It was possible for an unprivileged task to invoke any function with
privilege by passing it as a parameter to MPU_xTaskCreate,
MPU_xTaskCreateStatic, MPU_xTimerCreate, MPU_xTimerCreateStatic, or
MPU_xTimerPendFunctionCall.
This commit ensures that MPU_xTaskCreate and MPU_xTaskCreateStatic can
only create unprivileged tasks. It also removes the following APIs:
1. MPU_xTimerCreate
2. MPU_xTimerCreateStatic
3. MPU_xTimerPendFunctionCall
We thank Huazhong University of Science and Technology for reporting
this issue.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
It was possible for a third party that had already independently gained
the ability to execute injected code to achieve further privilege
escalation by branching directly inside a FreeRTOS MPU API wrapper
function with a manually crafted stack frame. This commit removes the
local stack variable `xRunningPrivileged` so that a manually crafted
stack frame cannot be used for privilege escalation by branching
directly inside a FreeRTOS MPU API wrapper.
We thank Certibit Consulting, LLC, Huazhong University of Science and
Technology and the SecLab team at Northeastern University for reporting
this issue.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
ARMv7-M allows overlapping MPU regions. When 2 MPU regions overlap, the
MPU configuration of the higher numbered MPU region is applied. For
example, if a memory area is covered by 2 MPU regions 0 and 1, the
memory permissions for MPU region 1 are applied.
We use 5 MPU regions for kernel code and kernel data protections and
leave the remaining for the application writer. We were using lowest
numbered MPU regions (0-4) for kernel protections and leaving the
remaining for the application writer. The application writer could
configure those higher numbered MPU regions to override kernel
protections.
This commit changes the code to use highest numbered MPU regions for
kernel protections and leave the remaining for the application writer.
This ensures that the application writer cannot override kernel
protections.
We thank the SecLab team at Northeastern University for reporting this
issue.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
The secure side context management code now checks that the secure
context being saved or restored belongs to the task being switched-out
or switched-in respectively.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
This commit improves ARMv8-M security by pre-allocating secure-side task
context structures and changing how tasks reference a secure-side
context structure when calling a secure function. The new configuration
constant secureconfigMAX_SECURE_CONTEXTS sets the number of secure
context structures to pre-allocate. secureconfigMAX_SECURE_CONTEXTS
defaults to 8 if left undefined.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
This change removes the FreeRTOS System Calls (aka MPU wrappers) for the
following kernel APIs:
- xTaskCreateRestricted
- xTaskCreateRestrictedStatic
- vTaskAllocateMPURegions
A system call allows an unprivileged task to execute a kernel API which
is otherwise accessible to privileged software only. The above 3 APIs
can create a new task with a different MPU configuration or alter the
MPU configuration of an existing task. This an be (mis)used by an
unprivileged task to grant itself access to a region which it does not
have access to.
Removing the system calls for these APIs ensures that an unprivileged
task cannot execute this APIs. If an unprivileged task attempts to
execute any of these API, it will result in a Memory Fault.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
Critical sections in FreeRTOS are implemented using the following two
functions:
void vPortEnterCritical( void )
{
portDISABLE_INTERRUPTS();
uxCriticalNesting++;
}
void vPortExitCritical( void )
{
uxCriticalNesting--;
if( uxCriticalNesting == 0 )
{
portENABLE_INTERRUPTS();
}
}
uxCriticalNesting is initialized to a large value at the start and set
to zero when the scheduler is started (xPortStartScheduler). As a
result, before the scheduler is started, a pair of enter/exit critical
section will leave the interrupts disabled because uxCriticalNesting
will not reach zero in the vPortExitCritical function. This is done to
ensure that the interrupts remain disabled from the time first FreeRTOS
API is called to the time when the scheduler is started. The scheduler
starting code is expected to enure that interrupts are enabled before
the first task starts executing.
Cortex-M33 ports were not enabling interrupts before starting the first
task and as a result, the first task was started with interrupts
disabled. This PR fixes the issue by ensuring that interrupts are
enabled before the first task is started.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
* Xtensa: fix the coproc_area incorrect issue
foss-xtensa/amazon-freertos#2 mentioned a issue:
1.
In function pxPortInitialiseStack(StackType_t *pxTopOfStack....)
p = (uint32_t *)(((uint32_t) pxTopOfStack - XT_CP_SIZE) & ~0xf);
In function prvInitialiseNewTask (file: task.c)
pxTopOfStack = (pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)
So the co-processor area is at
p = (uint32_t *)(((uint32_t)((pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)) - XT_CP_SIZE) & ~0xf);
2.
In function vPortStoreTaskMPUSettings( .... , StackType_t pxBottomOfStack ...)
xMPUSettings->coproc_area = (StackType_t)((((uint32_t)(pxBottomOfStack + usStackDepth - 1)) - XT_CP_SIZE) & ~0xf);
pxBottomOfStack = pxStack
=> xMPUSettings->coproc_area = (StackType_t*)((((uint32_t)(pxStack+ ulStackDepth - 1)) - XT_CP_SIZE ) & ~0xf);
The p is coproc_area that should be equal to xMPUSettings->coproc_area.
For example, assume pxStack is 0xa0000000, ulStackDepth is 0x2000,
portBYTE_ALIGNMENT_MASK is 0x7f, XT_CP_SIZE is 0x100.
The p = (uint32_t)(((uint32_t)((pxStack + (ulStackDepth - 1)) & (~portBYTE_ALIGNMENT_MASK)) - XT_CP_SIZE) & ~0xf)
= 0xa0001e80
The xMPUSettings->coproc_area = (StackType_t)((((uint32_t)(pxStack+ usStackDepth - 1)) - XT_CP_SIZE ) & ~0xf)
= 0xa0001ef0
Obviously, the p is not equal to the xMPUSettings->coproc_area, which will cause context switching error.
Signed-off-by: magicse7en <magicse7en@outlook.com>
* Update port.c
Co-authored-by: Carl Lundin <53273776+lundinc2@users.noreply.github.com>
In function pxPortInitialiseStack of port.c:
sp = ( StackType_t * ) ( ( ( UBaseType_t ) ( pxTopOfStack + 1 ) - XT_CP_SIZE - XT_STK_FRMSZ ) & ~0xf );
We assume XT_CP_SIZE is 0xE4, XT_STK_FRMSZ is 0xA0, pxTopOfStack is 0xA0000000, sp is 0x9FFFFE80.
From port.c, we know the frame->a1 as below:
frame->a1 = ( UBaseType_t ) sp + XT_STK_FRMSZ; /* physical top of stack frame */
So frame->a1 is 0x9FFFFF20. Therefore the interrupt stack frame range is 0x9FFFFE80 ~ 0x9FFFFF20.
The coproc_area is: p = ( uint32_t * ) ( ( ( uint32_t ) pxTopOfStack - XT_CP_SIZE ) & ~0xf );
So its value is 0x9FFFFF10. Obviously, the interrupt stack frame overlaps the coproc_area.
Co-authored-by: Carl Lundin <53273776+lundinc2@users.noreply.github.com>
If a higher priority task than the currently running task was resumed
using xTaskResumeFromISR and the user chose to ignore the return value
of xTaskResumeFromISR to initiate a context switch using
portYIELD_FROM_ISR, we were not doing the context switch on the next run
of the scheduler. This change fixes this by marking a yield as pending
to ensure that the context switch is performed on the next run of the
scheduler.
Signed-off-by: Gaurav Aggarwal <aggarg@amazon.com>
This PR:
Changes the INCLUDE_vTaskDelayUntil compile time constant to INCLUDE_xTaskDelayUntil.
Updates FreeRTOS.h to ensure backward compatibility for projects that already have INCLUDE_vTaskDelayUntil defined.
Updates the MPU prototypes, wrapper and implementation to use the updated xTaskDelayUntil() function.
Tests to be checked into the FreeRTOS/FreeRTOS repository after this PR.
* Renamed old port to ESP_IDF_V3
* Update ESP32 port files to support IDF v4.2.
* Add changes required to support ESP32-S2
Co-authored-by: Shubham Kulkarni <shubham.kulkarni@espressif.com>
