines
drivers/media/platform/qcom/venus/core.c +0 -0
 	venus_shutdown(core);
 
 	pm_runtime_put_sync(core->dev);
drivers/media/platform/qcom/venus/hfi.c +4 -1
index a211eb9..2eeb31f 100644
@@ -195,7 +195,7 @@ EXPORT_SYMBOL_GPL(hfi_session_create);
 int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 {
 	struct venus_core *core = inst->core;
-	const struct hfi_ops *ops = core->ops;
+	const struct hfi_ops *ops;
 	int ret;
 
 	if (inst->state != INST_UNINIT)
@@ -204,10 +204,13 @@ int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 	inst->hfi_codec = to_codec_type(pixfmt);
 	reinit_completion(&inst->done);
 
+	mutex_lock(&core->lock);
+	ops = core->ops;
 	ret = ops->session_init(inst, inst->session_type, inst->hfi_codec);
 	if (ret)
 		return ret;
 
+	mutex_unlock(&core->lock);
 	ret = wait_session_msg(inst);
 	if (ret)
 		return ret;
Signature · 2 lines
regards,
Stan
#4 lore raw

Re: [RESEND 1/3] venus: core: handle race condititon for core ops

Quoted 63 lines · Stanimir Varbanov

Hi Mansur,

Quoted 38 lines · Stanimir Varbanov

Hi Mansur,

Thanks for the patches!

Quoted 33 lines · Mansur Alisha Shaik

For core ops we are having only write protect but there is no read protect, because of this in multthreading and concurrency, one CPU core is reading without wait which is causing the NULL pointer dereferece crash.

one such scenario is as show below, where in one core core->ops becoming NULL and in another core calling core->ops->session_init().

CPU: core-7:

Call trace:
 hfi_session_init+0x180/0x1dc [venus_core]
 vdec_queue_setup+0x9c/0x364 [venus_dec]
 vb2_core_reqbufs+0x1e4/0x368 [videobuf2_common]
 vb2_reqbufs+0x4c/0x64 [videobuf2_v4l2]
 v4l2_m2m_reqbufs+0x50/0x84 [v4l2_mem2mem]
 v4l2_m2m_ioctl_reqbufs+0x2c/0x38 [v4l2_mem2mem]
 v4l_reqbufs+0x4c/0x5c
__video_do_ioctl+0x2b0/0x39c

CPU: core-0:

Call trace:
 venus_shutdown+0x98/0xfc [venus_core]
 venus_sys_error_handler+0x64/0x148 [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
---
 drivers/media/platform/qcom/venus/core.c | 2 +-
 drivers/media/platform/qcom/venus/hfi.c  | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)

See below comment, otherwise:

Quoted 9 lines
Quoted 9 lines
drivers/media/platform/qcom/venus/core.c +0 -0
index 203c653..fe99c83 100644
@@ -64,8 +64,8 @@ static void venus_sys_error_handler(struct 
 work_struct *work)
 	pm_runtime_get_sync(core->dev);
 
Quoted 7 lines

drivers/media/platform/qcom/venus/core.c static void venus_sys_error_handler(struct work_struct *work) · line 68

Quoted 4 lines
drivers/media/platform/qcom/venus/core.c +1 -1
 	hfi_core_deinit(core, true);
-	hfi_destroy(core);
 	mutex_lock(&core->lock);
+	hfi_destroy(core);

As my recovery fixes [1] touches this part also, could you please apply them on top of yours and re-test?

Quoted 3 lines

I'll drop above chunk from the patch because it is already taken into account in my recovery fixes series and queue up the patch for v5.10.

yes, you can drop. I have validated these patches on top of your recovery patch series. I will push V2 with dependency on "venus - recovery from frimware crash" series (https://lore.kernel.org/patchwork/project/lkml/list/?series=455962)

Quoted 38 lines
Quoted 38 lines

Otherwise this patch looks good to me.

[1] https://www.spinics.net/lists/linux-arm-msm/msg70092.html

Quoted 32 lines
drivers/media/platform/qcom/venus/core.c +0 -0
 	venus_shutdown(core);
 
drivers/media/platform/qcom/venus/core.c +0 -0
 	pm_runtime_put_sync(core->dev);
drivers/media/platform/qcom/venus/hfi.c +4 -1
index a211eb9..2eeb31f 100644
@@ -195,7 +195,7 @@ EXPORT_SYMBOL_GPL(hfi_session_create);
 int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 {
 	struct venus_core *core = inst->core;
-	const struct hfi_ops *ops = core->ops;
+	const struct hfi_ops *ops;
 	int ret;
 
 	if (inst->state != INST_UNINIT)
@@ -204,10 +204,13 @@ int hfi_session_init(struct venus_inst *inst, 
 u32 pixfmt)
 	inst->hfi_codec = to_codec_type(pixfmt);
 	reinit_completion(&inst->done);
 
+	mutex_lock(&core->lock);
+	ops = core->ops;
 	ret = ops->session_init(inst, inst->session_type, inst->hfi_codec);
 	if (ret)
 		return ret;
 
+	mutex_unlock(&core->lock);
 	ret = wait_session_msg(inst);
 	if (ret)

return ret;

#5 lore raw

Re: [RESEND 1/3] venus: core: handle race condititon for core ops

In reply to Mansur Alisha Shaik

Hi Mansur,

Quoted 5 lines · Mansur Alisha Shaik

For core ops we are having only write protect but there is no read protect, because of this in multthreading and concurrency, one CPU core is reading without wait which is causing the NULL pointer dereferece crash.

Quoted 7 lines

one such scenario is as show below, where in one core core->ops becoming NULL and in another core calling core->ops->session_init().

CPU: core-7:

Call trace:
 hfi_session_init+0x180/0x1dc [venus_core]

I thought more on this issue. I think we have to return error from hfi_session_init() in the case when the driver is in system-error-handler. Infact all userspace ioctls must end up with error while we are in recovery state. What do you think?

Quoted 64 lines
 vdec_queue_setup+0x9c/0x364 [venus_dec]
 vb2_core_reqbufs+0x1e4/0x368 [videobuf2_common]
 vb2_reqbufs+0x4c/0x64 [videobuf2_v4l2]
 v4l2_m2m_reqbufs+0x50/0x84 [v4l2_mem2mem]
 v4l2_m2m_ioctl_reqbufs+0x2c/0x38 [v4l2_mem2mem]
 v4l_reqbufs+0x4c/0x5c
__video_do_ioctl+0x2b0/0x39c

CPU: core-0:

Call trace:
 venus_shutdown+0x98/0xfc [venus_core]
 venus_sys_error_handler+0x64/0x148 [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
---
 drivers/media/platform/qcom/venus/core.c | 2 +-
 drivers/media/platform/qcom/venus/hfi.c  | 5 ++++-
 2 files changed, 5 insertions(+), 2 deletions(-)
drivers/media/platform/qcom/venus/core.c +1 -1
index 203c653..fe99c83 100644
@@ -64,8 +64,8 @@ static void venus_sys_error_handler(struct work_struct *work)
 	pm_runtime_get_sync(core->dev);
 
 	hfi_core_deinit(core, true);
-	hfi_destroy(core);
 	mutex_lock(&core->lock);
+	hfi_destroy(core);
 	venus_shutdown(core);
 
 	pm_runtime_put_sync(core->dev);
drivers/media/platform/qcom/venus/hfi.c +4 -1
index a211eb9..2eeb31f 100644
@@ -195,7 +195,7 @@ EXPORT_SYMBOL_GPL(hfi_session_create);
 int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 {
 	struct venus_core *core = inst->core;
-	const struct hfi_ops *ops = core->ops;
+	const struct hfi_ops *ops;
 	int ret;
 
 	if (inst->state != INST_UNINIT)
@@ -204,10 +204,13 @@ int hfi_session_init(struct venus_inst *inst, u32 pixfmt)
 	inst->hfi_codec = to_codec_type(pixfmt);
 	reinit_completion(&inst->done);
 
+	mutex_lock(&core->lock);
+	ops = core->ops;
 	ret = ops->session_init(inst, inst->session_type, inst->hfi_codec);
 	if (ret)
 		return ret;
 
+	mutex_unlock(&core->lock);
 	ret = wait_session_msg(inst);
 	if (ret)
 		return ret;
Signature · 2 lines
regards,
Stan
#6 lore raw

PATCH RESEND 2/3 venus: core: cancel pending work items in workqueue

In reply to Mansur Alisha Shaik

In concurrency usecase and reboot scenario we are
observing race condition and seeing NULL pointer
dereference crash. In shutdown path and system
recovery path we are destroying the same mutex
hence seeing crash.

This case is handled by mutex protection and
cancel delayed work items in work queue.

Below is the call trace for the crash

Call trace:
 venus_remove+0xdc/0xec [venus_core]
 venus_core_shutdown+0x1c/0x34 [venus_core]
 platform_drv_shutdown+0x28/0x34
 device_shutdown+0x154/0x1fc
 kernel_restart_prepare+0x40/0x4c
 kernel_restart+0x1c/0x64

Call trace:
 mutex_lock+0x34/0x60
 venus_hfi_destroy+0x28/0x98 [venus_core]
 hfi_destroy+0x1c/0x28 [venus_core]
 venus_sys_error_handler+0x60/0x14c [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18
---
 drivers/media/platform/qcom/venus/core.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/media/platform/qcom/venus/core.c +4 -0
index fe99c83..41a293e 100644
@@ -312,6 +312,8 @@ static int venus_remove(struct platform_device *pdev)
 	struct device *dev = core->dev;
 	int ret;
 
+	cancel_delayed_work_sync(&core->work);
+
 	ret = pm_runtime_get_sync(dev);
 	WARN_ON(ret < 0);
 
@@ -329,7 +331,9 @@ static int venus_remove(struct platform_device *pdev)
 	if (pm_ops->core_put)
 		pm_ops->core_put(dev);
 
+	mutex_lock(&core->lock);
 	hfi_destroy(core);
+	mutex_unlock(&core->lock);
 
 	icc_put(core->video_path);
 	icc_put(core->cpucfg_path);
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member 
of Code Aurora Forum, hosted by The Linux Foundation
#7 lore raw

PATCH RESEND 3/3 venus: handle use after free for iommu_map/iommu_unmap

In reply to Mansur Alisha Shaik

In concurrency usecase and reboot scenario we are trying to map fw.iommu_domain which is already unmapped during shutdown. This is causing NULL pointer dereference crash.

This case is handled by adding necessary checks.

Call trace:
 __iommu_map+0x4c/0x348
 iommu_map+0x5c/0x70
 venus_boot+0x184/0x230 [venus_core]
 venus_sys_error_handler+0xa0/0x14c [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18
---
 drivers/media/platform/qcom/venus/firmware.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
drivers/media/platform/qcom/venus/firmware.c +13 -4
index 8801a6a..c427e88 100644
@@ -171,9 +171,14 @@ static int venus_shutdown_no_tz(struct venus_core *core)
 
 	iommu = core->fw.iommu_domain;
 
-	unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
-	if (unmapped != mapped)
-		dev_err(dev, "failed to unmap firmware\n");
+	if (core->fw.mapped_mem_size && iommu) {
+		unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
+
+		if (unmapped != mapped)
+			dev_err(dev, "failed to unmap firmware\n");
+		else
+			core->fw.mapped_mem_size = 0;
+	}
 
 	return 0;
 }
@@ -288,7 +293,11 @@ void venus_firmware_deinit(struct venus_core *core)
 	iommu = core->fw.iommu_domain;
 
 	iommu_detach_device(iommu, core->fw.dev);
-	iommu_domain_free(iommu);
+
+	if (core->fw.iommu_domain) {
+		iommu_domain_free(iommu);
+		core->fw.iommu_domain = NULL;
+	}
 
 	platform_device_unregister(to_platform_device(core->fw.dev));
 }
Signature · 2 lines
QUALCOMM INDIA, on behalf of Qualcomm Innovation Center, Inc. is a member 
of Code Aurora Forum, hosted by The Linux Foundation
#8 lore raw

Re: [RESEND 3/3] venus: handle use after free for iommu_map/iommu_unmap

Hi,

Quoted 2 lines · Mansur Alisha Shaik

In concurrency usecase and reboot scenario we are trying to map fw.iommu_domain which is already unmapped during

I guess you want to say "to unmap iommu domain which is already unmapped" ?

Quoted 55 lines

shutdown. This is causing NULL pointer dereference crash.

This case is handled by adding necessary checks.

Call trace:
 __iommu_map+0x4c/0x348
 iommu_map+0x5c/0x70
 venus_boot+0x184/0x230 [venus_core]
 venus_sys_error_handler+0xa0/0x14c [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18
---
 drivers/media/platform/qcom/venus/firmware.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
drivers/media/platform/qcom/venus/firmware.c +13 -4
index 8801a6a..c427e88 100644
@@ -171,9 +171,14 @@ static int venus_shutdown_no_tz(struct venus_core *core)
 
 	iommu = core->fw.iommu_domain;
 
-	unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
-	if (unmapped != mapped)
-		dev_err(dev, "failed to unmap firmware\n");
+	if (core->fw.mapped_mem_size && iommu) {
+		unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
+
+		if (unmapped != mapped)
+			dev_err(dev, "failed to unmap firmware\n");
+		else
+			core->fw.mapped_mem_size = 0;
+	}
 
 	return 0;
 }
@@ -288,7 +293,11 @@ void venus_firmware_deinit(struct venus_core *core)
 	iommu = core->fw.iommu_domain;
 
 	iommu_detach_device(iommu, core->fw.dev);
-	iommu_domain_free(iommu);
+
+	if (core->fw.iommu_domain) {
+		iommu_domain_free(iommu);
+		core->fw.iommu_domain = NULL;
+	}
 
 	platform_device_unregister(to_platform_device(core->fw.dev));
 }
Signature · 2 lines
regards,
Stan
#9 lore raw

Re: [RESEND 3/3] venus: handle use after free for iommu_map/iommu_unmap

In reply to Mansur Alisha Shaik
Quoted 44 lines · Mansur Alisha Shaik

In concurrency usecase and reboot scenario we are trying to map fw.iommu_domain which is already unmapped during shutdown. This is causing NULL pointer dereference crash.

This case is handled by adding necessary checks.

Call trace:
 __iommu_map+0x4c/0x348
 iommu_map+0x5c/0x70
 venus_boot+0x184/0x230 [venus_core]
 venus_sys_error_handler+0xa0/0x14c [venus_core]
 process_one_work+0x210/0x3d0
 worker_thread+0x248/0x3f4
 kthread+0x11c/0x12c
 ret_from_fork+0x10/0x18
---
 drivers/media/platform/qcom/venus/firmware.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
drivers/media/platform/qcom/venus/firmware.c +8 -3
index 8801a6a..c427e88 100644
@@ -171,9 +171,14 @@ static int venus_shutdown_no_tz(struct venus_core *core)
 
 	iommu = core->fw.iommu_domain;
 
-	unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
-	if (unmapped != mapped)
-		dev_err(dev, "failed to unmap firmware\n");
+	if (core->fw.mapped_mem_size && iommu) {
+		unmapped = iommu_unmap(iommu, VENUS_FW_START_ADDR, mapped);
+
+		if (unmapped != mapped)
+			dev_err(dev, "failed to unmap firmware\n");
+		else
+			core->fw.mapped_mem_size = 0;
+	}
 
 	return 0;
 }

drivers/media/platform/qcom/venus/firmware.c void venus_firmware_deinit(struct venus_core *core) · line 297

Quoted 6 lines
drivers/media/platform/qcom/venus/firmware.c +2 -1
@@ -288,7 +293,11 @@ void venus_firmware_deinit(struct venus_core *core)
 	iommu = core->fw.iommu_domain;
 
 	iommu_detach_device(iommu, core->fw.dev);
-	iommu_domain_free(iommu);
+
+	if (core->fw.iommu_domain) {

why not just ?

if (iommu)

Quoted 7 lines
drivers/media/platform/qcom/venus/firmware.c +3 -0
+		iommu_domain_free(iommu);
+		core->fw.iommu_domain = NULL;
+	}
 
 	platform_device_unregister(to_platform_device(core->fw.dev));
 }
Signature · 2 lines
regards,
Stan