Commit 7f613c7d authored by Keith Owens's avatar Keith Owens Committed by Tony Luck
Browse files

[PATCH] MCA/INIT: use per cpu stacks 



The bulk of the change.  Use per cpu MCA/INIT stacks.  Change the SAL
to OS state (sos) to be per process.  Do all the assembler work on the
MCA/INIT stacks, leaving the original stack alone.  Pass per cpu state
data to the C handlers for MCA and INIT, which also means changing the
mca_drv interfaces slightly.  Lots of verification on whether the
original stack is usable before converting it to a sleeping process.

Signed-off-by: default avatarKeith Owens <kaos@sgi.com>
Signed-off-by: default avatarTony Luck <tony.luck@intel.com>
parent 289d773e

arch/ia64/kernel/asm-offsets.c

+32 −8
Original line number Diff line number Diff line
@@ -211,17 +211,41 @@ void foo(void) 
#endif



	BLANK();

	DEFINE(IA64_MCA_CPU_PROC_STATE_DUMP_OFFSET,

	       offsetof (struct ia64_mca_cpu, proc_state_dump));

	DEFINE(IA64_MCA_CPU_STACK_OFFSET,

	       offsetof (struct ia64_mca_cpu, stack));

	DEFINE(IA64_MCA_CPU_STACKFRAME_OFFSET,

	       offsetof (struct ia64_mca_cpu, stackframe));

	DEFINE(IA64_MCA_CPU_RBSTORE_OFFSET,

	       offsetof (struct ia64_mca_cpu, rbstore));

	DEFINE(IA64_MCA_CPU_MCA_STACK_OFFSET,

	       offsetof (struct ia64_mca_cpu, mca_stack));

	DEFINE(IA64_MCA_CPU_INIT_STACK_OFFSET,

	       offsetof (struct ia64_mca_cpu, init_stack));

	BLANK();

	DEFINE(IA64_SAL_OS_STATE_COMMON_OFFSET,

	       offsetof (struct ia64_sal_os_state, sal_ra));

	DEFINE(IA64_SAL_OS_STATE_OS_GP_OFFSET,

	       offsetof (struct ia64_sal_os_state, os_gp));

	DEFINE(IA64_SAL_OS_STATE_PAL_MIN_STATE_OFFSET,

	       offsetof (struct ia64_sal_os_state, pal_min_state));

	DEFINE(IA64_SAL_OS_STATE_PROC_STATE_PARAM_OFFSET,

	       offsetof (struct ia64_sal_os_state, proc_state_param));

	DEFINE(IA64_SAL_OS_STATE_SIZE,

	       sizeof (struct ia64_sal_os_state));

	DEFINE(IA64_PMSA_GR_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_gr));

	DEFINE(IA64_PMSA_BANK1_GR_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_bank1_gr));

	DEFINE(IA64_PMSA_PR_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_pr));

	DEFINE(IA64_PMSA_BR0_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_br0));

	DEFINE(IA64_PMSA_RSC_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_rsc));

	DEFINE(IA64_PMSA_IIP_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_iip));

	DEFINE(IA64_PMSA_IPSR_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_ipsr));

	DEFINE(IA64_PMSA_IFS_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_ifs));

	DEFINE(IA64_PMSA_XIP_OFFSET,

	       offsetof (struct pal_min_state_area_s, pmsa_xip));

	BLANK();



	/* used by fsys_gettimeofday in arch/ia64/kernel/fsys.S */

	DEFINE(IA64_TIME_INTERPOLATOR_ADDRESS_OFFSET, offsetof (struct time_interpolator, addr));

	DEFINE(IA64_TIME_INTERPOLATOR_SOURCE_OFFSET, offsetof (struct time_interpolator, source));

arch/ia64/kernel/mca.c

+505 −316

File changed.

Preview size limit exceeded, changes collapsed.

arch/ia64/kernel/mca_asm.S

+718 −640

File changed.

Preview size limit exceeded, changes collapsed.

arch/ia64/kernel/mca_drv.c

+17 −20
Original line number Diff line number Diff line
@@ -4,6 +4,8 @@ 
 *

 * Copyright (C) 2004 FUJITSU LIMITED

 * Copyright (C) Hidetoshi Seto (seto.hidetoshi@jp.fujitsu.com)

 * Copyright (C) 2005 Silicon Graphics, Inc

 * Copyright (C) 2005 Keith Owens <kaos@sgi.com>

 */

#include <linux/config.h>

#include <linux/types.h>
@@ -38,10 +40,6 @@ 
/* max size of SAL error record (default) */

static int sal_rec_max = 10000;



/* from mca.c */

static ia64_mca_sal_to_os_state_t *sal_to_os_handoff_state;

static ia64_mca_os_to_sal_state_t *os_to_sal_handoff_state;



/* from mca_drv_asm.S */

extern void *mca_handler_bhhook(void);
@@ -316,7 +314,8 @@ init_record_index_pools(void) 
 */



static mca_type_t

is_mca_global(peidx_table_t *peidx, pal_bus_check_info_t *pbci)

is_mca_global(peidx_table_t *peidx, pal_bus_check_info_t *pbci,

	      struct ia64_sal_os_state *sos)

{

	pal_processor_state_info_t *psp = (pal_processor_state_info_t*)peidx_psp(peidx);
@@ -327,7 +326,7 @@ is_mca_global(peidx_table_t *peidx, pal_bus_check_info_t *pbci) 
	 * Therefore it is local MCA when rendezvous has not been requested.

	 * Failed to rendezvous, the system must be down.

	 */

	switch (sal_to_os_handoff_state->imsto_rendez_state) {

	switch (sos->rv_rc) {

		case -1: /* SAL rendezvous unsuccessful */

			return MCA_IS_GLOBAL;

		case  0: /* SAL rendezvous not required */
@@ -388,7 +387,8 @@ is_mca_global(peidx_table_t *peidx, pal_bus_check_info_t *pbci) 
 */



static int

recover_from_read_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci)

recover_from_read_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci,

			struct ia64_sal_os_state *sos)

{

	sal_log_mod_error_info_t *smei;

	pal_min_state_area_t *pmsa;
@@ -426,7 +426,7 @@ recover_from_read_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_chec 
			 *  setup for resume to bottom half of MCA,

			 * "mca_handler_bhhook"

			 */

			pmsa = (pal_min_state_area_t *)(sal_to_os_handoff_state->pal_min_state | (6ul<<61));

			pmsa = sos->pal_min_state;

			/* pass to bhhook as 1st argument (gr8) */

			pmsa->pmsa_gr[8-1] = smei->target_identifier;

			/* set interrupted return address (but no use) */
@@ -459,7 +459,8 @@ recover_from_read_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_chec 
 */



static int

recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci)

recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci,

			    struct ia64_sal_os_state *sos)

{

	int status = 0;

	pal_processor_state_info_t *psp = (pal_processor_state_info_t*)peidx_psp(peidx);
@@ -469,7 +470,7 @@ recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_ 
		case 1: /* partial read */

		case 3: /* full line(cpu) read */

		case 9: /* I/O space read */

			status = recover_from_read_error(slidx, peidx, pbci);

			status = recover_from_read_error(slidx, peidx, pbci, sos);

			break;

		case 0: /* unknown */

		case 2: /* partial write */
@@ -508,7 +509,8 @@ recover_from_platform_error(slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_ 
 */



static int

recover_from_processor_error(int platform, slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci)

recover_from_processor_error(int platform, slidx_table_t *slidx, peidx_table_t *peidx, pal_bus_check_info_t *pbci,

			     struct ia64_sal_os_state *sos)

{

	pal_processor_state_info_t *psp = (pal_processor_state_info_t*)peidx_psp(peidx);
@@ -545,7 +547,7 @@ recover_from_processor_error(int platform, slidx_table_t *slidx, peidx_table_t * 
	 * This means "there are some platform errors".

	 */

	if (platform) 

		return recover_from_platform_error(slidx, peidx, pbci);

		return recover_from_platform_error(slidx, peidx, pbci, sos);

	/* 

	 * On account of strange SAL error record, we cannot recover. 

	 */
@@ -562,8 +564,7 @@ recover_from_processor_error(int platform, slidx_table_t *slidx, peidx_table_t * 


static int

mca_try_to_recover(void *rec, 

	ia64_mca_sal_to_os_state_t *sal_to_os_state,

	ia64_mca_os_to_sal_state_t *os_to_sal_state)

	struct ia64_sal_os_state *sos)

{

	int platform_err;

	int n_proc_err;
@@ -571,10 +572,6 @@ mca_try_to_recover(void *rec, 
	peidx_table_t peidx;

	pal_bus_check_info_t pbci;



	/* handoff state from/to mca.c */

	sal_to_os_handoff_state = sal_to_os_state;

	os_to_sal_handoff_state = os_to_sal_state;



	/* Make index of SAL error record */

	platform_err = mca_make_slidx(rec, &slidx);
@@ -597,11 +594,11 @@ mca_try_to_recover(void *rec, 
	*((u64*)&pbci) = peidx_check_info(&peidx, bus_check, 0);



	/* Check whether MCA is global or not */

	if (is_mca_global(&peidx, &pbci))

	if (is_mca_global(&peidx, &pbci, sos))

		return 0;

	

	/* Try to recover a processor error */

	return recover_from_processor_error(platform_err, &slidx, &peidx, &pbci);

	return recover_from_processor_error(platform_err, &slidx, &peidx, &pbci, sos);

}



/*

include/asm-ia64/mca.h

+64 −38
Original line number Diff line number Diff line
@@ -11,8 +11,6 @@ 
#ifndef _ASM_IA64_MCA_H

#define _ASM_IA64_MCA_H



#define IA64_MCA_STACK_SIZE	8192



#if !defined(__ASSEMBLY__)



#include <linux/interrupt.h>
@@ -48,7 +46,8 @@ typedef union cmcv_reg_u { 


enum {

	IA64_MCA_RENDEZ_CHECKIN_NOTDONE	=	0x0,

	IA64_MCA_RENDEZ_CHECKIN_DONE	=	0x1

	IA64_MCA_RENDEZ_CHECKIN_DONE	=	0x1,

	IA64_MCA_RENDEZ_CHECKIN_INIT	=	0x2,

};



/* Information maintained by the MC infrastructure */
@@ -63,18 +62,42 @@ typedef struct ia64_mc_info_s { 


} ia64_mc_info_t;



typedef struct ia64_mca_sal_to_os_state_s {

	u64		imsto_os_gp;		/* GP of the os registered with the SAL */

	u64		imsto_pal_proc;		/* PAL_PROC entry point - physical addr */

	u64		imsto_sal_proc;		/* SAL_PROC entry point - physical addr */

	u64		imsto_sal_gp;		/* GP of the SAL - physical */

	u64		imsto_rendez_state;	/* Rendez state information */

	u64		imsto_sal_check_ra;	/* Return address in SAL_CHECK while going

						 * back to SAL from OS after MCA handling.

/* Handover state from SAL to OS and vice versa, for both MCA and INIT events.

 * Besides the handover state, it also contains some saved registers from the

 * time of the event.

 * Note: mca_asm.S depends on the precise layout of this structure.

 */



struct ia64_sal_os_state {

	/* SAL to OS, must be at offset 0 */

	u64			os_gp;			/* GP of the os registered with the SAL, physical */

	u64			pal_proc;		/* PAL_PROC entry point, physical */

	u64			sal_proc;		/* SAL_PROC entry point, physical */

	u64			rv_rc;			/* MCA - Rendezvous state, INIT - reason code */

	u64			proc_state_param;	/* from R18 */

	u64			monarch;		/* 1 for a monarch event, 0 for a slave */

	/* common, must follow SAL to OS */

	u64			sal_ra;			/* Return address in SAL, physical */

	u64			sal_gp;			/* GP of the SAL - physical */

	pal_min_state_area_t	*pal_min_state;		/* from R17.  physical in asm, virtual in C */

	u64			prev_IA64_KR_CURRENT;	/* previous value of IA64_KR(CURRENT) */

	struct task_struct	*prev_task;		/* previous task, NULL if it is not useful */

	/* Some interrupt registers are not saved in minstate, pt_regs or

	 * switch_stack.  Because MCA/INIT can occur when interrupts are

	 * disabled, we need to save the additional interrupt registers over

	 * MCA/INIT and resume.

	 */

	u64		pal_min_state;		/* from PAL in r17 */

	u64		proc_state_param;	/* from PAL in r18. See SDV 2:268 11.3.2.1 */

} ia64_mca_sal_to_os_state_t;

	u64			isr;

	u64			ifa;

	u64			itir;

	u64			iipa;

	u64			iim;

	u64			iha;

	/* OS to SAL, must follow common */

	u64			os_status;		/* OS status to SAL, enum below */

	u64			context;		/* 0 if return to same context

							   1 if return to new context */

};



enum {

	IA64_MCA_CORRECTED	=	0x0,	/* Error has been corrected by OS_MCA */
@@ -83,36 +106,22 @@ enum { 
	IA64_MCA_HALT		=	-3	/* System to be halted by SAL */

};



enum {

	IA64_INIT_RESUME	=	0x0,	/* Resume after return from INIT */

	IA64_INIT_WARM_BOOT	=	-1,	/* Warm boot of the system need from SAL */

};



enum {

	IA64_MCA_SAME_CONTEXT	=	0x0,	/* SAL to return to same context */

	IA64_MCA_NEW_CONTEXT	=	-1	/* SAL to return to new context */

};



typedef struct ia64_mca_os_to_sal_state_s {

	u64		imots_os_status;	/*   OS status to SAL as to what happened

						 *   with the MCA handling.

						 */

	u64		imots_sal_gp;		/* GP of the SAL - physical */

	u64		imots_context;		/* 0 if return to same context

						   1 if return to new context */

	u64		*imots_new_min_state;	/* Pointer to structure containing

						 * new values of registers in the min state

						 * save area.

						 */

	u64		imots_sal_check_ra;	/* Return address in SAL_CHECK while going

						 * back to SAL from OS after MCA handling.

						 */

} ia64_mca_os_to_sal_state_t;



/* Per-CPU MCA state that is too big for normal per-CPU variables.  */



struct ia64_mca_cpu {

	u64 stack[IA64_MCA_STACK_SIZE/8];	/* MCA memory-stack */

	u64 proc_state_dump[512];

	u64 stackframe[32];

	u64 rbstore[IA64_MCA_STACK_SIZE/8];	/* MCA reg.-backing store */

	u64 mca_stack[KERNEL_STACK_SIZE/8];

	u64 init_stack[KERNEL_STACK_SIZE/8];

} __attribute__ ((aligned(16)));

};



/* Array of physical addresses of each CPU's MCA area.  */

extern unsigned long __per_cpu_mca[NR_CPUS];
@@ -121,12 +130,29 @@ extern void ia64_mca_init(void); 
extern void ia64_mca_cpu_init(void *);

extern void ia64_os_mca_dispatch(void);

extern void ia64_os_mca_dispatch_end(void);

extern void ia64_mca_ucmc_handler(void);

extern void ia64_mca_ucmc_handler(struct pt_regs *, struct ia64_sal_os_state *);

extern void ia64_init_handler(struct pt_regs *,

			      struct switch_stack *,

			      struct ia64_sal_os_state *);

extern void ia64_monarch_init_handler(void);

extern void ia64_slave_init_handler(void);

extern void ia64_mca_cmc_vector_setup(void);

extern int  ia64_reg_MCA_extension(void*);

extern int  ia64_reg_MCA_extension(int (*fn)(void *, struct ia64_sal_os_state *));

extern void ia64_unreg_MCA_extension(void);

extern u64 ia64_get_rnat(u64 *);



#else	/* __ASSEMBLY__ */



#define IA64_MCA_CORRECTED	0x0	/* Error has been corrected by OS_MCA */

#define IA64_MCA_WARM_BOOT	-1	/* Warm boot of the system need from SAL */

#define IA64_MCA_COLD_BOOT	-2	/* Cold boot of the system need from SAL */

#define IA64_MCA_HALT		-3	/* System to be halted by SAL */



#define IA64_INIT_RESUME	0x0	/* Resume after return from INIT */

#define IA64_INIT_WARM_BOOT	-1	/* Warm boot of the system need from SAL */



#define IA64_MCA_SAME_CONTEXT	0x0	/* SAL to return to same context */

#define IA64_MCA_NEW_CONTEXT	-1	/* SAL to return to new context */



#endif /* !__ASSEMBLY__ */

#endif /* _ASM_IA64_MCA_H */

CRA Git | Maintained and supported by SUSTech CRA and CCSE