[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <45F70E25.7050704@goop.org>
Date: Tue, 13 Mar 2007 13:48:37 -0700
From: Jeremy Fitzhardinge <jeremy@...p.org>
To: Rusty Russell <rusty@...tcorp.com.au>
CC: lkml - Kernel Mailing List <linux-kernel@...r.kernel.org>,
Zachary Amsden <zach@...are.com>,
Jeremy Fitzhardinge <jeremy@...source.com>,
Ingo Molnar <mingo@...e.hu>,
Andrew Morton <akpm@...ux-foundation.org>,
Andi Kleen <ak@...e.de>
Subject: Re: [PATCH 0/8] x86 boot, pda and gdt cleanups
Rusty Russell wrote:
> Hi all,
>
> The GDT stuff on x86 is a little more complex than it need be, but
> playing with boot code is always dangerous. These compile and boot on
> UP and SMP for me, but Andrew should let the cook in -mm for a while.
>
Hi Rusty,
This is my rough hacking patch I needed to get things into a Xen-shape
state.
There are a few of things here:
* init_gdt should always use write_gdt_entry when touching the gdt;
if it doesn't and it ends up touching an already-installed gdt
under Xen, it will get a write fault. This happens because
init_gdt ends up getting called twice in SMP (see below).
* init_gdt should always be called before bringing up the cpu,
rather than by the cpu itself (and therefore, cpu_init() shouldn't
call it). Obviously the the boot cpu is an exception.
* secondary_cpu_init stops being necessary.
* On SMP, init_gdt can get called twice: first time in
smp_prepare_boot_cpu, and a second time in trap_init. On UP,
trap_init is the only caller.
diff -r 8dcd1dc9b298 arch/i386/kernel/cpu/common.c
--- a/arch/i386/kernel/cpu/common.c Tue Mar 13 00:33:37 2007 -0700
+++ b/arch/i386/kernel/cpu/common.c Tue Mar 13 13:13:08 2007 -0700
@@ -620,14 +620,24 @@ __cpuinit void init_gdt(int cpu, struct
__cpuinit void init_gdt(int cpu, struct task_struct *idle)
{
struct desc_struct *gdt = per_cpu(gdt_page, cpu).gdt;
+ int i;
/* Based on boot_gdt_table: set percpu so it can be used immediately */
- memcpy(gdt, boot_gdt_table, GDT_SIZE);
+ for(i = 0; i < GDT_ENTRIES; i++) {
+ const struct desc_struct *bgdtt = &boot_gdt_table[i];
+ write_gdt_entry(gdt, i, bgdtt->a, bgdtt->b);
+ }
+
#ifdef CONFIG_SMP
- pack_descriptor((u32 *)&gdt[GDT_ENTRY_PERCPU].a,
- (u32 *)&gdt[GDT_ENTRY_PERCPU].b,
- __per_cpu_offset[cpu], 0xFFFFF,
- 0x80 | DESCTYPE_S | 0x2, 0); /* present read-write data segment */
+ {
+ u32 a, b;
+
+ /* present read-write data segment, pagesize granularity */
+ pack_descriptor(&a, &b, __per_cpu_offset[cpu], 0xFFFFF,
+ 0x80 | DESCTYPE_S | 0x2, 0x8);
+ write_gdt_entry(gdt, GDT_ENTRY_PERCPU, a, b);
+ }
+
per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu];
per_cpu(cpu_number, cpu) = cpu;
#endif /* SMP*/
@@ -709,30 +719,21 @@ static void __cpuinit _cpu_init(int cpu,
mxcsr_feature_mask_init();
}
-/* Entrypoint to initialize secondary CPU */
-void __cpuinit secondary_cpu_init(void)
-{
- int cpu = smp_processor_id();
- struct task_struct *curr = current;
-
- _cpu_init(cpu, curr);
-}
-
/*
* cpu_init() initializes state that is per-CPU. Some data is already
* initialized (naturally) in the bootstrap process, such as the GDT
* and IDT. We reload them nevertheless, this function acts as a
* 'CPU state barrier', nothing should get across.
+ *
+ * Note: we expect the gdt itself to be already set up for this
+ * processor; we just point the cpu at it.
*/
void __cpuinit cpu_init(void)
{
int cpu = smp_processor_id();
- struct task_struct *curr = current;
-
- /* Set up the real GDT so we can transition from the boot_gdt_table. */
- init_gdt(cpu, curr);
+
cpu_set_gdt(cpu);
- _cpu_init(cpu, curr);
+ _cpu_init(cpu, current);
}
#ifdef CONFIG_HOTPLUG_CPU
diff -r 8dcd1dc9b298 arch/i386/kernel/smpboot.c
--- a/arch/i386/kernel/smpboot.c Tue Mar 13 00:33:37 2007 -0700
+++ b/arch/i386/kernel/smpboot.c Tue Mar 13 13:13:08 2007 -0700
@@ -381,14 +381,14 @@ static void __cpuinit start_secondary(vo
static void __cpuinit start_secondary(void *unused)
{
/*
- * Don't put *anything* before secondary_cpu_init(), SMP
- * booting is too fragile that we want to limit the
- * things done here to the most necessary things.
+ * Don't put *anything* before cpu_init(), SMP booting is too
+ * fragile that we want to limit the things done here to the
+ * most necessary things.
*/
#ifdef CONFIG_VMI
vmi_bringup();
#endif
- secondary_cpu_init();
+ cpu_init();
preempt_disable();
smp_callin();
while (!cpu_isset(smp_processor_id(), smp_commenced_mask))
@@ -1165,15 +1165,17 @@ void __init native_smp_prepare_cpus(unsi
void __devinit native_smp_prepare_boot_cpu(void)
{
- cpu_set(smp_processor_id(), cpu_online_map);
- cpu_set(smp_processor_id(), cpu_callout_map);
- cpu_set(smp_processor_id(), cpu_present_map);
- cpu_set(smp_processor_id(), cpu_possible_map);
- per_cpu(cpu_state, smp_processor_id()) = CPU_ONLINE;
+ int cpu = smp_processor_id();
+
+ cpu_set(cpu, cpu_online_map);
+ cpu_set(cpu, cpu_callout_map);
+ cpu_set(cpu, cpu_present_map);
+ cpu_set(cpu, cpu_possible_map);
+ per_cpu(cpu_state, cpu) = CPU_ONLINE;
/* Set up %fs to point to our per-CPU area now it's allocated */
- init_gdt(smp_processor_id(), &init_task);
- cpu_set_gdt(smp_processor_id());
+ init_gdt(cpu, &init_task);
+ cpu_set_gdt(cpu);
}
#ifdef CONFIG_HOTPLUG_CPU
diff -r 8dcd1dc9b298 arch/i386/kernel/traps.c
--- a/arch/i386/kernel/traps.c Tue Mar 13 00:33:37 2007 -0700
+++ b/arch/i386/kernel/traps.c Tue Mar 13 13:13:08 2007 -0700
@@ -1181,6 +1181,7 @@ void __init trap_init(void)
/*
* Should be a barrier for any external CPU state.
*/
+ init_gdt(smp_processor_id(), current);
cpu_init();
trap_init_hook();
diff -r 8dcd1dc9b298 include/asm-i386/processor.h
--- a/include/asm-i386/processor.h Tue Mar 13 00:33:37 2007 -0700
+++ b/include/asm-i386/processor.h Tue Mar 13 13:13:08 2007 -0700
@@ -744,6 +744,6 @@ extern int sysenter_setup(void);
extern void init_gdt(int cpu, struct task_struct *idle);
extern void cpu_set_gdt(int);
-extern void secondary_cpu_init(void);
+extern void cpu_init(void);
#endif /* __ASM_I386_PROCESSOR_H */
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists