[PATCH 2/5] powerpc: Make it possible to move the interrupt handlers away from the kernel
Paul Mackerras
paulus at samba.org
Wed Aug 13 11:29:31 EST 2008
This changes the way that the exception prologs transfer control to
the handlers in 64-bit kernels with the aim of making it possible to
have the prologs separate from the main body of the kernel. Now,
instead of computing the address of the handler by taking the top
32 bits of the paca address (to get the 0xc0000000........ part) and
ORing in something in the bottom 16 bits, we get the base address of
the kernel by doing a load from the paca and add an offset.
This also replaces an mfmsr and an ori to compute the MSR value for
the handler with a load from the paca. That makes it unnecessary to
have a separate version of EXCEPTION_PROLOG_PSERIES that forces 64-bit
mode.
We can no longer use a direct branches in the exception prolog code,
which means that the SLB miss handlers can't branch directly to
.slb_miss_realmode any more. Instead we have to compute the address
and do an indirect branch.
Since the secondary CPUs on pSeries start execution in the first 0x100
bytes of real memory and then have to get to wherever the kernel is,
we can't use a direct branch to get there. Instead this changes
__secondary_hold_spinloop from a flag to a function pointer. When it
is set to a non-NULL value, the secondary CPUs jump to the function
pointed to by that value.
Finally this eliminates one code difference between 32-bit and 64-bit
by making __secondary_hold be the text address of the secondary CPU
spinloop rather than a function descriptor for it.
Signed-off-by: Paul Mackerras <paulus at samba.org>
---
diff --git a/arch/powerpc/include/asm/exception.h b/arch/powerpc/include/asm/exception.h
index 329148b..d3d4534 100644
--- a/arch/powerpc/include/asm/exception.h
+++ b/arch/powerpc/include/asm/exception.h
@@ -53,14 +53,8 @@
* low halfword of the address, but for Kdump we need the whole low
* word.
*/
-#ifdef CONFIG_CRASH_DUMP
#define LOAD_HANDLER(reg, label) \
- oris reg,reg,(label)@h; /* virt addr of handler ... */ \
- ori reg,reg,(label)@l; /* .. and the rest */
-#else
-#define LOAD_HANDLER(reg, label) \
- ori reg,reg,(label)@l; /* virt addr of handler ... */
-#endif
+ addi reg,reg,(label)-_stext; /* virt addr of handler ... */
#define EXCEPTION_PROLOG_1(area) \
mfspr r13,SPRN_SPRG3; /* get paca address into r13 */ \
@@ -72,37 +66,12 @@
std r9,area+EX_R13(r13); \
mfcr r9
-/*
- * Equal to EXCEPTION_PROLOG_PSERIES, except that it forces 64bit mode.
- * The firmware calls the registered system_reset_fwnmi and
- * machine_check_fwnmi handlers in 32bit mode if the cpu happens to run
- * a 32bit application at the time of the event.
- * This firmware bug is present on POWER4 and JS20.
- */
-#define EXCEPTION_PROLOG_PSERIES_FORCE_64BIT(area, label) \
- EXCEPTION_PROLOG_1(area); \
- clrrdi r12,r13,32; /* get high part of &label */ \
- mfmsr r10; \
- /* force 64bit mode */ \
- li r11,5; /* MSR_SF_LG|MSR_ISF_LG */ \
- rldimi r10,r11,61,0; /* insert into top 3 bits */ \
- /* done 64bit mode */ \
- mfspr r11,SPRN_SRR0; /* save SRR0 */ \
- LOAD_HANDLER(r12,label) \
- ori r10,r10,MSR_IR|MSR_DR|MSR_RI; \
- mtspr SPRN_SRR0,r12; \
- mfspr r12,SPRN_SRR1; /* and SRR1 */ \
- mtspr SPRN_SRR1,r10; \
- rfid; \
- b . /* prevent speculative execution */
-
#define EXCEPTION_PROLOG_PSERIES(area, label) \
EXCEPTION_PROLOG_1(area); \
- clrrdi r12,r13,32; /* get high part of &label */ \
- mfmsr r10; \
+ ld r12,PACAKBASE(r13); /* get high part of &label */ \
+ ld r10,PACAKMSR(r13); /* get MSR value for kernel */ \
mfspr r11,SPRN_SRR0; /* save SRR0 */ \
LOAD_HANDLER(r12,label) \
- ori r10,r10,MSR_IR|MSR_DR|MSR_RI; \
mtspr SPRN_SRR0,r12; \
mfspr r12,SPRN_SRR1; /* and SRR1 */ \
mtspr SPRN_SRR1,r10; \
@@ -210,11 +179,10 @@ label##_pSeries: \
std r10,PACA_EXGEN+EX_R13(r13); \
std r11,PACA_EXGEN+EX_R11(r13); \
std r12,PACA_EXGEN+EX_R12(r13); \
- clrrdi r12,r13,32; /* get high part of &label */ \
- mfmsr r10; \
+ ld r12,PACAKBASE(r13); /* get high part of &label */ \
+ ld r10,PACAKMSR(r13); /* get MSR value for kernel */ \
mfspr r11,SPRN_SRR0; /* save SRR0 */ \
LOAD_HANDLER(r12,label##_common) \
- ori r10,r10,MSR_IR|MSR_DR|MSR_RI; \
mtspr SPRN_SRR0,r12; \
mfspr r12,SPRN_SRR1; /* and SRR1 */ \
mtspr SPRN_SRR1,r10; \
diff --git a/arch/powerpc/include/asm/paca.h b/arch/powerpc/include/asm/paca.h
index 6493a39..082b3ae 100644
--- a/arch/powerpc/include/asm/paca.h
+++ b/arch/powerpc/include/asm/paca.h
@@ -62,6 +62,8 @@ struct paca_struct {
u16 paca_index; /* Logical processor number */
u64 kernel_toc; /* Kernel TOC address */
+ u64 kernelbase; /* Base address of kernel */
+ u64 kernel_msr; /* MSR while running in kernel */
u64 stab_real; /* Absolute address of segment table */
u64 stab_addr; /* Virtual address of segment table */
void *emergency_sp; /* pointer to emergency stack */
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 92768d3..e9c4044 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -122,6 +122,8 @@ int main(void)
DEFINE(PACASTABRR, offsetof(struct paca_struct, stab_rr));
DEFINE(PACAR1, offsetof(struct paca_struct, saved_r1));
DEFINE(PACATOC, offsetof(struct paca_struct, kernel_toc));
+ DEFINE(PACAKBASE, offsetof(struct paca_struct, kernelbase));
+ DEFINE(PACAKMSR, offsetof(struct paca_struct, kernel_msr));
DEFINE(PACASOFTIRQEN, offsetof(struct paca_struct, soft_enabled));
DEFINE(PACAHARDIRQEN, offsetof(struct paca_struct, hard_enabled));
DEFINE(PACASLBCACHE, offsetof(struct paca_struct, slb_cache));
diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S
index 27935d1..229ccd1 100644
--- a/arch/powerpc/kernel/head_64.S
+++ b/arch/powerpc/kernel/head_64.S
@@ -82,7 +82,11 @@ END_FTR_SECTION(0, 1)
/* Catch branch to 0 in real mode */
trap
- /* Secondary processors spin on this value until it goes to 1. */
+ /* Secondary processors spin on this value until it becomes nonzero.
+ * When it does it contains the real address of the descriptor
+ * of the function that the cpu should jump to to continue
+ * initialization.
+ */
.globl __secondary_hold_spinloop
__secondary_hold_spinloop:
.llong 0x0
@@ -109,8 +113,11 @@ __secondary_hold_acknowledge:
* before the bulk of the kernel has been relocated. This code
* is relocated to physical address 0x60 before prom_init is run.
* All of it must fit below the first exception vector at 0x100.
+ * Use .globl here not _GLOBAL because we want __secondary_hold
+ * to be the actual text address, not a descriptor.
*/
-_GLOBAL(__secondary_hold)
+ .globl __secondary_hold
+__secondary_hold:
mfmsr r24
ori r24,r24,MSR_RI
mtmsrd r24 /* RI on */
@@ -126,11 +133,11 @@ _GLOBAL(__secondary_hold)
/* All secondary cpus wait here until told to start. */
100: ld r4,__secondary_hold_spinloop at l(0)
- cmpdi 0,r4,1
- bne 100b
+ cmpdi 0,r4,0
+ beq 100b
#if defined(CONFIG_SMP) || defined(CONFIG_KEXEC)
- LOAD_REG_IMMEDIATE(r4, .generic_secondary_smp_init)
+ ld r4,0(r4) /* deref function descriptor */
mtctr r4
mr r3,r24
bctr
@@ -147,6 +154,10 @@ exception_marker:
/*
* This is the start of the interrupt handlers for pSeries
* This code runs with relocation off.
+ * Code from here to __end_interrupts gets copied down to real
+ * address 0x100 when we are running a relocatable kernel.
+ * Therefore any relative branches in this section must only
+ * branch to labels in this section.
*/
. = 0x100
.globl __start_interrupts
@@ -200,7 +211,16 @@ data_access_slb_pSeries:
mfspr r10,SPRN_SPRG1
std r10,PACA_EXSLB+EX_R13(r13)
mfspr r12,SPRN_SRR1 /* and SRR1 */
- b .slb_miss_realmode /* Rel. branch works in real mode */
+ /*
+ * We can't just use a direct branch to .slb_miss_realmode
+ * because the distance from here to there depends on where
+ * the kernel ends up being put.
+ */
+ mfctr r11
+ ld r10,PACAKBASE(r13)
+ LOAD_HANDLER(r10, .slb_miss_realmode)
+ mtctr r10
+ bctr
STD_EXCEPTION_PSERIES(0x400, instruction_access)
@@ -225,7 +245,12 @@ instruction_access_slb_pSeries:
mfspr r10,SPRN_SPRG1
std r10,PACA_EXSLB+EX_R13(r13)
mfspr r12,SPRN_SRR1 /* and SRR1 */
- b .slb_miss_realmode /* Rel. branch works in real mode */
+ mfctr r11
+ ld r10,PACAKBASE(r13)
+ LOAD_HANDLER(r10, .slb_miss_realmode)
+ mtctr r10
+ bctr
+
MASKABLE_EXCEPTION_PSERIES(0x500, hardware_interrupt)
STD_EXCEPTION_PSERIES(0x600, alignment)
@@ -244,14 +269,12 @@ BEGIN_FTR_SECTION
beq- 1f
END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE)
mr r9,r13
- mfmsr r10
mfspr r13,SPRN_SPRG3
mfspr r11,SPRN_SRR0
- clrrdi r12,r13,32
- oris r12,r12,system_call_common at h
- ori r12,r12,system_call_common at l
+ ld r12,PACAKBASE(r13)
+ ld r10,PACAKMSR(r13)
+ LOAD_HANDLER(r12, system_call_entry)
mtspr SPRN_SRR0,r12
- ori r10,r10,MSR_IR|MSR_DR|MSR_RI
mfspr r12,SPRN_SRR1
mtspr SPRN_SRR1,r10
rfid
@@ -379,7 +402,12 @@ __end_interrupts:
/*
* Code from here down to __end_handlers is invoked from the
- * exception prologs above.
+ * exception prologs above. Because the prologs assemble the
+ * addresses of these handlers using the LOAD_HANDLER macro,
+ * which uses an addi instruction, these handlers must be in
+ * the first 32k of the kernel image. (This restriction could
+ * be lifted if we used an addis + addi in LOAD_HANDLER instead
+ * of just an addi.)
*/
/*** Common interrupt handlers ***/
@@ -419,6 +447,10 @@ machine_check_common:
STD_EXCEPTION_COMMON(0x1800, cbe_thermal, .cbe_thermal_exception)
#endif /* CONFIG_CBE_RAS */
+ .align 7
+system_call_entry:
+ b system_call_common
+
/*
* Here we have detected that the kernel stack pointer is bad.
* R9 contains the saved CR, r13 points to the paca,
@@ -562,6 +594,7 @@ unrecov_user_slb:
*/
_GLOBAL(slb_miss_realmode)
mflr r10
+ mtctr r11
stw r9,PACA_EXSLB+EX_CCR(r13) /* save CR in exc. frame */
std r10,PACA_EXSLB+EX_LR(r13) /* save LR */
@@ -612,11 +645,10 @@ BEGIN_FW_FTR_SECTION
END_FW_FTR_SECTION_IFSET(FW_FEATURE_ISERIES)
#endif /* CONFIG_PPC_ISERIES */
mfspr r11,SPRN_SRR0
- clrrdi r10,r13,32
+ ld r10,PACAKBASE(r13)
LOAD_HANDLER(r10,unrecov_slb)
mtspr SPRN_SRR0,r10
- mfmsr r10
- ori r10,r10,MSR_IR|MSR_DR|MSR_RI
+ ld r10,PACAKMSR(r13)
mtspr SPRN_SRR1,r10
rfid
b .
diff --git a/arch/powerpc/kernel/paca.c b/arch/powerpc/kernel/paca.c
index c9bf17e..623e8c3 100644
--- a/arch/powerpc/kernel/paca.c
+++ b/arch/powerpc/kernel/paca.c
@@ -79,6 +79,8 @@ void __init initialise_pacas(void)
new_paca->lock_token = 0x8000;
new_paca->paca_index = cpu;
new_paca->kernel_toc = kernel_toc;
+ new_paca->kernelbase = KERNELBASE;
+ new_paca->kernel_msr = MSR_KERNEL;
new_paca->hw_cpu_id = 0xffff;
new_paca->slb_shadow_ptr = &slb_shadow[cpu];
new_paca->__current = &init_task;
diff --git a/arch/powerpc/kernel/prom_init.c b/arch/powerpc/kernel/prom_init.c
index b72849a..1f89885 100644
--- a/arch/powerpc/kernel/prom_init.c
+++ b/arch/powerpc/kernel/prom_init.c
@@ -1321,7 +1321,7 @@ static void __init prom_initialize_tce_table(void)
*
* -- Cort
*/
-extern void __secondary_hold(void);
+extern char __secondary_hold;
extern unsigned long __secondary_hold_spinloop;
extern unsigned long __secondary_hold_acknowledge;
@@ -1342,13 +1342,7 @@ static void __init prom_hold_cpus(void)
= (void *) LOW_ADDR(__secondary_hold_spinloop);
unsigned long *acknowledge
= (void *) LOW_ADDR(__secondary_hold_acknowledge);
-#ifdef CONFIG_PPC64
- /* __secondary_hold is actually a descriptor, not the text address */
- unsigned long secondary_hold
- = __pa(*PTRRELOC((unsigned long *)__secondary_hold));
-#else
unsigned long secondary_hold = LOW_ADDR(__secondary_hold);
-#endif
prom_debug("prom_hold_cpus: start...\n");
prom_debug(" 1) spinloop = 0x%x\n", (unsigned long)spinloop);
diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c
index 8b25f51..843c0af 100644
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -255,9 +255,11 @@ void early_setup_secondary(void)
#endif /* CONFIG_SMP */
#if defined(CONFIG_SMP) || defined(CONFIG_KEXEC)
+extern unsigned long __secondary_hold_spinloop;
+extern void generic_secondary_smp_init(void);
+
void smp_release_cpus(void)
{
- extern unsigned long __secondary_hold_spinloop;
unsigned long *ptr;
DBG(" -> smp_release_cpus()\n");
@@ -266,12 +268,11 @@ void smp_release_cpus(void)
* all now so they can start to spin on their individual paca
* spinloops. For non SMP kernels, the secondary cpus never get out
* of the common spinloop.
- * This is useless but harmless on iSeries, secondaries are already
- * waiting on their paca spinloops. */
+ */
ptr = (unsigned long *)((unsigned long)&__secondary_hold_spinloop
- PHYSICAL_START);
- *ptr = 1;
+ *ptr = __pa(generic_secondary_smp_init);
mb();
DBG(" <- smp_release_cpus()\n");
More information about the Linuxppc-dev
mailing list