diff -urN linux-2.5.1/CREDITS linux/CREDITS
--- linux-2.5.1/CREDITS Sun Dec 16 17:39:29 2001
+++ linux/CREDITS Sun Dec 16 17:41:45 2001
@@ -973,8 +973,8 @@
N: Nigel Gamble
E: nigel@nrg.org
-E: nigel@sgi.com
D: Interrupt-driven printer driver
+D: Preemptible kernel
S: 120 Alley Way
S: Mountain View, California 94040
S: USA
diff -urN linux-2.5.1/Documentation/Configure.help linux/Documentation/Configure.help
--- linux-2.5.1/Documentation/Configure.help Sun Dec 16 17:40:22 2001
+++ linux/Documentation/Configure.help Sun Dec 16 17:41:45 2001
@@ -266,6 +266,19 @@
If you have a system with several CPUs, you do not need to say Y
here: the local APIC will be used automatically.
+Preemptible Kernel
+CONFIG_PREEMPT
+ This option reduces the latency of the kernel when reacting to
+ real-time or interactive events by allowing a low priority process to
+ be preempted even if it is in kernel mode executing a system call.
+ This allows applications to run more reliably even when the system is
+ under load due to other, lower priority, processes.
+
+ Say Y here if you are building a kernel for a desktop system, embedded
+ system or real-time system. Say N if you are building a kernel for a
+ system where throughput is more important than interactive response,
+ such as a server system. Say N if you are unsure.
+
Kernel math emulation
CONFIG_MATH_EMULATION
Linux can emulate a math coprocessor (used for floating point
diff -urN linux-2.5.1/Documentation/preempt-locking.txt linux/Documentation/preempt-locking.txt
--- linux-2.5.1/Documentation/preempt-locking.txt Wed Dec 31 19:00:00 1969
+++ linux/Documentation/preempt-locking.txt Sun Dec 16 17:41:45 2001
@@ -0,0 +1,94 @@
+ Proper Locking Under a Preemptible Kernel:
+ Keeping Kernel Code Preempt-Safe
+ Robert Love <rml@tech9.net>
+ Last Updated: 21 Oct 2001
+
+
+INTRODUCTION
+
+
+A preemptible kernel creates new locking issues. The issues are the same as
+those under SMP: concurrency and reentrancy. Thankfully, the Linux preemptible
+kernel model leverages existing SMP locking mechanisms. Thus, the kernel
+requires explicit additional locking for very few additional situations.
+
+This document is for all kernel hackers. Developing code in the kernel
+requires protecting these situations. As you will see, these situations would
+normally require a lock, where they not per-CPU.
+
+
+RULE #1: Per-CPU data structures need explicit protection
+
+
+Two similar problems arise. An example code snippet:
+
+ struct this_needs_locking tux[NR_CPUS];
+ tux[smp_processor_id()] = some_value;
+ /* task is preempted here... */
+ something = tux[smp_processor_id()];
+
+First, since the data is per-CPU, it may not have explicit SMP locking, but
+require it otherwise. Second, when a preempted task is finally rescheduled,
+the previous value of smp_processor_id may not equal the current. You must
+protect these situations by disabling preemption around them.
+
+
+RULE #2: CPU state must be protected.
+
+
+Under preemption, the state of the CPU must be protected. This is arch-
+dependent, but includes CPU structures and state not preserved over a context
+switch. For example, on x86, entering and exiting FPU mode is now a critical
+section that must occur while preemption is disabled. Think what would happen
+if the kernel is executing a floating-point instruction and is then preempted.
+Remember, the kernel does not save FPU state except for user tasks. Therefore,
+upon preemption, the FPU registers will be sold to the lowest bidder. Thus,
+preemption must be disabled around such regions.i
+
+Note, some FPU functions are already explicitly preempt safe. For example,
+kernel_fpu_begin and kernel_fpu_end will disable and enable preemption.
+However, math_state_restore must be called with preemption disabled.
+
+
+SOLUTION
+
+
+Data protection under preemption is achieved by disabling preemption for the
+duration of the critical region.
+
+preempt_enable() decrement the preempt counter
+preempt_disable() increment the preempt counter
+preempt_enable_no_resched() decrement, but do not immediately preempt
+
+The functions are nestable. In other words, you can call preempt_disable
+n-times in a code path, and preemption will not be reenabled until the n-th
+call to preempt_enable. The preempt statements define to nothing if
+preemption is not enabled.
+
+Note that you do not need to explicitly prevent preemption if you are holding
+any locks or interrupts are disabled, since preemption is implicitly disabled
+in those cases.
+
+Example:
+
+ cpucache_t *cc; /* this is per-CPU */
+ preempt_disable();
+ cc = cc_data(searchp);
+ if (cc && cc->avail) {
+ __free_block(searchp, cc_entry(cc), cc->avail);
+ cc->avail = 0;
+ }
+ preempt_enable();
+ return 0;
+
+Notice how the preemption statements must encompass every reference of the
+critical variables. Another example:
+
+ int buf[NR_CPUS];
+ set_cpu_val(buf);
+ if (buf[smp_processor_id()] == -1) printf(KERN_INFO "wee!\n");
+ spin_lock(&buf_lock);
+ /* ... */
+
+This code is not preempt-safe, but see how easily we can fix it by simply
+moving the spin_lock up two lines.
diff -urN linux-2.5.1/MAINTAINERS linux/MAINTAINERS
--- linux-2.5.1/MAINTAINERS Sun Dec 16 17:39:39 2001
+++ linux/MAINTAINERS Sun Dec 16 17:41:45 2001
@@ -1230,6 +1230,14 @@
M: mostrows@styx.uwaterloo.ca
S: Maintained
+PREEMPTIBLE KERNEL
+P: Robert M. Love
+M: rml@tech9.net
+L: linux-kernel@vger.kernel.org
+L: kpreempt-tech@lists.sourceforge.net
+W: http://tech9.net/rml/linux
+S: Maintained
+
PROMISE DC4030 CACHING DISK CONTROLLER DRIVER
P: Peter Denison
M: promise@pnd-pc.demon.co.uk
diff -urN linux-2.5.1/arch/arm/config.in linux/arch/arm/config.in
--- linux-2.5.1/arch/arm/config.in Sun Dec 16 17:40:13 2001
+++ linux/arch/arm/config.in Sun Dec 16 17:41:45 2001
@@ -437,6 +437,7 @@
if [ "$CONFIG_CPU_32" = "y" -a "$CONFIG_ARCH_EBSA110" != "y" ]; then
bool 'Kernel-mode alignment trap handler' CONFIG_ALIGNMENT_TRAP
fi
+dep_bool 'Preemptible Kernel (experimental)' CONFIG_PREEMPT $CONFIG_CPU_32 $CONFIG_EXPERIMENTAL
endmenu
source drivers/parport/Config.in
diff -urN linux-2.5.1/arch/arm/kernel/entry-armv.S linux/arch/arm/kernel/entry-armv.S
--- linux-2.5.1/arch/arm/kernel/entry-armv.S Sun Dec 16 17:40:13 2001
+++ linux/arch/arm/kernel/entry-armv.S Sun Dec 16 17:41:45 2001
@@ -672,6 +672,12 @@
add r4, sp, #S_SP
mov r6, lr
stmia r4, {r5, r6, r7, r8, r9} @ save sp_SVC, lr_SVC, pc, cpsr, old_ro
+#ifdef CONFIG_PREEMPT
+ get_current_task r9
+ ldr r8, [r9, #TSK_PREEMPT]
+ add r8, r8, #1
+ str r8, [r9, #TSK_PREEMPT]
+#endif
1: get_irqnr_and_base r0, r6, r5, lr
movne r1, sp
@
@@ -679,6 +685,25 @@
@
adrsvc ne, lr, 1b
bne do_IRQ
+#ifdef CONFIG_PREEMPT
+2: ldr r8, [r9, #TSK_PREEMPT]
+ subs r8, r8, #1
+ bne 3f
+ ldr r7, [r9, #TSK_NEED_RESCHED]
+ teq r7, #0
+ beq 3f
+ ldr r6, .LCirqstat
+ ldr r0, [r6, #IRQSTAT_BH_COUNT]
+ teq r0, #0
+ bne 3f
+ mov r0, #MODE_SVC
+ msr cpsr_c, r0 @ enable interrupts
+ bl SYMBOL_NAME(preempt_schedule)
+ mov r0, #I_BIT | MODE_SVC
+ msr cpsr_c, r0 @ disable interrupts
+ b 2b
+3: str r8, [r9, #TSK_PREEMPT]
+#endif
ldr r0, [sp, #S_PSR] @ irqs are already disabled
msr spsr, r0
ldmia sp, {r0 - pc}^ @ load r0 - pc, cpsr
@@ -736,6 +761,9 @@
.LCprocfns: .word SYMBOL_NAME(processor)
#endif
.LCfp: .word SYMBOL_NAME(fp_enter)
+#ifdef CONFIG_PREEMPT
+.LCirqstat: .word SYMBOL_NAME(irq_stat)
+#endif
irq_prio_table
@@ -775,6 +803,12 @@
stmdb r8, {sp, lr}^
alignment_trap r4, r7, __temp_irq
zero_fp
+ get_current_task tsk
+#ifdef CONFIG_PREEMPT
+ ldr r0, [tsk, #TSK_PREEMPT]
+ add r0, r0, #1
+ str r0, [tsk, #TSK_PREEMPT]
+#endif
1: get_irqnr_and_base r0, r6, r5, lr
movne r1, sp
adrsvc ne, lr, 1b
@@ -782,8 +816,12 @@
@ routine called with r0 = irq number, r1 = struct pt_regs *
@
bne do_IRQ
+#ifdef CONFIG_PREEMPT
+ ldr r0, [tsk, #TSK_PREEMPT]
+ sub r0, r0, #1
+ str r0, [tsk, #TSK_PREEMPT]
+#endif
mov why, #0
- get_current_task tsk
b ret_to_user
.align 5
diff -urN linux-2.5.1/arch/arm/tools/getconstants.c linux/arch/arm/tools/getconstants.c
--- linux-2.5.1/arch/arm/tools/getconstants.c Sun Dec 16 17:40:14 2001
+++ linux/arch/arm/tools/getconstants.c Sun Dec 16 17:41:45 2001
@@ -13,6 +13,7 @@
#include <asm/pgtable.h>
#include <asm/uaccess.h>
+#include <asm/hardirq.h>
/*
* Make sure that the compiler and target are compatible.
@@ -39,6 +40,11 @@
DEFN("TSS_SAVE", OFF_TSK(thread.save));
DEFN("TSS_FPESAVE", OFF_TSK(thread.fpstate.soft.save));
+#ifdef CONFIG_PREEMPT
+DEFN("TSK_PREEMPT", OFF_TSK(preempt_count));
+DEFN("IRQSTAT_BH_COUNT", (unsigned long)&(((irq_cpustat_t *)0)->__local_bh_count));
+#endif
+
#ifdef CONFIG_CPU_32
DEFN("TSS_DOMAIN", OFF_TSK(thread.domain));
diff -urN linux-2.5.1/arch/i386/config.in linux/arch/i386/config.in
--- linux-2.5.1/arch/i386/config.in Sun Dec 16 17:40:08 2001
+++ linux/arch/i386/config.in Sun Dec 16 17:41:45 2001
@@ -170,6 +170,7 @@
bool 'Math emulation' CONFIG_MATH_EMULATION
bool 'MTRR (Memory Type Range Register) support' CONFIG_MTRR
bool 'Symmetric multi-processing support' CONFIG_SMP
+bool 'Preemptible Kernel' CONFIG_PREEMPT
if [ "$CONFIG_SMP" != "y" ]; then
bool 'Local APIC support on uniprocessors' CONFIG_X86_UP_APIC
dep_bool 'IO-APIC support on uniprocessors' CONFIG_X86_UP_IOAPIC $CONFIG_X86_UP_APIC
@@ -183,9 +184,12 @@
bool 'Multiquad NUMA system' CONFIG_MULTIQUAD
fi
-if [ "$CONFIG_SMP" = "y" -a "$CONFIG_X86_CMPXCHG" = "y" ]; then
- define_bool CONFIG_HAVE_DEC_LOCK y
+if [ "$CONFIG_SMP" = "y" -o "$CONFIG_PREEMPT" = "y" ]; then
+ if [ "$CONFIG_X86_CMPXCHG" = "y" ]; then
+ define_bool CONFIG_HAVE_DEC_LOCK y
+ fi
fi
+
endmenu
mainmenu_option next_comment
diff -urN linux-2.5.1/arch/i386/kernel/entry.S linux/arch/i386/kernel/entry.S
--- linux-2.5.1/arch/i386/kernel/entry.S Sun Dec 16 17:40:08 2001
+++ linux/arch/i386/kernel/entry.S Sun Dec 16 17:41:45 2001
@@ -71,7 +71,7 @@
* these are offsets into the task-struct.
*/
state = 0
-flags = 4
+preempt_count = 4
sigpending = 8
addr_limit = 12
exec_domain = 16
@@ -79,8 +79,28 @@
tsk_ptrace = 24
processor = 52
+ /* These are offsets into the irq_stat structure
+ * There is one per cpu and it is aligned to 32
+ * byte boundry (we put that here as a shift count)
+ */
+irq_array_shift = CONFIG_X86_L1_CACHE_SHIFT
+
+irq_stat_local_irq_count = 4
+irq_stat_local_bh_count = 8
+
ENOSYS = 38
+#ifdef CONFIG_SMP
+#define GET_CPU_INDX movl processor(%ebx),%eax; \
+ shll $irq_array_shift,%eax
+#define GET_CURRENT_CPU_INDX GET_CURRENT(%ebx); \
+ GET_CPU_INDX
+#define CPU_INDX (,%eax)
+#else
+#define GET_CPU_INDX
+#define GET_CURRENT_CPU_INDX GET_CURRENT(%ebx)
+#define CPU_INDX
+#endif
#define SAVE_ALL \
cld; \
@@ -247,12 +267,30 @@
ALIGN
ENTRY(ret_from_intr)
GET_CURRENT(%ebx)
+#ifdef CONFIG_PREEMPT
+ cli
+ decl preempt_count(%ebx)
+#endif
ret_from_exception:
movl EFLAGS(%esp),%eax # mix EFLAGS and CS
movb CS(%esp),%al
testl $(VM_MASK | 3),%eax # return to VM86 mode or non-supervisor?
jne ret_from_sys_call
+#ifdef CONFIG_PREEMPT
+ cmpl $0,preempt_count(%ebx)
+ jnz restore_all
+ cmpl $0,need_resched(%ebx)
+ jz restore_all
+ movl SYMBOL_NAME(irq_stat)+irq_stat_local_bh_count CPU_INDX,%ecx
+ addl SYMBOL_NAME(irq_stat)+irq_stat_local_irq_count CPU_INDX,%ecx
+ jnz restore_all
+ incl preempt_count(%ebx)
+ sti
+ call SYMBOL_NAME(preempt_schedule)
+ jmp ret_from_intr
+#else
jmp restore_all
+#endif
ALIGN
reschedule:
@@ -289,6 +327,9 @@
GET_CURRENT(%ebx)
call *%edi
addl $8,%esp
+#ifdef CONFIG_PREEMPT
+ cli
+#endif
jmp ret_from_exception
ENTRY(coprocessor_error)
@@ -308,12 +349,18 @@
movl %cr0,%eax
testl $0x4,%eax # EM (math emulation bit)
jne device_not_available_emulate
+#ifdef CONFIG_PREEMPT
+ cli
+#endif
call SYMBOL_NAME(math_state_restore)
jmp ret_from_exception
device_not_available_emulate:
pushl $0 # temporary storage for ORIG_EIP
call SYMBOL_NAME(math_emulate)
addl $4,%esp
+#ifdef CONFIG_PREEMPT
+ cli
+#endif
jmp ret_from_exception
ENTRY(debug)
diff -urN linux-2.5.1/arch/i386/kernel/i387.c linux/arch/i386/kernel/i387.c
--- linux-2.5.1/arch/i386/kernel/i387.c Sun Dec 16 17:40:09 2001
+++ linux/arch/i386/kernel/i387.c Sun Dec 16 17:41:45 2001
@@ -10,6 +10,7 @@
#include <linux/config.h>
#include <linux/sched.h>
+#include <linux/spinlock.h>
#include <asm/processor.h>
#include <asm/i387.h>
#include <asm/math_emu.h>
@@ -65,6 +66,8 @@
{
struct task_struct *tsk = current;
+ preempt_disable();
+
if (tsk->flags & PF_USEDFPU) {
__save_init_fpu(tsk);
return;
diff -urN linux-2.5.1/arch/i386/kernel/traps.c linux/arch/i386/kernel/traps.c
--- linux-2.5.1/arch/i386/kernel/traps.c Sun Dec 16 17:40:08 2001
+++ linux/arch/i386/kernel/traps.c Sun Dec 16 17:42:50 2001
@@ -694,6 +694,8 @@
*
* Careful.. There are problems with IBM-designed IRQ13 behaviour.
* Don't touch unless you *really* know how it works.
+ *
+ * Must be called with kernel preemption disabled.
*/
asmlinkage void math_state_restore(struct pt_regs regs)
{
diff -urN linux-2.5.1/arch/i386/lib/dec_and_lock.c linux/arch/i386/lib/dec_and_lock.c
--- linux-2.5.1/arch/i386/lib/dec_and_lock.c Sun Dec 16 17:40:08 2001
+++ linux/arch/i386/lib/dec_and_lock.c Sun Dec 16 17:41:45 2001
@@ -8,6 +8,7 @@
*/
#include <linux/spinlock.h>
+#include <linux/sched.h>
#include <asm/atomic.h>
int atomic_dec_and_lock(atomic_t *atomic, spinlock_t *lock)
diff -urN linux-2.5.1/arch/sh/config.in linux/arch/sh/config.in
--- linux-2.5.1/arch/sh/config.in Sun Dec 16 17:40:14 2001
+++ linux/arch/sh/config.in Sun Dec 16 17:41:45 2001
@@ -124,6 +124,8 @@
hex 'Physical memory start address' CONFIG_MEMORY_START 08000000
hex 'Physical memory size' CONFIG_MEMORY_SIZE 00400000
fi
+# Preemptible kernel feature
+bool 'Preemptible Kernel' CONFIG_PREEMPT
endmenu
if [ "$CONFIG_SH_HP690" = "y" ]; then
diff -urN linux-2.5.1/arch/sh/kernel/entry.S linux/arch/sh/kernel/entry.S
--- linux-2.5.1/arch/sh/kernel/entry.S Sun Dec 16 17:40:14 2001
+++ linux/arch/sh/kernel/entry.S Sun Dec 16 17:41:45 2001
@@ -60,10 +60,18 @@
/*
* These are offsets into the task-struct.
*/
-flags = 4
+preempt_count = 4
sigpending = 8
need_resched = 20
tsk_ptrace = 24
+flags = 84
+
+/*
+ * And these offsets are into irq_stat.
+ * (Find irq_cpustat_t in asm-sh/hardirq.h)
+ */
+local_irq_count = 8
+local_bh_count = 12
PT_TRACESYS = 0x00000002
PF_USEDFPU = 0x00100000
@@ -143,7 +151,7 @@
mov.l __INV_IMASK, r11; \
stc sr, r10; \
and r11, r10; \
- stc k_g_imask, r11; \
+ stc k_g_imask, r11; \
or r11, r10; \
ldc r10, sr
@@ -304,8 +312,8 @@
mov.l @(tsk_ptrace,r0), r0 ! Is current PTRACE_SYSCALL'd?
mov #PT_TRACESYS, r1
tst r1, r0
- bt ret_from_syscall
- bra syscall_ret_trace
+ bf syscall_ret_trace
+ bra ret_from_syscall
nop
.align 2
@@ -505,8 +513,6 @@
.long syscall_ret_trace
__syscall_ret:
.long syscall_ret
-__INV_IMASK:
- .long 0xffffff0f ! ~(IMASK)
.align 2
@@ -518,7 +524,84 @@
.align 2
1: .long SYMBOL_NAME(schedule)
+#ifdef CONFIG_PREEMPT
+ !
+ ! Returning from interrupt during kernel mode: check if
+ ! preempt_schedule should be called. If need_resched flag
+ ! is set, preempt_count is zero, and we're not currently
+ ! in an interrupt handler (local irq or bottom half) then
+ ! call preempt_schedule.
+ !
+ ! Increment preempt_count to prevent a nested interrupt
+ ! from reentering preempt_schedule, then decrement after
+ ! and drop through to regular interrupt return which will
+ ! jump back and check again in case such an interrupt did
+ ! come in (and didn't preempt due to preempt_count).
+ !
+ ! NOTE: because we just checked that preempt_count was
+ ! zero before getting to the call, can't we use immediate
+ ! values (1 and 0) rather than inc/dec? Also, rather than
+ ! drop through to ret_from_irq, we already know this thread
+ ! is kernel mode, can't we go direct to ret_from_kirq? In
+ ! fact, with proper interrupt nesting and so forth could
+ ! the loop simply be on the need_resched w/o checking the
+ ! other stuff again? Optimize later...
+ !
+ .align 2
+ret_from_kirq:
+ ! Nonzero preempt_count prevents scheduling
+ stc k_current, r1
+ mov.l @(preempt_count,r1), r0
+ cmp/eq #0, r0
+ bf restore_all
+ ! Zero need_resched prevents scheduling
+ mov.l @(need_resched,r1), r0
+ cmp/eq #0, r0
+ bt restore_all
+ ! If in_interrupt(), don't schedule
+ mov.l __irq_stat, r1
+ mov.l @(local_irq_count,r1), r0
+ mov.l @(local_bh_count,r1), r1
+ or r1, r0
+ cmp/eq #0, r0
+ bf restore_all
+ ! Allow scheduling using preempt_schedule
+ ! Adjust preempt_count and SR as needed.
+ stc k_current, r1
+ mov.l @(preempt_count,r1), r0 ! Could replace this ...
+ add #1, r0 ! ... and this w/mov #1?
+ mov.l r0, @(preempt_count,r1)
+ STI()
+ mov.l __preempt_schedule, r0
+ jsr @r0
+ nop
+ /* CLI */
+ stc sr, r0
+ or #0xf0, r0
+ ldc r0, sr
+ !
+ stc k_current, r1
+ mov.l @(preempt_count,r1), r0 ! Could replace this ...
+ add #-1, r0 ! ... and this w/mov #0?
+ mov.l r0, @(preempt_count,r1)
+ ! Maybe should bra ret_from_kirq, or loop over need_resched?
+ ! For now, fall through to ret_from_irq again...
+#endif /* CONFIG_PREEMPT */
+
ret_from_irq:
+ mov #OFF_SR, r0
+ mov.l @(r0,r15), r0 ! get status register
+ shll r0
+ shll r0 ! kernel space?
+#ifndef CONFIG_PREEMPT
+ bt restore_all ! Yes, it's from kernel, go back soon
+#else /* CONFIG_PREEMPT */
+ bt ret_from_kirq ! From kernel: maybe preempt_schedule
+#endif /* CONFIG_PREEMPT */
+ !
+ bra ret_from_syscall
+ nop
+
ret_from_exception:
mov #OFF_SR, r0
mov.l @(r0,r15), r0 ! get status register
@@ -564,6 +647,13 @@
.long SYMBOL_NAME(do_signal)
__irq_stat:
.long SYMBOL_NAME(irq_stat)
+#ifdef CONFIG_PREEMPT
+__preempt_schedule:
+ .long SYMBOL_NAME(preempt_schedule)
+#endif /* CONFIG_PREEMPT */
+__INV_IMASK:
+ .long 0xffffff0f ! ~(IMASK)
+
.align 2
restore_all:
@@ -679,7 +769,7 @@
__fpu_prepare_fd:
.long SYMBOL_NAME(fpu_prepare_fd)
__init_task_flags:
- .long SYMBOL_NAME(init_task_union)+4
+ .long SYMBOL_NAME(init_task_union)+flags
__PF_USEDFPU:
.long PF_USEDFPU
#endif
diff -urN linux-2.5.1/arch/sh/kernel/irq.c linux/arch/sh/kernel/irq.c
--- linux-2.5.1/arch/sh/kernel/irq.c Sun Dec 16 17:40:14 2001
+++ linux/arch/sh/kernel/irq.c Sun Dec 16 17:41:45 2001
@@ -229,6 +229,14 @@
struct irqaction * action;
unsigned int status;
+ /*
+ * At this point we're now about to actually call handlers,
+ * and interrupts might get reenabled during them... bump
+ * preempt_count to prevent any preemption while the handler
+ * called here is pending...
+ */
+ preempt_disable();
+
/* Get IRQ number */
asm volatile("stc r2_bank, %0\n\t"
"shlr2 %0\n\t"
@@ -298,8 +306,17 @@
desc->handler->end(irq);
spin_unlock(&desc->lock);
+
if (softirq_pending(cpu))
do_softirq();
+
+ /*
+ * We're done with the handlers, interrupts should be
+ * currently disabled; decrement preempt_count now so
+ * as we return preemption may be allowed...
+ */
+ preempt_enable_no_resched();
+
return 1;
}
diff -urN linux-2.5.1/drivers/char/console.c linux/drivers/char/console.c
--- linux-2.5.1/drivers/char/console.c Sun Dec 16 17:39:42 2001
+++ linux/drivers/char/console.c Sun Dec 16 17:41:45 2001
@@ -2356,8 +2356,14 @@
return;
pm_access(pm_con);
+
+ /*
+ * If we raced with con_close(), `vt' may be null.
+ * Hence this bandaid. - akpm
+ */
acquire_console_sem();
- set_cursor(vt->vc_num);
+ if (vt)
+ set_cursor(vt->vc_num);
release_console_sem();
}
diff -urN linux-2.5.1/drivers/ieee1394/csr.c linux/drivers/ieee1394/csr.c
--- linux-2.5.1/drivers/ieee1394/csr.c Sun Dec 16 17:40:02 2001
+++ linux/drivers/ieee1394/csr.c Sun Dec 16 17:41:45 2001
@@ -10,6 +10,7 @@
*/
#include <linux/string.h>
+#include <linux/sched.h>
#include "ieee1394_types.h"
#include "hosts.h"
diff -urN linux-2.5.1/fs/adfs/map.c linux/fs/adfs/map.c
--- linux-2.5.1/fs/adfs/map.c Sun Dec 16 17:39:29 2001
+++ linux/fs/adfs/map.c Sun Dec 16 17:41:45 2001
@@ -12,6 +12,7 @@
#include <linux/fs.h>
#include <linux/adfs_fs.h>
#include <linux/spinlock.h>
+#include <linux/sched.h>
#include "adfs.h"
diff -urN linux-2.5.1/fs/exec.c linux/fs/exec.c
--- linux-2.5.1/fs/exec.c Sun Dec 16 17:39:28 2001
+++ linux/fs/exec.c Sun Dec 16 17:41:45 2001
@@ -420,8 +420,8 @@
active_mm = current->active_mm;
current->mm = mm;
current->active_mm = mm;
- task_unlock(current);
activate_mm(active_mm, mm);
+ task_unlock(current);
mm_release();
if (old_mm) {
if (active_mm != old_mm) BUG();
diff -urN linux-2.5.1/fs/fat/cache.c linux/fs/fat/cache.c
--- linux-2.5.1/fs/fat/cache.c Sun Dec 16 17:39:28 2001
+++ linux/fs/fat/cache.c Sun Dec 16 17:41:45 2001
@@ -14,6 +14,7 @@
#include <linux/string.h>
#include <linux/stat.h>
#include <linux/fat_cvf.h>
+#include <linux/sched.h>
#if 0
# define PRINTK(x) printk x
diff -urN linux-2.5.1/include/asm-arm/dma.h linux/include/asm-arm/dma.h
--- linux-2.5.1/include/asm-arm/dma.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/dma.h Sun Dec 16 17:41:45 2001
@@ -5,6 +5,7 @@
#include <linux/config.h>
#include <linux/spinlock.h>
+#include <linux/sched.h>
#include <asm/system.h>
#include <asm/memory.h>
#include <asm/scatterlist.h>
diff -urN linux-2.5.1/include/asm-arm/hardirq.h linux/include/asm-arm/hardirq.h
--- linux-2.5.1/include/asm-arm/hardirq.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/hardirq.h Sun Dec 16 17:41:45 2001
@@ -34,6 +34,7 @@
#define irq_exit(cpu,irq) (local_irq_count(cpu)--)
#define synchronize_irq() do { } while (0)
+#define release_irqlock(cpu) do { } while (0)
#else
#error SMP not supported
diff -urN linux-2.5.1/include/asm-arm/mmu_context.h linux/include/asm-arm/mmu_context.h
--- linux-2.5.1/include/asm-arm/mmu_context.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/mmu_context.h Sun Dec 16 17:41:45 2001
@@ -42,6 +42,10 @@
switch_mm(struct mm_struct *prev, struct mm_struct *next,
struct task_struct *tsk, unsigned int cpu)
{
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disable() == 0)
+ BUG();
+#endif
if (prev != next) {
cpu_switch_mm(next->pgd, tsk);
clear_bit(cpu, &prev->cpu_vm_mask);
diff -urN linux-2.5.1/include/asm-arm/pgalloc.h linux/include/asm-arm/pgalloc.h
--- linux-2.5.1/include/asm-arm/pgalloc.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/pgalloc.h Sun Dec 16 17:41:45 2001
@@ -57,40 +57,48 @@
{
unsigned long *ret;
+ preempt_disable();
if ((ret = pgd_quicklist) != NULL) {
pgd_quicklist = (unsigned long *)__pgd_next(ret);
ret[1] = ret[2];
clean_dcache_entry(ret + 1);
pgtable_cache_size--;
}
+ preempt_enable();
return (pgd_t *)ret;
}
static inline void free_pgd_fast(pgd_t *pgd)
{
+ preempt_disable();
__pgd_next(pgd) = (unsigned long) pgd_quicklist;
pgd_quicklist = (unsigned long *) pgd;
pgtable_cache_size++;
+ preempt_enable();
}
static inline pte_t *pte_alloc_one_fast(struct mm_struct *mm, unsigned long address)
{
unsigned long *ret;
+ preempt_disable();
if((ret = pte_quicklist) != NULL) {
pte_quicklist = (unsigned long *)__pte_next(ret);
ret[0] = 0;
clean_dcache_entry(ret);
pgtable_cache_size--;
}
+ preempt_enable();
return (pte_t *)ret;
}
static inline void free_pte_fast(pte_t *pte)
{
+ preempt_disable();
__pte_next(pte) = (unsigned long) pte_quicklist;
pte_quicklist = (unsigned long *) pte;
pgtable_cache_size++;
+ preempt_enable();
}
#else /* CONFIG_NO_PGT_CACHE */
diff -urN linux-2.5.1/include/asm-arm/smplock.h linux/include/asm-arm/smplock.h
--- linux-2.5.1/include/asm-arm/smplock.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/smplock.h Sun Dec 16 17:41:45 2001
@@ -3,12 +3,17 @@
*
* Default SMP lock implementation
*/
+#include <linux/config.h>
#include <linux/interrupt.h>
#include <linux/spinlock.h>
extern spinlock_t kernel_flag;
+#ifdef CONFIG_PREEMPT
+#define kernel_locked() preempt_is_disable()
+#else
#define kernel_locked() spin_is_locked(&kernel_flag)
+#endif
/*
* Release global kernel lock and global interrupt lock
@@ -40,8 +45,14 @@
*/
static inline void lock_kernel(void)
{
+#ifdef CONFIG_PREEMPT
+ if (current->lock_depth == -1)
+ spin_lock(&kernel_flag);
+ ++current->lock_depth;
+#else
if (!++current->lock_depth)
spin_lock(&kernel_flag);
+#endif
}
static inline void unlock_kernel(void)
diff -urN linux-2.5.1/include/asm-arm/softirq.h linux/include/asm-arm/softirq.h
--- linux-2.5.1/include/asm-arm/softirq.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-arm/softirq.h Sun Dec 16 17:41:45 2001
@@ -5,20 +5,22 @@
#include <asm/hardirq.h>
#define __cpu_bh_enable(cpu) \
- do { barrier(); local_bh_count(cpu)--; } while (0)
+ do { barrier(); local_bh_count(cpu)--; preempt_enable(); } while (0)
#define cpu_bh_disable(cpu) \
- do { local_bh_count(cpu)++; barrier(); } while (0)
+ do { preempt_disable(); local_bh_count(cpu)++; barrier(); } while (0)
#define local_bh_disable() cpu_bh_disable(smp_processor_id())
#define __local_bh_enable() __cpu_bh_enable(smp_processor_id())
#define in_softirq() (local_bh_count(smp_processor_id()) != 0)
-#define local_bh_enable() \
+#define _local_bh_enable() \
do { \
unsigned int *ptr = &local_bh_count(smp_processor_id()); \
if (!--*ptr && ptr[-2]) \
__asm__("bl%? __do_softirq": : : "lr");/* out of line */\
} while (0)
+#define local_bh_enable() do { _local_bh_enable(); preempt_enable(); } while (0)
+
#endif /* __ASM_SOFTIRQ_H */
diff -urN linux-2.5.1/include/asm-i386/hardirq.h linux/include/asm-i386/hardirq.h
--- linux-2.5.1/include/asm-i386/hardirq.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/hardirq.h Sun Dec 16 17:41:45 2001
@@ -36,6 +36,8 @@
#define synchronize_irq() barrier()
+#define release_irqlock(cpu) do { } while (0)
+
#else
#include <asm/atomic.h>
diff -urN linux-2.5.1/include/asm-i386/highmem.h linux/include/asm-i386/highmem.h
--- linux-2.5.1/include/asm-i386/highmem.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/highmem.h Sun Dec 16 17:41:45 2001
@@ -88,6 +88,7 @@
enum fixed_addresses idx;
unsigned long vaddr;
+ preempt_disable();
if (page < highmem_start_page)
return page_address(page);
@@ -109,8 +110,10 @@
unsigned long vaddr = (unsigned long) kvaddr;
enum fixed_addresses idx = type + KM_TYPE_NR*smp_processor_id();
- if (vaddr < FIXADDR_START) // FIXME
+ if (vaddr < FIXADDR_START) { // FIXME
+ preempt_enable();
return;
+ }
if (vaddr != __fix_to_virt(FIX_KMAP_BEGIN+idx))
BUG();
@@ -122,6 +125,8 @@
pte_clear(kmap_pte-idx);
__flush_tlb_one(vaddr);
#endif
+
+ preempt_enable();
}
#endif /* __KERNEL__ */
diff -urN linux-2.5.1/include/asm-i386/hw_irq.h linux/include/asm-i386/hw_irq.h
--- linux-2.5.1/include/asm-i386/hw_irq.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/hw_irq.h Sun Dec 16 17:41:45 2001
@@ -95,6 +95,18 @@
#define __STR(x) #x
#define STR(x) __STR(x)
+#define GET_CURRENT \
+ "movl %esp, %ebx\n\t" \
+ "andl $-8192, %ebx\n\t"
+
+#ifdef CONFIG_PREEMPT
+#define BUMP_LOCK_COUNT \
+ GET_CURRENT \
+ "incl 4(%ebx)\n\t"
+#else
+#define BUMP_LOCK_COUNT
+#endif
+
#define SAVE_ALL \
"cld\n\t" \
"pushl %es\n\t" \
@@ -108,15 +120,12 @@
"pushl %ebx\n\t" \
"movl $" STR(__KERNEL_DS) ",%edx\n\t" \
"movl %edx,%ds\n\t" \
- "movl %edx,%es\n\t"
+ "movl %edx,%es\n\t" \
+ BUMP_LOCK_COUNT
#define IRQ_NAME2(nr) nr##_interrupt(void)
#define IRQ_NAME(nr) IRQ_NAME2(IRQ##nr)
-#define GET_CURRENT \
- "movl %esp, %ebx\n\t" \
- "andl $-8192, %ebx\n\t"
-
/*
* SMP has a few special interrupts for IPI messages
*/
diff -urN linux-2.5.1/include/asm-i386/i387.h linux/include/asm-i386/i387.h
--- linux-2.5.1/include/asm-i386/i387.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/i387.h Sun Dec 16 17:41:45 2001
@@ -12,6 +12,7 @@
#define __ASM_I386_I387_H
#include <linux/sched.h>
+#include <linux/spinlock.h>
#include <asm/processor.h>
#include <asm/sigcontext.h>
#include <asm/user.h>
@@ -24,7 +25,7 @@
extern void restore_fpu( struct task_struct *tsk );
extern void kernel_fpu_begin(void);
-#define kernel_fpu_end() stts()
+#define kernel_fpu_end() do { stts(); preempt_enable(); } while(0)
#define unlazy_fpu( tsk ) do { \
diff -urN linux-2.5.1/include/asm-i386/mmu_context.h linux/include/asm-i386/mmu_context.h
--- linux-2.5.1/include/asm-i386/mmu_context.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/mmu_context.h Sun Dec 16 17:41:45 2001
@@ -27,6 +27,10 @@
static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next, struct task_struct *tsk, unsigned cpu)
{
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disabled() == 0)
+ BUG();
+#endif
if (prev != next) {
/* stop flush ipis for the previous mm */
clear_bit(cpu, &prev->cpu_vm_mask);
diff -urN linux-2.5.1/include/asm-i386/pgalloc.h linux/include/asm-i386/pgalloc.h
--- linux-2.5.1/include/asm-i386/pgalloc.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/pgalloc.h Sun Dec 16 17:41:45 2001
@@ -75,20 +75,26 @@
{
unsigned long *ret;
+ preempt_disable();
if ((ret = pgd_quicklist) != NULL) {
pgd_quicklist = (unsigned long *)(*ret);
ret[0] = 0;
pgtable_cache_size--;
- } else
+ preempt_enable();
+ } else {
+ preempt_enable();
ret = (unsigned long *)get_pgd_slow();
+ }
return (pgd_t *)ret;
}
static inline void free_pgd_fast(pgd_t *pgd)
{
+ preempt_disable();
*(unsigned long *)pgd = (unsigned long) pgd_quicklist;
pgd_quicklist = (unsigned long *) pgd;
pgtable_cache_size++;
+ preempt_enable();
}
static inline void free_pgd_slow(pgd_t *pgd)
@@ -119,19 +125,23 @@
{
unsigned long *ret;
+ preempt_disable();
if ((ret = (unsigned long *)pte_quicklist) != NULL) {
pte_quicklist = (unsigned long *)(*ret);
ret[0] = ret[1];
pgtable_cache_size--;
}
+ preempt_enable();
return (pte_t *)ret;
}
static inline void pte_free_fast(pte_t *pte)
{
+ preempt_disable();
*(unsigned long *)pte = (unsigned long) pte_quicklist;
pte_quicklist = (unsigned long *) pte;
pgtable_cache_size++;
+ preempt_enable();
}
static __inline__ void pte_free_slow(pte_t *pte)
diff -urN linux-2.5.1/include/asm-i386/processor.h linux/include/asm-i386/processor.h
--- linux-2.5.1/include/asm-i386/processor.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/processor.h Sun Dec 16 17:41:45 2001
@@ -502,7 +502,10 @@
{
__asm__ __volatile__ ("prefetchw (%0)" : : "r"(x));
}
-#define spin_lock_prefetch(x) prefetchw(x)
+#define spin_lock_prefetch(x) do { \
+ prefetchw(x); \
+ preempt_prefetch(¤t->preempt_count); \
+} while(0)
#endif
diff -urN linux-2.5.1/include/asm-i386/smplock.h linux/include/asm-i386/smplock.h
--- linux-2.5.1/include/asm-i386/smplock.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/smplock.h Sun Dec 16 17:41:45 2001
@@ -10,7 +10,15 @@
extern spinlock_t kernel_flag;
+#ifdef CONFIG_SMP
#define kernel_locked() spin_is_locked(&kernel_flag)
+#else
+#ifdef CONFIG_PREEMPT
+#define kernel_locked() preempt_is_disabled()
+#else
+#define kernel_locked() 1
+#endif
+#endif
/*
* Release global kernel lock and global interrupt lock
@@ -42,6 +50,11 @@
*/
static __inline__ void lock_kernel(void)
{
+#ifdef CONFIG_PREEMPT
+ if (current->lock_depth == -1)
+ spin_lock(&kernel_flag);
+ ++current->lock_depth;
+#else
#if 1
if (!++current->lock_depth)
spin_lock(&kernel_flag);
@@ -54,6 +67,7 @@
:"=m" (__dummy_lock(&kernel_flag)),
"=m" (current->lock_depth));
#endif
+#endif
}
static __inline__ void unlock_kernel(void)
diff -urN linux-2.5.1/include/asm-i386/softirq.h linux/include/asm-i386/softirq.h
--- linux-2.5.1/include/asm-i386/softirq.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/softirq.h Sun Dec 16 17:41:45 2001
@@ -5,9 +5,9 @@
#include <asm/hardirq.h>
#define __cpu_bh_enable(cpu) \
- do { barrier(); local_bh_count(cpu)--; } while (0)
+ do { barrier(); local_bh_count(cpu)--; preempt_enable(); } while (0)
#define cpu_bh_disable(cpu) \
- do { local_bh_count(cpu)++; barrier(); } while (0)
+ do { preempt_disable(); local_bh_count(cpu)++; barrier(); } while (0)
#define local_bh_disable() cpu_bh_disable(smp_processor_id())
#define __local_bh_enable() __cpu_bh_enable(smp_processor_id())
@@ -22,7 +22,7 @@
* If you change the offsets in irq_stat then you have to
* update this code as well.
*/
-#define local_bh_enable() \
+#define _local_bh_enable() \
do { \
unsigned int *ptr = &local_bh_count(smp_processor_id()); \
\
@@ -45,4 +45,6 @@
/* no registers clobbered */ ); \
} while (0)
+#define local_bh_enable() do { _local_bh_enable(); preempt_enable(); } while (0)
+
#endif /* __ASM_SOFTIRQ_H */
diff -urN linux-2.5.1/include/asm-i386/spinlock.h linux/include/asm-i386/spinlock.h
--- linux-2.5.1/include/asm-i386/spinlock.h Sun Dec 16 17:39:30 2001
+++ linux/include/asm-i386/spinlock.h Sun Dec 16 17:41:45 2001
@@ -77,7 +77,7 @@
:"=m" (lock->lock) : : "memory"
-static inline void spin_unlock(spinlock_t *lock)
+static inline void _raw_spin_unlock(spinlock_t *lock)
{
#if SPINLOCK_DEBUG
if (lock->magic != SPINLOCK_MAGIC)
@@ -97,7 +97,7 @@
:"=q" (oldval), "=m" (lock->lock) \
:"0" (oldval) : "memory"
-static inline void spin_unlock(spinlock_t *lock)
+static inline void _raw_spin_unlock(spinlock_t *lock)
{
char oldval = 1;
#if SPINLOCK_DEBUG
@@ -113,7 +113,7 @@
#endif
-static inline int spin_trylock(spinlock_t *lock)
+static inline int _raw_spin_trylock(spinlock_t *lock)
{
char oldval;
__asm__ __volatile__(
@@ -123,7 +123,7 @@
return oldval > 0;
}
-static inline void spin_lock(spinlock_t *lock)
+static inline void _raw_spin_lock(spinlock_t *lock)
{
#if SPINLOCK_DEBUG
__label__ here;
@@ -179,7 +179,7 @@
*/
/* the spinlock helpers are in arch/i386/kernel/semaphore.c */
-static inline void read_lock(rwlock_t *rw)
+static inline void _raw_read_lock(rwlock_t *rw)
{
#if SPINLOCK_DEBUG
if (rw->magic != RWLOCK_MAGIC)
@@ -188,7 +188,7 @@
__build_read_lock(rw, "__read_lock_failed");
}
-static inline void write_lock(rwlock_t *rw)
+static inline void _raw_write_lock(rwlock_t *rw)
{
#if SPINLOCK_DEBUG
if (rw->magic != RWLOCK_MAGIC)
@@ -197,10 +197,10 @@
__build_write_lock(rw, "__write_lock_failed");
}
-#define read_unlock(rw) asm volatile("lock ; incl %0" :"=m" ((rw)->lock) : : "memory")
-#define write_unlock(rw) asm volatile("lock ; addl $" RW_LOCK_BIAS_STR ",%0":"=m" ((rw)->lock) : : "memory")
+#define _raw_read_unlock(rw) asm volatile("lock ; incl %0" :"=m" ((rw)->lock) : : "memory")
+#define _raw_write_unlock(rw) asm volatile("lock ; addl $" RW_LOCK_BIAS_STR ",%0":"=m" ((rw)->lock) : : "memory")
-static inline int write_trylock(rwlock_t *lock)
+static inline int _raw_write_trylock(rwlock_t *lock)
{
atomic_t *count = (atomic_t *)lock;
if (atomic_sub_and_test(RW_LOCK_BIAS, count))
diff -urN linux-2.5.1/include/asm-sh/hardirq.h linux/include/asm-sh/hardirq.h
--- linux-2.5.1/include/asm-sh/hardirq.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-sh/hardirq.h Sun Dec 16 17:41:45 2001
@@ -34,6 +34,8 @@
#define synchronize_irq() barrier()
+#define release_irqlock(cpu) do { } while (0)
+
#else
#error Super-H SMP is not available
diff -urN linux-2.5.1/include/asm-sh/mmu_context.h linux/include/asm-sh/mmu_context.h
--- linux-2.5.1/include/asm-sh/mmu_context.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-sh/mmu_context.h Sun Dec 16 17:41:45 2001
@@ -166,6 +166,10 @@
struct mm_struct *next,
struct task_struct *tsk, unsigned int cpu)
{
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disabled() == 0)
+ BUG();
+#endif
if (prev != next) {
unsigned long __pgdir = (unsigned long)next->pgd;
diff -urN linux-2.5.1/include/asm-sh/smplock.h linux/include/asm-sh/smplock.h
--- linux-2.5.1/include/asm-sh/smplock.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-sh/smplock.h Sun Dec 16 17:41:45 2001
@@ -9,15 +9,88 @@
#include <linux/config.h>
-#ifndef CONFIG_SMP
-
+#if !defined(CONFIG_SMP) && !defined(CONFIG_PREEMPT)
+/*
+ * Should never happen, since linux/smp_lock.h catches this case;
+ * but in case this file is included directly with neither SMP nor
+ * PREEMPT configuration, provide same dummys as linux/smp_lock.h
+ */
#define lock_kernel() do { } while(0)
#define unlock_kernel() do { } while(0)
-#define release_kernel_lock(task, cpu, depth) ((depth) = 1)
-#define reacquire_kernel_lock(task, cpu, depth) do { } while(0)
+#define release_kernel_lock(task, cpu) do { } while(0)
+#define reacquire_kernel_lock(task) do { } while(0)
+#define kernel_locked() 1
+
+#else /* CONFIG_SMP || CONFIG_PREEMPT */
+
+#if CONFIG_SMP
+#error "We do not support SMP on SH yet"
+#endif
+/*
+ * Default SMP lock implementation (i.e. the i386 version)
+ */
+
+#include <linux/interrupt.h>
+#include <linux/spinlock.h>
+
+extern spinlock_t kernel_flag;
+#define lock_bkl() spin_lock(&kernel_flag)
+#define unlock_bkl() spin_unlock(&kernel_flag)
+#ifdef CONFIG_SMP
+#define kernel_locked() spin_is_locked(&kernel_flag)
+#elif CONFIG_PREEMPT
+#define kernel_locked() preempt_is_disabled()
+#else /* neither */
+#define kernel_locked() 1
+#endif
+
+/*
+ * Release global kernel lock and global interrupt lock
+ */
+#define release_kernel_lock(task, cpu) \
+do { \
+ if (task->lock_depth >= 0) \
+ spin_unlock(&kernel_flag); \
+ release_irqlock(cpu); \
+ __sti(); \
+} while (0)
+
+/*
+ * Re-acquire the kernel lock
+ */
+#define reacquire_kernel_lock(task) \
+do { \
+ if (task->lock_depth >= 0) \
+ spin_lock(&kernel_flag); \
+} while (0)
+
+/*
+ * Getting the big kernel lock.
+ *
+ * This cannot happen asynchronously,
+ * so we only need to worry about other
+ * CPU's.
+ */
+static __inline__ void lock_kernel(void)
+{
+#ifdef CONFIG_PREEMPT
+ if (current->lock_depth == -1)
+ spin_lock(&kernel_flag);
+ ++current->lock_depth;
#else
-#error "We do not support SMP on SH"
-#endif /* CONFIG_SMP */
+ if (!++current->lock_depth)
+ spin_lock(&kernel_flag);
+#endif
+}
+
+static __inline__ void unlock_kernel(void)
+{
+ if (current->lock_depth < 0)
+ BUG();
+ if (--current->lock_depth < 0)
+ spin_unlock(&kernel_flag);
+}
+#endif /* CONFIG_SMP || CONFIG_PREEMPT */
#endif /* __ASM_SH_SMPLOCK_H */
diff -urN linux-2.5.1/include/asm-sh/softirq.h linux/include/asm-sh/softirq.h
--- linux-2.5.1/include/asm-sh/softirq.h Sun Dec 16 17:39:31 2001
+++ linux/include/asm-sh/softirq.h Sun Dec 16 17:41:45 2001
@@ -6,6 +6,7 @@
#define local_bh_disable() \
do { \
+ preempt_disable(); \
local_bh_count(smp_processor_id())++; \
barrier(); \
} while (0)
@@ -14,6 +15,7 @@
do { \
barrier(); \
local_bh_count(smp_processor_id())--; \
+ preempt_enable(); \
} while (0)
#define local_bh_enable() \
@@ -22,6 +24,7 @@
if (!--local_bh_count(smp_processor_id()) \
&& softirq_pending(smp_processor_id())) { \
do_softirq(); \
+ preempt_enable(); \
} \
} while (0)
diff -urN linux-2.5.1/include/linux/brlock.h linux/include/linux/brlock.h
--- linux-2.5.1/include/linux/brlock.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/brlock.h Sun Dec 16 17:41:45 2001
@@ -171,11 +171,11 @@
}
#else
-# define br_read_lock(idx) ((void)(idx))
-# define br_read_unlock(idx) ((void)(idx))
-# define br_write_lock(idx) ((void)(idx))
-# define br_write_unlock(idx) ((void)(idx))
-#endif
+# define br_read_lock(idx) ({ (void)(idx); preempt_disable(); })
+# define br_read_unlock(idx) ({ (void)(idx); preempt_enable(); })
+# define br_write_lock(idx) ({ (void)(idx); preempt_disable(); })
+# define br_write_unlock(idx) ({ (void)(idx); preempt_enable(); })
+#endif /* CONFIG_SMP */
/*
* Now enumerate all of the possible sw/hw IRQ protected
diff -urN linux-2.5.1/include/linux/dcache.h linux/include/linux/dcache.h
--- linux-2.5.1/include/linux/dcache.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/dcache.h Sun Dec 16 17:41:45 2001
@@ -126,31 +126,6 @@
extern spinlock_t dcache_lock;
-/**
- * d_drop - drop a dentry
- * @dentry: dentry to drop
- *
- * d_drop() unhashes the entry from the parent
- * dentry hashes, so that it won't be found through
- * a VFS lookup any more. Note that this is different
- * from deleting the dentry - d_delete will try to
- * mark the dentry negative if possible, giving a
- * successful _negative_ lookup, while d_drop will
- * just make the cache lookup fail.
- *
- * d_drop() is used mainly for stuff that wants
- * to invalidate a dentry for some reason (NFS
- * timeouts or autofs deletes).
- */
-
-static __inline__ void d_drop(struct dentry * dentry)
-{
- spin_lock(&dcache_lock);
- list_del(&dentry->d_hash);
- INIT_LIST_HEAD(&dentry->d_hash);
- spin_unlock(&dcache_lock);
-}
-
static __inline__ int dname_external(struct dentry *d)
{
return d->d_name.name != d->d_iname;
@@ -275,3 +250,34 @@
#endif /* __KERNEL__ */
#endif /* __LINUX_DCACHE_H */
+
+#if !defined(__LINUX_DCACHE_H_INLINES) && defined(_TASK_STRUCT_DEFINED)
+#define __LINUX_DCACHE_H_INLINES
+
+#ifdef __KERNEL__
+/**
+ * d_drop - drop a dentry
+ * @dentry: dentry to drop
+ *
+ * d_drop() unhashes the entry from the parent
+ * dentry hashes, so that it won't be found through
+ * a VFS lookup any more. Note that this is different
+ * from deleting the dentry - d_delete will try to
+ * mark the dentry negative if possible, giving a
+ * successful _negative_ lookup, while d_drop will
+ * just make the cache lookup fail.
+ *
+ * d_drop() is used mainly for stuff that wants
+ * to invalidate a dentry for some reason (NFS
+ * timeouts or autofs deletes).
+ */
+
+static __inline__ void d_drop(struct dentry * dentry)
+{
+ spin_lock(&dcache_lock);
+ list_del(&dentry->d_hash);
+ INIT_LIST_HEAD(&dentry->d_hash);
+ spin_unlock(&dcache_lock);
+}
+#endif
+#endif
diff -urN linux-2.5.1/include/linux/fs_struct.h linux/include/linux/fs_struct.h
--- linux-2.5.1/include/linux/fs_struct.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/fs_struct.h Sun Dec 16 17:41:45 2001
@@ -20,6 +20,15 @@
extern void exit_fs(struct task_struct *);
extern void set_fs_altroot(void);
+struct fs_struct *copy_fs_struct(struct fs_struct *old);
+void put_fs_struct(struct fs_struct *fs);
+
+#endif
+#endif
+
+#if !defined(_LINUX_FS_STRUCT_H_INLINES) && defined(_TASK_STRUCT_DEFINED)
+#define _LINUX_FS_STRUCT_H_INLINES
+#ifdef __KERNEL__
/*
* Replace the fs->{rootmnt,root} with {mnt,dentry}. Put the old values.
* It can block. Requires the big lock held.
@@ -65,9 +74,5 @@
mntput(old_pwdmnt);
}
}
-
-struct fs_struct *copy_fs_struct(struct fs_struct *old);
-void put_fs_struct(struct fs_struct *fs);
-
#endif
#endif
diff -urN linux-2.5.1/include/linux/sched.h linux/include/linux/sched.h
--- linux-2.5.1/include/linux/sched.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/sched.h Sun Dec 16 17:41:45 2001
@@ -87,6 +87,7 @@
#define TASK_UNINTERRUPTIBLE 2
#define TASK_ZOMBIE 4
#define TASK_STOPPED 8
+#define PREEMPT_ACTIVE 0x40000000
#define __set_task_state(tsk, state_value) \
do { (tsk)->state = (state_value); } while (0)
@@ -153,6 +154,9 @@
#define MAX_SCHEDULE_TIMEOUT LONG_MAX
extern signed long FASTCALL(schedule_timeout(signed long timeout));
asmlinkage void schedule(void);
+#ifdef CONFIG_PREEMPT
+asmlinkage void preempt_schedule(void);
+#endif
extern int schedule_task(struct tq_struct *task);
extern void flush_scheduled_tasks(void);
@@ -282,7 +286,17 @@
* offsets of these are hardcoded elsewhere - touch with care
*/
volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
- unsigned long flags; /* per process flags, defined below */
+ /*
+ * We want the preempt_count in this cache line, but we
+ * a) don't want to mess up the offsets in asm code, and
+ * b) the alignment of the next line below,
+ * so we move "flags" down
+ *
+ * Also note we don't make preempt_count volatile, but we do
+ * need to make sure it is never hiding in a register when
+ * we have an interrupt, so we need to use barrier()
+ */
+ int preempt_count; /* 0=> preemptable, < 0 => BUG */
int sigpending;
mm_segment_t addr_limit; /* thread address space:
0-0xBFFFFFFF for user-thead
@@ -324,6 +338,7 @@
struct mm_struct *active_mm;
struct list_head local_pages;
unsigned int allocation_order, nr_local_pages;
+ unsigned long flags;
/* task state */
struct linux_binfmt *binfmt;
@@ -925,6 +940,11 @@
return res;
}
+#define _TASK_STRUCT_DEFINED
+#include <linux/dcache.h>
+#include <linux/tqueue.h>
+#include <linux/fs_struct.h>
+
#endif /* __KERNEL__ */
#endif
diff -urN linux-2.5.1/include/linux/smp.h linux/include/linux/smp.h
--- linux-2.5.1/include/linux/smp.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/smp.h Sun Dec 16 17:41:45 2001
@@ -81,7 +81,9 @@
#define smp_processor_id() 0
#define hard_smp_processor_id() 0
#define smp_threads_ready 1
+#ifndef CONFIG_PREEMPT
#define kernel_lock()
+#endif
#define cpu_logical_map(cpu) 0
#define cpu_number_map(cpu) 0
#define smp_call_function(func,info,retry,wait) ({ 0; })
diff -urN linux-2.5.1/include/linux/smp_lock.h linux/include/linux/smp_lock.h
--- linux-2.5.1/include/linux/smp_lock.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/smp_lock.h Sun Dec 16 17:41:45 2001
@@ -3,7 +3,7 @@
#include <linux/config.h>
-#ifndef CONFIG_SMP
+#if !defined(CONFIG_SMP) && !defined(CONFIG_PREEMPT)
#define lock_kernel() do { } while(0)
#define unlock_kernel() do { } while(0)
diff -urN linux-2.5.1/include/linux/spinlock.h linux/include/linux/spinlock.h
--- linux-2.5.1/include/linux/spinlock.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/spinlock.h Sun Dec 16 17:41:45 2001
@@ -2,6 +2,7 @@
#define __LINUX_SPINLOCK_H
#include <linux/config.h>
+#include <linux/compiler.h>
/*
* These are the generic versions of the spinlocks and read-write
@@ -45,8 +46,10 @@
#if (DEBUG_SPINLOCKS < 1)
+#ifndef CONFIG_PREEMPT
#define atomic_dec_and_lock(atomic,lock) atomic_dec_and_test(atomic)
#define ATOMIC_DEC_AND_LOCK
+#endif
/*
* Your basic spinlocks, allowing only a single CPU anywhere
@@ -62,11 +65,11 @@
#endif
#define spin_lock_init(lock) do { } while(0)
-#define spin_lock(lock) (void)(lock) /* Not "unused variable". */
+#define _raw_spin_lock(lock) (void)(lock) /* Not "unused variable". */
#define spin_is_locked(lock) (0)
-#define spin_trylock(lock) ({1; })
+#define _raw_spin_trylock(lock) ({1; })
#define spin_unlock_wait(lock) do { } while(0)
-#define spin_unlock(lock) do { } while(0)
+#define _raw_spin_unlock(lock) do { } while(0)
#elif (DEBUG_SPINLOCKS < 2)
@@ -125,13 +128,77 @@
#endif
#define rwlock_init(lock) do { } while(0)
-#define read_lock(lock) (void)(lock) /* Not "unused variable". */
-#define read_unlock(lock) do { } while(0)
-#define write_lock(lock) (void)(lock) /* Not "unused variable". */
-#define write_unlock(lock) do { } while(0)
+#define _raw_read_lock(lock) (void)(lock) /* Not "unused variable". */
+#define _raw_read_unlock(lock) do { } while(0)
+#define _raw_write_lock(lock) (void)(lock) /* Not "unused variable". */
+#define _raw_write_unlock(lock) do { } while(0)
#endif /* !SMP */
+#ifdef CONFIG_PREEMPT
+
+#define preempt_is_disabled() (current->preempt_count)
+#define preempt_prefetch(a) prefetchw(a)
+
+#define preempt_disable() \
+do { \
+ ++current->preempt_count; \
+ barrier(); \
+} while (0)
+
+#define preempt_enable_no_resched() \
+do { \
+ --current->preempt_count; \
+ barrier(); \
+} while (0)
+
+#define preempt_enable() \
+do { \
+ --current->preempt_count; \
+ barrier(); \
+ if (unlikely((current->preempt_count == 0) && current->need_resched)) \
+ preempt_schedule(); \
+} while (0)
+
+#define spin_lock(lock) \
+do { \
+ preempt_disable(); \
+ _raw_spin_lock(lock); \
+} while(0)
+#define spin_trylock(lock) ({preempt_disable(); _raw_spin_trylock(lock) ? \
+ 1 : ({preempt_enable(); 0;});})
+#define spin_unlock(lock) \
+do { \
+ _raw_spin_unlock(lock); \
+ preempt_enable(); \
+} while (0)
+
+#define read_lock(lock) ({preempt_disable(); _raw_read_lock(lock);})
+#define read_unlock(lock) ({_raw_read_unlock(lock); preempt_enable();})
+#define write_lock(lock) ({preempt_disable(); _raw_write_lock(lock);})
+#define write_unlock(lock) ({_raw_write_unlock(lock); preempt_enable();})
+#define write_trylock(lock) ({preempt_disable(); _raw_write_trylock(lock) ? \
+ 1 : ({preempt_enable(); 0;});})
+
+#else
+
+#define preempt_is_disabled() do { } while (0)
+#define preempt_disable() do { } while (0)
+#define preempt_enable_no_resched()
+#define preempt_enable() do { } while (0)
+#define preempt_prefetch(a)
+
+#define spin_lock(lock) _raw_spin_lock(lock)
+#define spin_trylock(lock) _raw_spin_trylock(lock)
+#define spin_unlock(lock) _raw_spin_unlock(lock)
+
+#define read_lock(lock) _raw_read_lock(lock)
+#define read_unlock(lock) _raw_read_unlock(lock)
+#define write_lock(lock) _raw_write_lock(lock)
+#define write_unlock(lock) _raw_write_unlock(lock)
+#define write_trylock(lock) _raw_write_trylock(lock)
+#endif
+
/* "lock on reference count zero" */
#ifndef ATOMIC_DEC_AND_LOCK
#include <asm/atomic.h>
diff -urN linux-2.5.1/include/linux/tqueue.h linux/include/linux/tqueue.h
--- linux-2.5.1/include/linux/tqueue.h Sun Dec 16 17:39:29 2001
+++ linux/include/linux/tqueue.h Sun Dec 16 17:41:45 2001
@@ -94,6 +94,22 @@
extern spinlock_t tqueue_lock;
/*
+ * Call all "bottom halfs" on a given list.
+ */
+
+extern void __run_task_queue(task_queue *list);
+
+static inline void run_task_queue(task_queue *list)
+{
+ if (TQ_ACTIVE(*list))
+ __run_task_queue(list);
+}
+
+#endif /* _LINUX_TQUEUE_H */
+
+#if !defined(_LINUX_TQUEUE_H_INLINES) && defined(_TASK_STRUCT_DEFINED)
+#define _LINUX_TQUEUE_H_INLINES
+/*
* Queue a task on a tq. Return non-zero if it was successfully
* added.
*/
@@ -109,17 +125,4 @@
}
return ret;
}
-
-/*
- * Call all "bottom halfs" on a given list.
- */
-
-extern void __run_task_queue(task_queue *list);
-
-static inline void run_task_queue(task_queue *list)
-{
- if (TQ_ACTIVE(*list))
- __run_task_queue(list);
-}
-
-#endif /* _LINUX_TQUEUE_H */
+#endif
diff -urN linux-2.5.1/kernel/exit.c linux/kernel/exit.c
--- linux-2.5.1/kernel/exit.c Sun Dec 16 17:39:29 2001
+++ linux/kernel/exit.c Sun Dec 16 17:41:45 2001
@@ -286,6 +286,10 @@
struct mm_struct * start_lazy_tlb(void)
{
struct mm_struct *mm = current->mm;
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disabled() == 0)
+ BUG();
+#endif
current->mm = NULL;
/* active_mm is still 'mm' */
atomic_inc(&mm->mm_count);
@@ -297,6 +301,10 @@
{
struct mm_struct *active_mm = current->active_mm;
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disabled() == 0)
+ BUG();
+#endif
current->mm = mm;
if (mm != active_mm) {
current->active_mm = mm;
@@ -320,8 +328,8 @@
/* more a memory barrier than a real lock */
task_lock(tsk);
tsk->mm = NULL;
- task_unlock(tsk);
enter_lazy_tlb(mm, current, smp_processor_id());
+ task_unlock(tsk);
mmput(mm);
}
}
diff -urN linux-2.5.1/kernel/fork.c linux/kernel/fork.c
--- linux-2.5.1/kernel/fork.c Sun Dec 16 17:39:29 2001
+++ linux/kernel/fork.c Sun Dec 16 17:41:45 2001
@@ -604,6 +604,12 @@
if (p->binfmt && p->binfmt->module)
__MOD_INC_USE_COUNT(p->binfmt->module);
+#ifdef CONFIG_PREEMPT
+ /* Since we are keeping the context switch off state as part
+ * of the context, make sure we start with it off.
+ */
+ p->preempt_count = 1;
+#endif
p->did_exec = 0;
p->swappable = 0;
p->state = TASK_UNINTERRUPTIBLE;
diff -urN linux-2.5.1/kernel/ksyms.c linux/kernel/ksyms.c
--- linux-2.5.1/kernel/ksyms.c Sun Dec 16 17:39:29 2001
+++ linux/kernel/ksyms.c Sun Dec 16 17:41:45 2001
@@ -436,6 +436,9 @@
EXPORT_SYMBOL(interruptible_sleep_on);
EXPORT_SYMBOL(interruptible_sleep_on_timeout);
EXPORT_SYMBOL(schedule);
+#ifdef CONFIG_PREEMPT
+EXPORT_SYMBOL(preempt_schedule);
+#endif
EXPORT_SYMBOL(schedule_timeout);
EXPORT_SYMBOL(jiffies);
EXPORT_SYMBOL(xtime);
diff -urN linux-2.5.1/kernel/sched.c linux/kernel/sched.c
--- linux-2.5.1/kernel/sched.c Sun Dec 16 17:39:29 2001
+++ linux/kernel/sched.c Sun Dec 16 17:41:45 2001
@@ -476,7 +476,7 @@
task_lock(prev);
task_release_cpu(prev);
mb();
- if (prev->state == TASK_RUNNING)
+ if (task_on_runqueue(prev))
goto needs_resched;
out_unlock:
@@ -506,7 +506,7 @@
goto out_unlock;
spin_lock_irqsave(&runqueue_lock, flags);
- if ((prev->state == TASK_RUNNING) && !task_has_cpu(prev))
+ if (task_on_runqueue(prev) && !task_has_cpu(prev))
reschedule_idle(prev);
spin_unlock_irqrestore(&runqueue_lock, flags);
goto out_unlock;
@@ -519,6 +519,7 @@
asmlinkage void schedule_tail(struct task_struct *prev)
{
__schedule_tail(prev);
+ preempt_enable();
}
/*
@@ -541,6 +542,8 @@
spin_lock_prefetch(&runqueue_lock);
+ preempt_disable();
+
if (!current->active_mm) BUG();
need_resched_back:
prev = current;
@@ -568,6 +571,9 @@
move_last_runqueue(prev);
}
+#ifdef CONFIG_PREEMPT
+ if (preempt_is_disabled() & PREEMPT_ACTIVE) goto treat_like_run;
+#endif
switch (prev->state) {
case TASK_INTERRUPTIBLE:
if (signal_pending(prev)) {
@@ -578,6 +584,9 @@
del_from_runqueue(prev);
case TASK_RUNNING:;
}
+#ifdef CONFIG_PREEMPT
+ treat_like_run:
+#endif
prev->need_resched = 0;
/*
@@ -686,6 +695,7 @@
reacquire_kernel_lock(current);
if (current->need_resched)
goto need_resched_back;
+ preempt_enable_no_resched();
return;
}
@@ -964,6 +974,34 @@
return setscheduler(pid, -1, param);
}
+#ifdef CONFIG_PREEMPT
+
+#ifdef CONFIG_SMP
+#define lock_to_this_cpu() \
+ unsigned long old_cpus_allowed = current->cpus_allowed; \
+ current->cpus_allowed = 1UL << smp_processor_id()
+#define restore_cpus_allowed() current->cpus_allowed = old_cpus_allowed
+#else
+#define lock_to_this_cpu()
+#define restore_cpus_allowed()
+#endif /* !CONFIG_SMP */
+
+asmlinkage void preempt_schedule(void)
+{
+ while (current->need_resched) {
+ /* it would be ideal not to lock tasks to their cpu here,
+ * but only around the data that needs such locking */
+ lock_to_this_cpu();
+ current->preempt_count += PREEMPT_ACTIVE + 1;
+ barrier();
+ schedule();
+ current->preempt_count -= PREEMPT_ACTIVE + 1;
+ barrier();
+ restore_cpus_allowed();
+ }
+}
+#endif /* CONFIG_PREEMPT */
+
asmlinkage long sys_sched_getscheduler(pid_t pid)
{
struct task_struct *p;
diff -urN linux-2.5.1/lib/dec_and_lock.c linux/lib/dec_and_lock.c
--- linux-2.5.1/lib/dec_and_lock.c Sun Dec 16 17:39:29 2001
+++ linux/lib/dec_and_lock.c Sun Dec 16 17:41:45 2001
@@ -1,5 +1,6 @@
#include <linux/module.h>
#include <linux/spinlock.h>
+#include <linux/sched.h>
#include <asm/atomic.h>
/*
diff -urN linux-2.5.1/mm/slab.c linux/mm/slab.c
--- linux-2.5.1/mm/slab.c Sun Dec 16 17:39:29 2001
+++ linux/mm/slab.c Sun Dec 16 17:41:45 2001
@@ -49,7 +49,9 @@
* constructors and destructors are called without any locking.
* Several members in kmem_cache_t and slab_t never change, they
* are accessed without any locking.
- * The per-cpu arrays are never accessed from the wrong cpu, no locking.
+ * The per-cpu arrays are never accessed from the wrong cpu, no locking,
+ * they are however called with local interrupts disabled so no
+ * preempt_disable needed.
* The non-constant members are protected with a per-cache irq spinlock.
*
* Further notes from the original documentation:
diff -urN linux-2.5.1/net/socket.c linux/net/socket.c
--- linux-2.5.1/net/socket.c Sun Dec 16 17:39:38 2001
+++ linux/net/socket.c Sun Dec 16 17:41:45 2001
@@ -135,7 +135,7 @@
static struct net_proto_family *net_families[NPROTO];
-#ifdef CONFIG_SMP
+#if defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)
static atomic_t net_family_lockct = ATOMIC_INIT(0);
static spinlock_t net_family_lock = SPIN_LOCK_UNLOCKED;