sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
3477c347e8
nuttx/arch/arm/src/armv7-r/arm_syscall.c
ligd e2df52390a SMP: fix crash when switch to new task which is still running 
Situation:

Assume we have 2 cpus, and busy run task0.

CPU0                                CPU1
task0 -> task1                      task2 -> task0
1. remove task0 form runninglist
2. take task1 as new tcb
3. add task0 to blocklist
4. clear spinlock
                                    4.1 remove task2 form runninglist
                                    4.2 take task0 as new tcb
                                    4.3 add task2 to blocklist
                                    4.4 use svc ISR swith to task0
                                    4.5 crash
5. use svc ISR swith to task1

Fix:
Move clear spinlock to the end of svc ISR

Signed-off-by: ligd <liguiding1@xiaomi.com>
2022-09-17 17:37:47 +09:00

561 lines
18 KiB
C
Raw Blame History

/****************************************************************************
* arch/arm/src/armv7-r/arm_syscall.c
*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <nuttx/config.h>
#include <stdint.h>
#include <string.h>
#include <assert.h>
#include <debug.h>
#include <syscall.h>
#include <nuttx/arch.h>
#include <nuttx/sched.h>
#include "addrenv.h"
#include "arm.h"
#include "arm_internal.h"
#include "signal/signal.h"
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Name: dump_syscall
*
* Description:
* Dump the syscall registers
*
****************************************************************************/
static void dump_syscall(const char *tag, uint32_t cmd, const uint32_t *regs)
{
/* The SVCall software interrupt is called with R0 = system call command
* and R1..R7 = variable number of arguments depending on the system call.
*/
#ifdef CONFIG_LIB_SYSCALL
if (cmd >= CONFIG_SYS_RESERVED)
{
svcinfo("SYSCALL %s: regs: %p cmd: %" PRId32 " name: %s\n", tag,
regs, cmd, g_funcnames[cmd - CONFIG_SYS_RESERVED]);
}
else
#endif
{
svcinfo("SYSCALL %s: regs: %p cmd: %" PRId32 "\n", tag, regs, cmd);
}
svcinfo(" R0: %08" PRIx32 " %08" PRIx32 " %08" PRIx32 " %08" PRIx32
" %08" PRIx32 " %08" PRIx32 " %08" PRIx32 " %08" PRIx32 "\n",
regs[REG_R0], regs[REG_R1], regs[REG_R2], regs[REG_R3],
regs[REG_R4], regs[REG_R5], regs[REG_R6], regs[REG_R7]);
svcinfo(" R8: %08" PRIx32 " %08" PRIx32 " %08" PRIx32 " %08" PRIx32
" %08" PRIx32 " %08" PRIx32 " %08" PRIx32 " %08" PRIx32 "\n",
regs[REG_R8], regs[REG_R9], regs[REG_R10], regs[REG_R11],
regs[REG_R12], regs[REG_R13], regs[REG_R14], regs[REG_R15]);
svcinfo("CPSR: %08" PRIx32 "\n", regs[REG_CPSR]);
}
/****************************************************************************
* Name: dispatch_syscall
*
* Description:
* Call the stub function corresponding to the system call. NOTE the non-
* standard parameter passing:
*
* R0 = SYS_ call number
* R1 = parm0
* R2 = parm1
* R3 = parm2
* R4 = parm3
* R5 = parm4
* R6 = parm5
*
* The values of R4-R5 may be preserved in the proxy called by the user
* code if they are used (but otherwise will not be).
*
* WARNING: There are hard-coded values in this logic!
*
* Register usage:
*
* R0 - Need not be preserved.
* R1-R3 - Need to be preserved until the stub is called. The values of
* R0 and R1 returned by the stub must be preserved.
* R4-R11 must be preserved to support the expectations of the user-space
* callee. R4-R6 may have been preserved by the proxy, but don't know
* for sure.
* R12 - Need not be preserved
* R13 - (stack pointer)
* R14 - Need not be preserved
* R15 - (PC)
*
****************************************************************************/
#ifdef CONFIG_LIB_SYSCALL
static void dispatch_syscall(void) naked_function;
static void dispatch_syscall(void)
{
__asm__ __volatile__
(
" sub sp, sp, #16\n" /* Create a stack frame to hold 3 parms + lr */
" str r4, [sp, #0]\n" /* Move parameter 4 (if any) into position */
" str r5, [sp, #4]\n" /* Move parameter 5 (if any) into position */
" str r6, [sp, #8]\n" /* Move parameter 6 (if any) into position */
" str lr, [sp, #12]\n" /* Save lr in the stack frame */
" ldr ip, =g_stublookup\n" /* R12=The base of the stub lookup table */
" ldr ip, [ip, r0, lsl #2]\n" /* R12=The address of the stub for this SYSCALL */
" blx ip\n" /* Call the stub (modifies lr) */
" ldr lr, [sp, #12]\n" /* Restore lr */
" add sp, sp, #16\n" /* Destroy the stack frame */
" mov r2, r0\n" /* R2=Save return value in R2 */
" mov r0, %0\n" /* R0=SYS_syscall_return */
" svc %1\n"::"i"(SYS_syscall_return),
"i"(SYS_syscall) /* Return from the SYSCALL */
);
}
#endif
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: arm_syscall
*
* Description:
* SVC interrupts will vector here with insn=the SVC instruction and
* xcp=the interrupt context
*
* The handler may get the SVC number be de-referencing the return
* address saved in the xcp and decoding the SVC instruction
*
****************************************************************************/
uint32_t *arm_syscall(uint32_t *regs)
{
uint32_t cmd;
#ifdef CONFIG_BUILD_PROTECTED
uint32_t cpsr;
#endif
/* Nested interrupts are not supported */
DEBUGASSERT(CURRENT_REGS == NULL);
/* Current regs non-zero indicates that we are processing an interrupt;
* CURRENT_REGS is also used to manage interrupt level context switches.
*/
CURRENT_REGS = regs;
/* The SYSCALL command is in R0 on entry. Parameters follow in R1..R7 */
cmd = regs[REG_R0];
/* The SVCall software interrupt is called with R0 = system call command
* and R1..R7 = variable number of arguments depending on the system call.
*/
dump_syscall("Entry", cmd, regs);
/* Handle the SVCall according to the command in R0 */
switch (cmd)
{
/* R0=SYS_syscall_return: This a SYSCALL return command:
*
* void arm_syscall_return(void);
*
* At this point, the following values are saved in context:
*
* R0 = SYS_syscall_return
*
* We need to restore the saved return address and return in
* unprivileged thread mode.
*/
#ifdef CONFIG_LIB_SYSCALL
case SYS_syscall_return:
{
struct tcb_s *rtcb = nxsched_self();
int index = (int)rtcb->xcp.nsyscalls - 1;
/* Make sure that there is a saved SYSCALL return address. */
DEBUGASSERT(index >= 0);
/* Setup to return to the saved SYSCALL return address in
* the original mode.
*/
regs[REG_PC] = rtcb->xcp.syscall[index].sysreturn;
#ifdef CONFIG_BUILD_PROTECTED
regs[REG_CPSR] = rtcb->xcp.syscall[index].cpsr;
#endif
/* The return value must be in R0-R1. dispatch_syscall()
* temporarily moved the value for R0 into R2.
*/
regs[REG_R0] = regs[REG_R2];
#ifdef CONFIG_ARCH_KERNEL_STACK
/* If this is the outermost SYSCALL and if there is a saved user
* stack pointer, then restore the user stack pointer on this
* final return to user code.
*/
if (index == 0 && rtcb->xcp.ustkptr != NULL)
{
regs[REG_SP] = (uint32_t)rtcb->xcp.ustkptr;
rtcb->xcp.ustkptr = NULL;
}
#endif
/* Save the new SYSCALL nesting level */
rtcb->xcp.nsyscalls = index;
/* Handle any signal actions that were deferred while processing
* the system call.
*/
rtcb->flags &= ~TCB_FLAG_SYSCALL;
nxsig_unmask_pendingsignal();
}
break;
#endif
/* R0=SYS_restore_context: Restore task context
*
* void arm_fullcontextrestore(uint32_t *restoreregs)
* noreturn_function;
*
* At this point, the following values are saved in context:
*
* R0 = SYS_restore_context
* R1 = restoreregs
*/
case SYS_restore_context:
{
/* Replace 'regs' with the pointer to the register set in
* regs[REG_R1]. On return from the system call, that register
* set will determine the restored context.
*/
CURRENT_REGS = (uint32_t *)regs[REG_R1];
DEBUGASSERT(CURRENT_REGS);
}
break;
/* R0=SYS_switch_context: This a switch context command:
*
* void arm_switchcontext(uint32_t **saveregs,
* uint32_t *restoreregs);
*
* At this point, the following values are saved in context:
*
* R0 = SYS_switch_context
* R1 = saveregs
* R2 = restoreregs
*
* In this case, we do both: We save the context registers to the save
* register area reference by the saved contents of R1 and then set
* regs to the save register area referenced by the saved
* contents of R2.
*/
case SYS_switch_context:
{
DEBUGASSERT(regs[REG_R1] != 0 && regs[REG_R2] != 0);
*(uint32_t **)regs[REG_R1] = regs;
CURRENT_REGS = (uint32_t *)regs[REG_R2];
}
break;
/* R0=SYS_task_start: This a user task start
*
* void up_task_start(main_t taskentry, int argc, char *argv[])
* noreturn_function;
*
* At this point, the following values are saved in context:
*
* R0 = SYS_task_start
* R1 = taskentry
* R2 = argc
* R3 = argv
*/
#ifdef CONFIG_BUILD_PROTECTED
case SYS_task_start:
{
/* Set up to return to the user-space _start function in
* unprivileged mode. We need:
*
* R0 = argc
* R1 = argv
* PC = taskentry
* CSPR = user mode
*/
regs[REG_PC] = regs[REG_R1];
regs[REG_R0] = regs[REG_R2];
regs[REG_R1] = regs[REG_R3];
cpsr = regs[REG_CPSR] & ~PSR_MODE_MASK;
regs[REG_CPSR] = cpsr | PSR_MODE_USR;
}
break;
#endif
/* R0=SYS_pthread_start: This a user pthread start
*
* void up_pthread_start(pthread_startroutine_t entrypt,
* pthread_addr_t arg) noreturn_function;
*
* At this point, the following values are saved in context:
*
* R0 = SYS_pthread_start
* R1 = entrypt
* R2 = arg
*/
#if !defined(CONFIG_BUILD_FLAT) && !defined(CONFIG_DISABLE_PTHREAD)
case SYS_pthread_start:
{
/* Set up to enter the user-space pthread start-up function in
* unprivileged mode. We need:
*
* R0 = startup
* R1 = arg
* PC = entrypt
* CSPR = user mode
*/
regs[REG_PC] = regs[REG_R1];
regs[REG_R0] = regs[REG_R2];
regs[REG_R1] = regs[REG_R3];
cpsr = regs[REG_CPSR] & ~PSR_MODE_MASK;
regs[REG_CPSR] = cpsr | PSR_MODE_USR;
}
break;
#endif
#ifdef CONFIG_BUILD_PROTECTED
/* R0=SYS_signal_handler: This a user signal handler callback
*
* void signal_handler(_sa_sigaction_t sighand, int signo,
* siginfo_t *info, void *ucontext);
*
* At this point, the following values are saved in context:
*
* R0 = SYS_signal_handler
* R1 = sighand
* R2 = signo
* R3 = info
* R4 = ucontext
*/
case SYS_signal_handler:
{
struct tcb_s *rtcb = nxsched_self();
/* Remember the caller's return address */
DEBUGASSERT(rtcb->xcp.sigreturn == 0);
rtcb->xcp.sigreturn = regs[REG_PC];
/* Set up to return to the user-space trampoline function in
* unprivileged mode.
*/
regs[REG_PC] = (uint32_t)ARCH_DATA_RESERVE->ar_sigtramp;
cpsr = regs[REG_CPSR] & ~PSR_MODE_MASK;
regs[REG_CPSR] = cpsr | PSR_MODE_USR;
/* Change the parameter ordering to match the expectation of struct
* userpace_s signal_handler.
*/
regs[REG_R0] = regs[REG_R1]; /* sighand */
regs[REG_R1] = regs[REG_R2]; /* signal */
regs[REG_R2] = regs[REG_R3]; /* info */
regs[REG_R3] = regs[REG_R4]; /* ucontext */
#ifdef CONFIG_ARCH_KERNEL_STACK
/* If we are signalling a user process, then we must be operating
* on the kernel stack now. We need to switch back to the user
* stack before dispatching the signal handler to the user code.
* The existence of an allocated kernel stack is sufficient
* information to make this decision.
*/
if (rtcb->xcp.kstack != NULL)
{
DEBUGASSERT(rtcb->xcp.kstkptr == NULL &&
rtcb->xcp.ustkptr != NULL);
rtcb->xcp.kstkptr = (uint32_t *)regs[REG_SP];
regs[REG_SP] = (uint32_t)rtcb->xcp.ustkptr;
}
#endif
}
break;
#endif
#ifdef CONFIG_BUILD_PROTECTED
/* R0=SYS_signal_handler_return: This a user signal handler callback
*
* void signal_handler_return(void);
*
* At this point, the following values are saved in context:
*
* R0 = SYS_signal_handler_return
*/
case SYS_signal_handler_return:
{
struct tcb_s *rtcb = nxsched_self();
/* Set up to return to the kernel-mode signal dispatching logic. */
DEBUGASSERT(rtcb->xcp.sigreturn != 0);
regs[REG_PC] = rtcb->xcp.sigreturn;
cpsr = regs[REG_CPSR] & ~PSR_MODE_MASK;
regs[REG_CPSR] = cpsr | PSR_MODE_SYS;
rtcb->xcp.sigreturn = 0;
#ifdef CONFIG_ARCH_KERNEL_STACK
/* We must enter here be using the user stack. We need to switch
* to back to the kernel user stack before returning to the kernel
* mode signal trampoline.
*/
if (rtcb->xcp.kstack != NULL)
{
DEBUGASSERT(rtcb->xcp.kstkptr != NULL &&
(uint32_t)rtcb->xcp.ustkptr == regs[REG_SP]);
regs[REG_SP] = (uint32_t)rtcb->xcp.kstkptr;
rtcb->xcp.kstkptr = NULL;
}
#endif
}
break;
#endif
/* This is not an architecture-specific system call. If NuttX is built
* as a standalone kernel with a system call interface, then all of the
* additional system calls must be handled as in the default case.
*/
default:
{
#ifdef CONFIG_LIB_SYSCALL
struct tcb_s *rtcb = nxsched_self();
int index = rtcb->xcp.nsyscalls;
/* Verify that the SYS call number is within range */
DEBUGASSERT(cmd >= CONFIG_SYS_RESERVED && cmd < SYS_maxsyscall);
/* Make sure that there is a no saved SYSCALL return address. We
* cannot yet handle nested system calls.
*/
DEBUGASSERT(index < CONFIG_SYS_NNEST);
/* Setup to return to dispatch_syscall in privileged mode. */
rtcb->xcp.syscall[index].sysreturn = regs[REG_PC];
#ifdef CONFIG_BUILD_PROTECTED
rtcb->xcp.syscall[index].cpsr = regs[REG_CPSR];
#endif
regs[REG_PC] = (uint32_t)dispatch_syscall;
#ifdef CONFIG_BUILD_PROTECTED
cpsr = regs[REG_CPSR] & ~PSR_MODE_MASK;
regs[REG_CPSR] = cpsr | PSR_MODE_SYS;
#endif
/* Offset R0 to account for the reserved values */
regs[REG_R0] -= CONFIG_SYS_RESERVED;
/* Indicate that we are in a syscall handler. */
rtcb->flags |= TCB_FLAG_SYSCALL;
#ifdef CONFIG_ARCH_KERNEL_STACK
/* If this is the first SYSCALL and if there is an allocated
* kernel stack, then switch to the kernel stack.
*/
if (index == 0 && rtcb->xcp.kstack != NULL)
{
rtcb->xcp.ustkptr = (uint32_t *)regs[REG_SP];
regs[REG_SP] = (uint32_t)rtcb->xcp.kstack +
ARCH_KERNEL_STACKSIZE;
}
#endif
/* Save the new SYSCALL nesting level */
rtcb->xcp.nsyscalls = index + 1;
#else
svcerr("ERROR: Bad SYS call: 0x%" PRIx32 "\n", regs[REG_R0]);
#endif
}
break;
}
/* Restore the cpu lock */
if (regs != CURRENT_REGS)
{
restore_critical_section();
regs = (uint32_t *)CURRENT_REGS;
}
/* Report what happened */
dump_syscall("Exit", cmd, regs);
/* Set CURRENT_REGS to NULL to indicate that we are no longer in an
* interrupt handler.
*/
CURRENT_REGS = NULL;
/* Return the last value of curent_regs. This supports context switches
* on return from the exception. That capability is only used with the
* SYS_context_switch system call.
*/
return regs;
}
Reference in New Issue View Git Blame Copy Permalink