1
0
Fork 0
mirror of https://gitlab.com/sortix/sortix.git synced 2023-02-13 20:55:38 -05:00
sortix--sortix/kernel/scheduler.cpp

386 lines
10 KiB
C++
Raw Normal View History

Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
/*******************************************************************************
2011-08-05 08:25:00 -04:00
Copyright(C) Jonas 'Sortie' Termansen 2011, 2012, 2013, 2014.
2011-08-05 08:25:00 -04:00
This file is part of Sortix.
2011-08-05 08:25:00 -04:00
Sortix is free software: you can redistribute it and/or modify it under the
terms of the GNU General Public License as published by the Free Software
Foundation, either version 3 of the License, or (at your option) any later
version.
2011-08-05 08:25:00 -04:00
Sortix is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
details.
2011-08-05 08:25:00 -04:00
You should have received a copy of the GNU General Public License along with
Sortix. If not, see <http://www.gnu.org/licenses/>.
2011-08-05 08:25:00 -04:00
scheduler.cpp
Decides the order to execute threads in and switching between them.
2011-08-05 08:25:00 -04:00
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
*******************************************************************************/
2011-08-05 08:25:00 -04:00
#include <sys/types.h>
#include <assert.h>
#include <string.h>
#include <timespec.h>
#if defined(__x86_64__)
#include <msr.h>
#endif
#include <sortix/clock.h>
#include <sortix/timespec.h>
2014-02-21 11:05:10 -05:00
#include <sortix/kernel/decl.h>
2013-01-09 17:30:36 -05:00
#include <sortix/kernel/interrupt.h>
2013-10-26 20:42:10 -04:00
#include <sortix/kernel/kernel.h>
#include <sortix/kernel/memorymanagement.h>
#include <sortix/kernel/process.h>
#include <sortix/kernel/registers.h>
2013-01-09 04:47:22 -05:00
#include <sortix/kernel/scheduler.h>
2013-05-12 18:52:58 -04:00
#include <sortix/kernel/signal.h>
2013-10-26 20:42:10 -04:00
#include <sortix/kernel/syscall.h>
#include <sortix/kernel/thread.h>
2013-10-26 20:42:10 -04:00
#include <sortix/kernel/time.h>
2013-01-08 18:41:35 -05:00
2014-02-21 11:05:10 -05:00
#if defined(__i386__) || defined(__x86_64__)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
#include "x86-family/gdt.h"
#include "x86-family/float.h"
2014-02-21 11:05:10 -05:00
#endif
2011-08-05 08:25:00 -04:00
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
namespace Sortix {
namespace Scheduler {
2014-02-21 11:05:10 -05:00
static Thread* current_thread;
void SaveInterruptedContext(const struct interrupt_context* intctx,
struct thread_registers* registers)
{
#if defined(__i386__)
registers->signal_pending = intctx->signal_pending;
registers->kerrno = intctx->kerrno;
registers->eax = intctx->eax;
registers->ebx = intctx->ebx;
registers->ecx = intctx->ecx;
registers->edx = intctx->edx;
registers->edi = intctx->edi;
registers->esi = intctx->esi;
registers->esp = intctx->esp;
registers->ebp = intctx->ebp;
registers->eip = intctx->eip;
registers->eflags = intctx->eflags;
registers->fsbase = (unsigned long) GDT::GetFSBase();
registers->gsbase = (unsigned long) GDT::GetGSBase();
asm ( "mov %%cr3, %0" : "=r"(registers->cr3) );
registers->kernel_stack = GDT::GetKernelStack();
registers->cs = intctx->cs;
registers->ds = intctx->ds;
registers->ss = intctx->ss;
asm volatile ("fxsave (%0)" : : "r"(registers->fpuenv));
#elif defined(__x86_64__)
registers->signal_pending = intctx->signal_pending;
registers->kerrno = intctx->kerrno;
registers->rax = intctx->rax;
registers->rbx = intctx->rbx;
registers->rcx = intctx->rcx;
registers->rdx = intctx->rdx;
registers->rdi = intctx->rdi;
registers->rsi = intctx->rsi;
registers->rsp = intctx->rsp;
registers->rbp = intctx->rbp;
registers->r8 = intctx->r8;
registers->r9 = intctx->r9;
registers->r10 = intctx->r10;
registers->r11 = intctx->r11;
registers->r12 = intctx->r12;
registers->r13 = intctx->r13;
registers->r14 = intctx->r14;
registers->r15 = intctx->r15;
registers->r15 = intctx->r15;
registers->rip = intctx->rip;
registers->rflags = intctx->rflags;
registers->fsbase = (unsigned long) rdmsr(MSRID_FSBASE);
registers->gsbase = (unsigned long) rdmsr(MSRID_GSBASE);
asm ( "mov %%cr3, %0" : "=r"(registers->cr3) );
registers->kernel_stack = GDT::GetKernelStack();
registers->cs = intctx->cs;
registers->ds = intctx->ds;
registers->ss = intctx->ss;
asm volatile ("fxsave (%0)" : : "r"(registers->fpuenv));
#else
#warning "You need to implement register saving"
#endif
}
void LoadInterruptedContext(struct interrupt_context* intctx,
const struct thread_registers* registers)
{
#if defined(__i386__)
intctx->signal_pending = registers->signal_pending;
intctx->kerrno = registers->kerrno;
intctx->eax = registers->eax;
intctx->ebx = registers->ebx;
intctx->ecx = registers->ecx;
intctx->edx = registers->edx;
intctx->edi = registers->edi;
intctx->esi = registers->esi;
intctx->esp = registers->esp;
intctx->ebp = registers->ebp;
intctx->eip = registers->eip;
intctx->eflags = registers->eflags;
GDT::SetFSBase(registers->fsbase);
GDT::SetGSBase(registers->gsbase);
asm volatile ( "mov %0, %%cr3" : : "r"(registers->cr3) );
GDT::SetKernelStack(registers->kernel_stack);
intctx->cs = registers->cs;
intctx->ds = registers->ds;
intctx->ss = registers->ss;
asm volatile ("fxrstor (%0)" : : "r"(registers->fpuenv));
#elif defined(__x86_64__)
intctx->signal_pending = registers->signal_pending;
intctx->kerrno = registers->kerrno;
intctx->rax = registers->rax;
intctx->rbx = registers->rbx;
intctx->rcx = registers->rcx;
intctx->rdx = registers->rdx;
intctx->rdi = registers->rdi;
intctx->rsi = registers->rsi;
intctx->rsp = registers->rsp;
intctx->rbp = registers->rbp;
intctx->r8 = registers->r8;
intctx->r9 = registers->r9;
intctx->r10 = registers->r10;
intctx->r11 = registers->r11;
intctx->r12 = registers->r12;
intctx->r13 = registers->r13;
intctx->r14 = registers->r14;
intctx->r15 = registers->r15;
intctx->r15 = registers->r15;
intctx->rip = registers->rip;
intctx->rflags = registers->rflags;
wrmsr(MSRID_FSBASE, registers->fsbase);
wrmsr(MSRID_GSBASE, registers->gsbase);
asm volatile ( "mov %0, %%cr3" : : "r"(registers->cr3) );
GDT::SetKernelStack(registers->kernel_stack);
intctx->cs = registers->cs;
intctx->ds = registers->ds;
intctx->ss = registers->ss;
asm volatile ("fxrstor (%0)" : : "r"(registers->fpuenv));
#else
#warning "You need to implement register loading"
#endif
}
static
void SwitchThread(struct interrupt_context* intctx, Thread* prev, Thread* next)
{
if ( prev == next )
return;
SaveInterruptedContext(intctx, &prev->registers);
if ( !prev->registers.cr3 )
Log::PrintF("Thread %p had cr3=0x%zx\n", prev, prev->registers.cr3);
if ( !next->registers.cr3 )
Log::PrintF("Thread %p has cr3=0x%zx\n", next, next->registers.cr3);
LoadInterruptedContext(intctx, &next->registers);
current_thread = next;
}
2014-02-21 11:05:10 -05:00
static Thread* idle_thread;
static Thread* first_runnable_thread;
static Thread* true_current_thread;
2014-02-21 11:05:10 -05:00
static Process* init_process;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
static Thread* FindRunnableThreadWithSystemTid(uintptr_t system_tid)
2011-08-05 08:25:00 -04:00
{
Thread* begun_thread = current_thread;
Thread* iter = begun_thread;
do
{
if ( iter->system_tid == system_tid )
return iter;
iter = iter->scheduler_list_next;
} while ( iter != begun_thread );
return NULL;
}
static Thread* PopNextThread(bool yielded)
{
Thread* result;
uintptr_t yield_to_tid = current_thread->yield_to_tid;
if ( yielded && yield_to_tid != 0 )
{
if ( (result = FindRunnableThreadWithSystemTid(yield_to_tid)) )
return result;
}
2014-02-21 11:05:10 -05:00
if ( first_runnable_thread )
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
result = first_runnable_thread;
2014-02-21 11:05:10 -05:00
first_runnable_thread = first_runnable_thread->scheduler_list_next;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
else
{
result = idle_thread;
}
true_current_thread = result;
2014-02-21 11:05:10 -05:00
return result;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
2011-08-05 08:25:00 -04:00
static void RealSwitch(struct interrupt_context* intctx, bool yielded)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
SwitchThread(intctx, CurrentThread(), PopNextThread(yielded));
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
if ( intctx->signal_pending && InUserspace(intctx) )
{
Interrupt::Enable();
Signal::DispatchHandler(intctx, NULL);
}
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
void Switch(struct interrupt_context* intctx)
{
RealSwitch(intctx, false);
}
void InterruptYieldCPU(struct interrupt_context* intctx, void* /*user*/)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
RealSwitch(intctx, true);
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
void ThreadExitCPU(struct interrupt_context* intctx, void* /*user*/)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
2014-02-21 11:05:10 -05:00
SetThreadState(current_thread, ThreadState::DEAD);
RealSwitch(intctx, false);
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
// The idle thread serves no purpose except being an infinite loop that does
// nothing, which is only run when the system has nothing to do.
void SetIdleThread(Thread* thread)
{
2014-02-21 11:05:10 -05:00
assert(!idle_thread);
idle_thread = thread;
2013-01-09 04:47:22 -05:00
SetThreadState(thread, ThreadState::NONE);
2014-02-21 11:05:10 -05:00
current_thread = thread;
true_current_thread = thread;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
void SetInitProcess(Process* init)
{
2014-02-21 11:05:10 -05:00
init_process = init;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
Process* GetInitProcess()
{
2014-02-21 11:05:10 -05:00
return init_process;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
Process* GetKernelProcess()
{
2014-02-21 11:05:10 -05:00
return idle_thread->process;
}
2013-01-09 04:47:22 -05:00
void SetThreadState(Thread* thread, ThreadState state)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
bool wasenabled = Interrupt::SetEnabled(false);
// Remove the thread from the list of runnable threads.
2013-01-09 04:47:22 -05:00
if ( thread->state == ThreadState::RUNNABLE &&
state != ThreadState::RUNNABLE )
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
2014-02-21 11:05:10 -05:00
if ( thread == first_runnable_thread )
first_runnable_thread = thread->scheduler_list_next;
if ( thread == first_runnable_thread )
first_runnable_thread = NULL;
assert(thread->scheduler_list_prev);
assert(thread->scheduler_list_next);
thread->scheduler_list_prev->scheduler_list_next = thread->scheduler_list_next;
thread->scheduler_list_next->scheduler_list_prev = thread->scheduler_list_prev;
thread->scheduler_list_prev = NULL;
thread->scheduler_list_next = NULL;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
// Insert the thread into the scheduler's carousel linked list.
2013-01-09 04:47:22 -05:00
if ( thread->state != ThreadState::RUNNABLE &&
state == ThreadState::RUNNABLE )
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
2014-02-21 11:05:10 -05:00
if ( first_runnable_thread == NULL )
first_runnable_thread = thread;
thread->scheduler_list_prev = first_runnable_thread->scheduler_list_prev;
thread->scheduler_list_next = first_runnable_thread;
first_runnable_thread->scheduler_list_prev = thread;
thread->scheduler_list_prev->scheduler_list_next = thread;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
thread->state = state;
2014-02-21 11:05:10 -05:00
assert(thread->state != ThreadState::RUNNABLE || thread->scheduler_list_prev);
assert(thread->state != ThreadState::RUNNABLE || thread->scheduler_list_next);
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
Interrupt::SetEnabled(wasenabled);
}
2013-01-09 04:47:22 -05:00
ThreadState GetThreadState(Thread* thread)
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
{
return thread->state;
}
} // namespace Scheduler
} // namespace Sortix
namespace Sortix {
int sys_sched_yield(void)
2013-09-16 15:16:35 -04:00
{
return kthread_yield(), 0;
}
} // namespace Sortix
namespace Sortix {
namespace Scheduler {
void ScheduleTrueThread()
{
bool wasenabled = Interrupt::SetEnabled(false);
if ( true_current_thread != current_thread )
{
current_thread->yield_to_tid = 0;
first_runnable_thread = true_current_thread;
kthread_yield();
}
Interrupt::SetEnabled(wasenabled);
}
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
void Init()
{
2014-02-21 11:05:10 -05:00
first_runnable_thread = NULL;
true_current_thread = NULL;
2014-02-21 11:05:10 -05:00
idle_thread = NULL;
Multithreaded kernel and improvement of signal handling. Pardon the big ass-commit, this took months to develop and debug and the refactoring got so far that a clean merge became impossible. The good news is that this commit does quite a bit of cleaning up and generally improves the kernel quality. This makes the kernel fully pre-emptive and multithreaded. This was done by rewriting the interrupt code, the scheduler, introducing new threading primitives, and rewriting large parts of the kernel. During the past few commits the kernel has had its device drivers thread secured; this commit thread secures large parts of the core kernel. There still remains some parts of the kernel that is _not_ thread secured, but this is not a problem at this point. Each user-space thread has an associated kernel stack that it uses when it goes into kernel mode. This stack is by default 8 KiB since that value works for me and is also used by Linux. Strange things tends to happen on x86 in case of a stack overflow - there is no ideal way to catch such a situation right now. The system call conventions were changed, too. The %edx register is now used to provide the errno value of the call, instead of the kernel writing it into a registered global variable. The system call code has also been updated to better reflect the native calling conventions: not all registers have to be preserved. This makes system calls faster and simplifies the assembly. In the kernel, there is no longer the event.h header or the hacky method of 'resuming system calls' that closely resembles cooperative multitasking. If a system call wants to block, it should just block. The signal handling was also improved significantly. At this point, signals cannot interrupt kernel threads (but can always interrupt user-space threads if enabled), which introduces some problems with how a SIGINT could interrupt a blocking read, for instance. This commit introduces and uses a number of new primitives such as kthread_lock_mutex_signal() that attempts to get the lock but fails if a signal is pending. In this manner, the kernel is safer as kernel threads cannot be shut down inconveniently, but in return for complexity as blocking operations must check they if they should fail. Process exiting has also been refactored significantly. The _exit(2) system call sets the exit code and sends SIGKILL to all the threads in the process. Once all the threads have cleaned themselves up and exited, a worker thread calls the process's LastPrayer() method that unmaps memory, deletes the address space, notifies the parent, etc. This provides a very robust way to terminate processes as even half-constructed processes (during a failing fork for instance) can be gracefully terminated. I have introduced a number of kernel threads to help avoid threading problems and simplify kernel design. For instance, there is now a functional generic kernel worker thread that any kernel thread can schedule jobs for. Interrupt handlers run with interrupts off (hence they cannot call kthread_ functions as it may deadlock the system if another thread holds the lock) therefore they cannot use the standard kernel worker threads. Instead, they use a special purpose interrupt worker thread that works much like the generic one expect that interrupt handlers can safely queue work with interrupts off. Note that this also means that interrupt handlers cannot allocate memory or print to the kernel log/screen as such mechanisms uses locks. I'll introduce a lock free algorithm for such cases later on. The boot process has also changed. The original kernel init thread in kernel.cpp creates a new bootstrap thread and becomes the system idle thread. Note that pid=0 now means the kernel, as there is no longer a system idle process. The bootstrap thread launches all the kernel worker threads and then creates a new process and loads /bin/init into it and then creates a thread in pid=1, which starts the system. The bootstrap thread then quietly waits for pid=1 to exit after which it shuts down/reboots/panics the system. In general, the introduction of race conditions and dead locks have forced me to revise a lot of the design and make sure it was thread secure. Since early parts of the kernel was quite hacky, I had to refactor such code. So it seems that the risk of dead locks forces me to write better code. Note that a real preemptive multithreaded kernel simplifies the construction of blocking system calls. My hope is that this will trigger a clean up of the filesystem code that current is almost beyond repair. Almost all of the kernel was modified during this refactoring. To the extent possible, these changes have been backported to older non-multithreaded kernel, but many changes were tightly coupled and went into this commit. Of interest is the implementation of the kthread_ api based on the design of pthreads; this library allows easy synchronization mechanisms and includes C++-style scoped locks. This commit also introduces new worker threads and tested mechanisms for interrupt handlers to schedule work in a kernel worker thread. A lot of code have been rewritten from scratch and has become a lot more stable and correct. Share and enjoy!
2012-08-01 11:30:34 -04:00
}
} // namespace Scheduler
} // namespace Sortix
2014-02-21 11:05:10 -05:00
namespace Sortix {
Thread* CurrentThread()
{
return Scheduler::current_thread;
}
Process* CurrentProcess()
{
return CurrentThread()->process;
}
} // namespace Sortix