haiku/src/system/kernel/syscalls.cpp

626 lines
14 KiB
C++
Raw Normal View History

/*
* Copyright 2008, Ingo Weinhold, ingo_weinhold@gmx.de.
* Copyright 2004-2006, Haiku Inc. All rights reserved.
* Distributed under the terms of the MIT License.
*/
/* Big case statement for dispatching syscalls */
#include <kernel.h>
#include <ksyscalls.h>
#include <syscalls.h>
#include <generic_syscall.h>
#include <debug.h>
#include <int.h>
#include <elf.h>
#include <vfs.h>
#include <vm.h>
#include <thread.h>
#include <posix/realtime_sem.h>
#include <sem.h>
#include <port.h>
#include <cpu.h>
#include <arch_config.h>
#include <disk_device_manager/ddm_userland_interface.h>
#include <sys/resource.h>
#include <fs/fd.h>
#include <fs/node_monitor.h>
#include <kimage.h>
#include <ksignal.h>
#include <real_time_clock.h>
#include <safemode.h>
#include <system_info.h>
#include <tracing.h>
#include <user_atomic.h>
#include <arch/system_info.h>
#include <messaging.h>
#include <frame_buffer_console.h>
#include <usergroup.h>
#include <wait_for_objects.h>
#include <malloc.h>
#include <string.h>
#include "syscall_numbers.h"
typedef struct generic_syscall generic_syscall;
struct generic_syscall {
list_link link;
char subsystem[B_FILE_NAME_LENGTH];
syscall_hook hook;
uint32 version;
uint32 flags;
generic_syscall *previous;
};
static struct mutex sGenericSyscallLock;
static struct list sGenericSyscalls;
#if SYSCALL_TRACING
static int dump_syscall_tracing(int argc, char** argv);
#endif
static generic_syscall *
find_generic_syscall(const char *subsystem)
{
generic_syscall *syscall = NULL;
ASSERT_LOCKED_MUTEX(&sGenericSyscallLock);
while ((syscall = (generic_syscall*)list_get_next_item(&sGenericSyscalls,
syscall)) != NULL) {
if (!strcmp(syscall->subsystem, subsystem))
return syscall;
}
return NULL;
}
/** Calls the generic syscall subsystem if any.
* Also handles the special generic syscall function \c B_SYSCALL_INFO.
* Returns \c B_NAME_NOT_FOUND if either the subsystem was not found, or
* the subsystem does not support the requested function.
* All other return codes are depending on the generic syscall implementation.
*/
static inline status_t
_user_generic_syscall(const char *userSubsystem, uint32 function,
void *buffer, size_t bufferSize)
{
char subsystem[B_FILE_NAME_LENGTH];
generic_syscall *syscall;
status_t status = B_NAME_NOT_FOUND;
if (!IS_USER_ADDRESS(userSubsystem)
|| user_strlcpy(subsystem, userSubsystem, sizeof(subsystem)) < B_OK)
return B_BAD_ADDRESS;
//dprintf("generic_syscall(subsystem = \"%s\", function = %lu)\n", subsystem, function);
mutex_lock(&sGenericSyscallLock);
syscall = find_generic_syscall(subsystem);
if (syscall == NULL)
goto out;
if (function >= B_RESERVED_SYSCALL_BASE) {
if (function != B_SYSCALL_INFO) {
// this is all we know
status = B_NAME_NOT_FOUND;
goto out;
}
// special info syscall
if (bufferSize != sizeof(uint32))
status = B_BAD_VALUE;
else {
uint32 requestedVersion;
// retrieve old version
status = user_memcpy(&requestedVersion, buffer, sizeof(uint32));
if (status == B_OK && requestedVersion != 0 && requestedVersion < syscall->version)
status = B_BAD_TYPE;
// return current version
if (status == B_OK)
status = user_memcpy(buffer, &syscall->version, sizeof(uint32));
}
} else {
while (syscall != NULL) {
generic_syscall *next;
mutex_unlock(&sGenericSyscallLock);
status = syscall->hook(subsystem, function, buffer, bufferSize);
mutex_lock(&sGenericSyscallLock);
if (status != B_BAD_HANDLER)
break;
// the syscall may have been removed in the mean time
next = find_generic_syscall(subsystem);
if (next == syscall)
syscall = syscall->previous;
else
syscall = next;
}
if (syscall == NULL)
status = B_NAME_NOT_FOUND;
}
out:
mutex_unlock(&sGenericSyscallLock);
return status;
}
static inline int
_user_is_computer_on(void)
{
return 1;
}
// map to the arch specific call
static inline int64
_user_restore_signal_frame()
{
Merged branch haiku/branches/developer/bonefish/optimization revision 23139 into trunk, with roughly the following changes (for details svn log the branch): * The int 99 syscall handler is now fully in assembly. * Added a sysenter/sysexit handler and use it on Pentiums that support it (via commpage). * Got rid of i386_handle_trap(). A bit of functionality was moved into the assembly handler which now uses a jump table to call C functions handling the respective interrupt. * Some optimizations to get user debugger support code out of the interrupt handling path. * Introduced a thread::flags fields which allows to skip handling of rare events (signals, user debug enabling/disabling) on the common interrupt handling path. * Got rid of the explicit iframe stack. The iframes can still be retrieved by iterating through the stack frames. * Made the commpage an architecture independent feature. It's used for the real time data stuff (instead of creating a separate area). * The x86 CPU modules can now provide processor optimized versions for common functions (currently memcpy() only). They are used in the kernel and are provided to the userland via commpage entries. * Introduced build system feature allowing easy use of C structure member offsets in assembly code. Changes after merging: * Fixed merge conflict in src/system/kernel/arch/x86/arch_debug.cpp (caused by refactoring and introduction of "call" debugger command). git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@23370 a95241bf-73f2-0310-859d-f6bbb57e9c96
2008-01-11 03:36:44 +03:00
syscall_64_bit_return_value();
return arch_restore_signal_frame();
}
// #pragma mark -
int32
syscall_dispatcher(uint32 call_num, void *args, uint64 *call_ret)
{
bigtime_t startTime;
// dprintf("syscall_dispatcher: thread 0x%x call 0x%x, arg0 0x%x, arg1 0x%x arg2 0x%x arg3 0x%x arg4 0x%x\n",
// thread_get_current_thread_id(), call_num, arg0, arg1, arg2, arg3, arg4);
user_debug_pre_syscall(call_num, args);
startTime = system_time();
switch (call_num) {
// the cases are auto-generated
#include "syscall_dispatcher.h"
default:
*call_ret = (uint64)B_BAD_VALUE;
}
user_debug_post_syscall(call_num, args, *call_ret, startTime);
// dprintf("syscall_dispatcher: done with syscall 0x%x\n", call_num);
return B_HANDLED_INTERRUPT;
}
status_t
generic_syscall_init(void)
{
list_init(&sGenericSyscalls);
mutex_init(&sGenericSyscallLock, "generic syscall");
#if SYSCALL_TRACING
add_debugger_command_etc("straced", &dump_syscall_tracing,
"Dump recorded syscall trace entries",
"Prints recorded trace entries. It is wrapper for the \"traced\"\n"
"command and supports all of its command line options (though\n"
"backward tracing doesn't really work). The difference is that if a\n"
"pre syscall trace entry is encountered, the corresponding post\n"
"syscall traced entry is also printed, even if it doesn't match the\n"
"given filter.\n", 0);
#endif // ENABLE_TRACING
return B_OK;
}
// #pragma mark -
// public API
status_t
register_generic_syscall(const char *subsystem, syscall_hook hook,
uint32 version, uint32 flags)
{
struct generic_syscall *previous, *syscall;
status_t status;
if (hook == NULL)
return B_BAD_VALUE;
mutex_lock(&sGenericSyscallLock);
previous = find_generic_syscall(subsystem);
if (previous != NULL) {
if ((flags & B_DO_NOT_REPLACE_SYSCALL) != 0
|| version < previous->version) {
status = B_NAME_IN_USE;
goto out;
}
if (previous->flags & B_SYSCALL_NOT_REPLACEABLE) {
status = B_NOT_ALLOWED;
goto out;
}
}
syscall = (generic_syscall *)malloc(sizeof(struct generic_syscall));
if (syscall == NULL) {
status = B_NO_MEMORY;
goto out;
}
strlcpy(syscall->subsystem, subsystem, sizeof(syscall->subsystem));
syscall->hook = hook;
syscall->version = version;
syscall->flags = flags;
syscall->previous = previous;
list_add_item(&sGenericSyscalls, syscall);
if (previous != NULL)
list_remove_link(&previous->link);
status = B_OK;
out:
mutex_unlock(&sGenericSyscallLock);
return status;
}
status_t
unregister_generic_syscall(const char *subsystem, uint32 version)
{
// ToDo: we should only remove the syscall with the matching version
generic_syscall *syscall;
status_t status;
mutex_lock(&sGenericSyscallLock);
syscall = find_generic_syscall(subsystem);
if (syscall != NULL) {
if (syscall->previous != NULL) {
// reestablish the old syscall
list_add_item(&sGenericSyscalls, syscall->previous);
}
list_remove_link(&syscall->link);
free(syscall);
status = B_OK;
} else
status = B_NAME_NOT_FOUND;
mutex_unlock(&sGenericSyscallLock);
return status;
}
// #pragma mark - syscall tracing
#if SYSCALL_TRACING
namespace SyscallTracing {
static const char*
get_syscall_name(uint32 syscall)
{
if (syscall >= (uint32)kSyscallCount)
return "<invalid syscall number>";
return kExtendedSyscallInfos[syscall].name;
}
class PreSyscall : public AbstractTraceEntry {
public:
PreSyscall(uint32 syscall, const void* parameters)
:
fSyscall(syscall),
fParameters(NULL)
{
if (syscall < (uint32)kSyscallCount) {
fParameters = alloc_tracing_buffer_memcpy(parameters,
kSyscallInfos[syscall].parameter_size, false);
// copy string parameters, if any
if (fParameters != NULL && syscall != SYSCALL_KTRACE_OUTPUT) {
int32 stringIndex = 0;
const extended_syscall_info& syscallInfo
= kExtendedSyscallInfos[fSyscall];
for (int i = 0; i < syscallInfo.parameter_count; i++) {
const syscall_parameter_info& paramInfo
= syscallInfo.parameters[i];
if (paramInfo.type != B_STRING_TYPE)
continue;
const uint8* data
= (uint8*)fParameters + paramInfo.offset;
if (stringIndex < MAX_PARAM_STRINGS) {
fParameterStrings[stringIndex++]
= alloc_tracing_buffer_strcpy(
*(const char**)data, 64, true);
}
}
}
}
Initialized();
}
virtual void AddDump(TraceOutput& out)
{
out.Print("syscall pre: %s(", get_syscall_name(fSyscall));
if (fParameters != NULL) {
int32 stringIndex = 0;
const extended_syscall_info& syscallInfo
= kExtendedSyscallInfos[fSyscall];
for (int i = 0; i < syscallInfo.parameter_count; i++) {
const syscall_parameter_info& paramInfo
= syscallInfo.parameters[i];
const uint8* data = (uint8*)fParameters + paramInfo.offset;
uint64 value = 0;
bool printValue = true;
switch (paramInfo.type) {
case B_INT8_TYPE:
value = *(uint8*)data;
break;
case B_INT16_TYPE:
value = *(uint16*)data;
break;
case B_INT32_TYPE:
value = *(uint32*)data;
break;
case B_INT64_TYPE:
value = *(uint64*)data;
break;
case B_POINTER_TYPE:
value = (uint64)*(void**)data;
break;
case B_STRING_TYPE:
if (stringIndex < MAX_PARAM_STRINGS
&& fSyscall != SYSCALL_KTRACE_OUTPUT) {
out.Print("%s\"%s\"",
(i == 0 ? "" : ", "),
fParameterStrings[stringIndex++]);
printValue = false;
} else
value = (uint64)*(void**)data;
break;
}
if (printValue)
out.Print("%s0x%llx", (i == 0 ? "" : ", "), value);
}
}
out.Print(")");
}
private:
enum { MAX_PARAM_STRINGS = 3 };
uint32 fSyscall;
void* fParameters;
const char* fParameterStrings[MAX_PARAM_STRINGS];
};
class PostSyscall : public AbstractTraceEntry {
public:
PostSyscall(uint32 syscall, uint64 returnValue)
:
fSyscall(syscall),
fReturnValue(returnValue)
{
Initialized();
#if 0
if (syscall < (uint32)kSyscallCount
&& returnValue != (returnValue & 0xffffffff)
&& kExtendedSyscallInfos[syscall].return_type.size <= 4) {
panic("syscall return value 64 bit although it should be 32 "
"bit");
}
#endif
}
virtual void AddDump(TraceOutput& out)
{
out.Print("syscall post: %s() -> 0x%llx",
get_syscall_name(fSyscall), fReturnValue);
}
private:
uint32 fSyscall;
uint64 fReturnValue;
};
} // namespace SyscallTracing
extern "C" void trace_pre_syscall(uint32 syscallNumber, const void* parameters);
void
trace_pre_syscall(uint32 syscallNumber, const void* parameters)
{
#if SYSCALL_TRACING_IGNORE_KTRACE_OUTPUT
if (syscallNumber != SYSCALL_KTRACE_OUTPUT)
#endif
{
new(std::nothrow) SyscallTracing::PreSyscall(syscallNumber, parameters);
}
}
extern "C" void trace_post_syscall(int syscallNumber, uint64 returnValue);
void
trace_post_syscall(int syscallNumber, uint64 returnValue)
{
#if SYSCALL_TRACING_IGNORE_KTRACE_OUTPUT
if (syscallNumber != SYSCALL_KTRACE_OUTPUT)
#endif
{
new(std::nothrow) SyscallTracing::PostSyscall(syscallNumber,
returnValue);
}
}
using namespace SyscallTracing;
class SyscallWrapperTraceFilter : public WrapperTraceFilter {
public:
virtual void Init(TraceFilter* filter, int direction, bool continued)
{
fFilter = filter;
fHitThreadLimit = false;
fDirection = direction;
if (!continued)
fPendingThreadCount = 0;
}
virtual bool Filter(const TraceEntry* _entry, LazyTraceOutput& out)
{
if (fFilter == NULL)
return true;
if (fDirection < 0)
return fFilter->Filter(_entry, out);
if (const PreSyscall* entry = dynamic_cast<const PreSyscall*>(_entry)) {
_RemovePendingThread(entry->Thread());
bool accepted = fFilter->Filter(entry, out);
if (accepted)
_AddPendingThread(entry->Thread());
return accepted;
} else if (const PostSyscall* entry
= dynamic_cast<const PostSyscall*>(_entry)) {
bool wasPending = _RemovePendingThread(entry->Thread());
return wasPending || fFilter->Filter(entry, out);
} else if (const AbstractTraceEntry* entry
= dynamic_cast<const AbstractTraceEntry*>(_entry)) {
bool isPending = _IsPendingThread(entry->Thread());
return isPending || fFilter->Filter(entry, out);
} else {
return fFilter->Filter(_entry, out);
}
}
bool HitThreadLimit() const
{
return fHitThreadLimit;
}
int Direction() const
{
return fDirection;
}
private:
enum {
MAX_PENDING_THREADS = 32
};
bool _AddPendingThread(thread_id thread)
{
int32 index = _PendingThreadIndex(thread);
if (index >= 0)
return true;
if (fPendingThreadCount == MAX_PENDING_THREADS) {
fHitThreadLimit = true;
return false;
}
fPendingThreads[fPendingThreadCount++] = thread;
return true;
}
bool _RemovePendingThread(thread_id thread)
{
int32 index = _PendingThreadIndex(thread);
if (index < 0)
return false;
if (index + 1 < fPendingThreadCount) {
memmove(fPendingThreads + index, fPendingThreads + index + 1,
fPendingThreadCount - index - 1);
}
fPendingThreadCount--;
return true;
}
bool _IsPendingThread(thread_id thread)
{
return _PendingThreadIndex(thread) >= 0;
}
int32 _PendingThreadIndex(thread_id thread)
{
for (int32 i = 0; i < fPendingThreadCount; i++) {
if (fPendingThreads[i] == thread)
return i;
}
return -1;
}
TraceFilter* fFilter;
thread_id fPendingThreads[MAX_PENDING_THREADS];
int32 fPendingThreadCount;
int fDirection;
bool fHitThreadLimit;
};
static SyscallWrapperTraceFilter sFilter;
static int
dump_syscall_tracing(int argc, char** argv)
{
new(&sFilter) SyscallWrapperTraceFilter;
int result = dump_tracing(argc, argv, &sFilter);
if (sFilter.HitThreadLimit()) {
kprintf("Warning: The thread buffer was too small to track all "
"threads!\n");
} else if (sFilter.HitThreadLimit()) {
kprintf("Warning: Can't track syscalls backwards!\n");
}
return result;
}
#endif // SYSCALL_TRACING
/*
* kSyscallCount and kSyscallInfos here
*/
// generated by gensyscalls
#include "syscall_table.h"