linux_dsm_epyc7002/drivers/acpi/apei/apei-base.c

766 lines
19 KiB
C
Raw Normal View History

/*
* apei-base.c - ACPI Platform Error Interface (APEI) supporting
* infrastructure
*
* APEI allows to report errors (for example from the chipset) to the
* the operating system. This improves NMI handling especially. In
* addition it supports error serialization and error injection.
*
* For more information about APEI, please refer to ACPI Specification
* version 4.0, chapter 17.
*
* This file has Common functions used by more than one APEI table,
* including framework of interpreter for ERST and EINJ; resource
* management for APEI registers.
*
* Copyright (C) 2009, Intel Corp.
* Author: Huang Ying <ying.huang@intel.com>
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License version
* 2 as published by the Free Software Foundation.
*
* This program 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.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/acpi.h>
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
#include <linux/acpi_io.h>
#include <linux/slab.h>
#include <linux/io.h>
#include <linux/kref.h>
#include <linux/rculist.h>
#include <linux/interrupt.h>
#include <linux/debugfs.h>
#include "apei-internal.h"
#define APEI_PFX "APEI: "
/*
* APEI ERST (Error Record Serialization Table) and EINJ (Error
* INJection) interpreter framework.
*/
#define APEI_EXEC_PRESERVE_REGISTER 0x1
void apei_exec_ctx_init(struct apei_exec_context *ctx,
struct apei_exec_ins_type *ins_table,
u32 instructions,
struct acpi_whea_header *action_table,
u32 entries)
{
ctx->ins_table = ins_table;
ctx->instructions = instructions;
ctx->action_table = action_table;
ctx->entries = entries;
}
EXPORT_SYMBOL_GPL(apei_exec_ctx_init);
int __apei_exec_read_register(struct acpi_whea_header *entry, u64 *val)
{
int rc;
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
rc = apei_read(val, &entry->register_region);
if (rc)
return rc;
*val >>= entry->register_region.bit_offset;
*val &= entry->mask;
return 0;
}
int apei_exec_read_register(struct apei_exec_context *ctx,
struct acpi_whea_header *entry)
{
int rc;
u64 val = 0;
rc = __apei_exec_read_register(entry, &val);
if (rc)
return rc;
ctx->value = val;
return 0;
}
EXPORT_SYMBOL_GPL(apei_exec_read_register);
int apei_exec_read_register_value(struct apei_exec_context *ctx,
struct acpi_whea_header *entry)
{
int rc;
rc = apei_exec_read_register(ctx, entry);
if (rc)
return rc;
ctx->value = (ctx->value == entry->value);
return 0;
}
EXPORT_SYMBOL_GPL(apei_exec_read_register_value);
int __apei_exec_write_register(struct acpi_whea_header *entry, u64 val)
{
int rc;
val &= entry->mask;
val <<= entry->register_region.bit_offset;
if (entry->flags & APEI_EXEC_PRESERVE_REGISTER) {
u64 valr = 0;
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
rc = apei_read(&valr, &entry->register_region);
if (rc)
return rc;
valr &= ~(entry->mask << entry->register_region.bit_offset);
val |= valr;
}
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
rc = apei_write(val, &entry->register_region);
return rc;
}
int apei_exec_write_register(struct apei_exec_context *ctx,
struct acpi_whea_header *entry)
{
return __apei_exec_write_register(entry, ctx->value);
}
EXPORT_SYMBOL_GPL(apei_exec_write_register);
int apei_exec_write_register_value(struct apei_exec_context *ctx,
struct acpi_whea_header *entry)
{
int rc;
ctx->value = entry->value;
rc = apei_exec_write_register(ctx, entry);
return rc;
}
EXPORT_SYMBOL_GPL(apei_exec_write_register_value);
int apei_exec_noop(struct apei_exec_context *ctx,
struct acpi_whea_header *entry)
{
return 0;
}
EXPORT_SYMBOL_GPL(apei_exec_noop);
/*
* Interpret the specified action. Go through whole action table,
* execute all instructions belong to the action.
*/
int __apei_exec_run(struct apei_exec_context *ctx, u8 action,
bool optional)
{
int rc = -ENOENT;
u32 i, ip;
struct acpi_whea_header *entry;
apei_exec_ins_func_t run;
ctx->ip = 0;
/*
* "ip" is the instruction pointer of current instruction,
* "ctx->ip" specifies the next instruction to executed,
* instruction "run" function may change the "ctx->ip" to
* implement "goto" semantics.
*/
rewind:
ip = 0;
for (i = 0; i < ctx->entries; i++) {
entry = &ctx->action_table[i];
if (entry->action != action)
continue;
if (ip == ctx->ip) {
if (entry->instruction >= ctx->instructions ||
!ctx->ins_table[entry->instruction].run) {
pr_warning(FW_WARN APEI_PFX
"Invalid action table, unknown instruction type: %d\n",
entry->instruction);
return -EINVAL;
}
run = ctx->ins_table[entry->instruction].run;
rc = run(ctx, entry);
if (rc < 0)
return rc;
else if (rc != APEI_EXEC_SET_IP)
ctx->ip++;
}
ip++;
if (ctx->ip < ip)
goto rewind;
}
return !optional && rc < 0 ? rc : 0;
}
EXPORT_SYMBOL_GPL(__apei_exec_run);
typedef int (*apei_exec_entry_func_t)(struct apei_exec_context *ctx,
struct acpi_whea_header *entry,
void *data);
static int apei_exec_for_each_entry(struct apei_exec_context *ctx,
apei_exec_entry_func_t func,
void *data,
int *end)
{
u8 ins;
int i, rc;
struct acpi_whea_header *entry;
struct apei_exec_ins_type *ins_table = ctx->ins_table;
for (i = 0; i < ctx->entries; i++) {
entry = ctx->action_table + i;
ins = entry->instruction;
if (end)
*end = i;
if (ins >= ctx->instructions || !ins_table[ins].run) {
pr_warning(FW_WARN APEI_PFX
"Invalid action table, unknown instruction type: %d\n",
ins);
return -EINVAL;
}
rc = func(ctx, entry, data);
if (rc)
return rc;
}
return 0;
}
static int pre_map_gar_callback(struct apei_exec_context *ctx,
struct acpi_whea_header *entry,
void *data)
{
u8 ins = entry->instruction;
if (ctx->ins_table[ins].flags & APEI_EXEC_INS_ACCESS_REGISTER)
return apei_map_generic_address(&entry->register_region);
return 0;
}
/*
* Pre-map all GARs in action table to make it possible to access them
* in NMI handler.
*/
int apei_exec_pre_map_gars(struct apei_exec_context *ctx)
{
int rc, end;
rc = apei_exec_for_each_entry(ctx, pre_map_gar_callback,
NULL, &end);
if (rc) {
struct apei_exec_context ctx_unmap;
memcpy(&ctx_unmap, ctx, sizeof(*ctx));
ctx_unmap.entries = end;
apei_exec_post_unmap_gars(&ctx_unmap);
}
return rc;
}
EXPORT_SYMBOL_GPL(apei_exec_pre_map_gars);
static int post_unmap_gar_callback(struct apei_exec_context *ctx,
struct acpi_whea_header *entry,
void *data)
{
u8 ins = entry->instruction;
if (ctx->ins_table[ins].flags & APEI_EXEC_INS_ACCESS_REGISTER)
apei_unmap_generic_address(&entry->register_region);
return 0;
}
/* Post-unmap all GAR in action table. */
int apei_exec_post_unmap_gars(struct apei_exec_context *ctx)
{
return apei_exec_for_each_entry(ctx, post_unmap_gar_callback,
NULL, NULL);
}
EXPORT_SYMBOL_GPL(apei_exec_post_unmap_gars);
/*
* Resource management for GARs in APEI
*/
struct apei_res {
struct list_head list;
unsigned long start;
unsigned long end;
};
/* Collect all resources requested, to avoid conflict */
struct apei_resources apei_resources_all = {
.iomem = LIST_HEAD_INIT(apei_resources_all.iomem),
.ioport = LIST_HEAD_INIT(apei_resources_all.ioport),
};
static int apei_res_add(struct list_head *res_list,
unsigned long start, unsigned long size)
{
struct apei_res *res, *resn, *res_ins = NULL;
unsigned long end = start + size;
if (end <= start)
return 0;
repeat:
list_for_each_entry_safe(res, resn, res_list, list) {
if (res->start > end || res->end < start)
continue;
else if (end <= res->end && start >= res->start) {
kfree(res_ins);
return 0;
}
list_del(&res->list);
res->start = start = min(res->start, start);
res->end = end = max(res->end, end);
kfree(res_ins);
res_ins = res;
goto repeat;
}
if (res_ins)
list_add(&res_ins->list, res_list);
else {
res_ins = kmalloc(sizeof(*res), GFP_KERNEL);
if (!res_ins)
return -ENOMEM;
res_ins->start = start;
res_ins->end = end;
list_add(&res_ins->list, res_list);
}
return 0;
}
static int apei_res_sub(struct list_head *res_list1,
struct list_head *res_list2)
{
struct apei_res *res1, *resn1, *res2, *res;
res1 = list_entry(res_list1->next, struct apei_res, list);
resn1 = list_entry(res1->list.next, struct apei_res, list);
while (&res1->list != res_list1) {
list_for_each_entry(res2, res_list2, list) {
if (res1->start >= res2->end ||
res1->end <= res2->start)
continue;
else if (res1->end <= res2->end &&
res1->start >= res2->start) {
list_del(&res1->list);
kfree(res1);
break;
} else if (res1->end > res2->end &&
res1->start < res2->start) {
res = kmalloc(sizeof(*res), GFP_KERNEL);
if (!res)
return -ENOMEM;
res->start = res2->end;
res->end = res1->end;
res1->end = res2->start;
list_add(&res->list, &res1->list);
resn1 = res;
} else {
if (res1->start < res2->start)
res1->end = res2->start;
else
res1->start = res2->end;
}
}
res1 = resn1;
resn1 = list_entry(resn1->list.next, struct apei_res, list);
}
return 0;
}
static void apei_res_clean(struct list_head *res_list)
{
struct apei_res *res, *resn;
list_for_each_entry_safe(res, resn, res_list, list) {
list_del(&res->list);
kfree(res);
}
}
void apei_resources_fini(struct apei_resources *resources)
{
apei_res_clean(&resources->iomem);
apei_res_clean(&resources->ioport);
}
EXPORT_SYMBOL_GPL(apei_resources_fini);
static int apei_resources_merge(struct apei_resources *resources1,
struct apei_resources *resources2)
{
int rc;
struct apei_res *res;
list_for_each_entry(res, &resources2->iomem, list) {
rc = apei_res_add(&resources1->iomem, res->start,
res->end - res->start);
if (rc)
return rc;
}
list_for_each_entry(res, &resources2->ioport, list) {
rc = apei_res_add(&resources1->ioport, res->start,
res->end - res->start);
if (rc)
return rc;
}
return 0;
}
int apei_resources_add(struct apei_resources *resources,
unsigned long start, unsigned long size,
bool iomem)
{
if (iomem)
return apei_res_add(&resources->iomem, start, size);
else
return apei_res_add(&resources->ioport, start, size);
}
EXPORT_SYMBOL_GPL(apei_resources_add);
/*
* EINJ has two groups of GARs (EINJ table entry and trigger table
* entry), so common resources are subtracted from the trigger table
* resources before the second requesting.
*/
int apei_resources_sub(struct apei_resources *resources1,
struct apei_resources *resources2)
{
int rc;
rc = apei_res_sub(&resources1->iomem, &resources2->iomem);
if (rc)
return rc;
return apei_res_sub(&resources1->ioport, &resources2->ioport);
}
EXPORT_SYMBOL_GPL(apei_resources_sub);
static int apei_get_nvs_callback(__u64 start, __u64 size, void *data)
{
struct apei_resources *resources = data;
return apei_res_add(&resources->iomem, start, size);
}
static int apei_get_nvs_resources(struct apei_resources *resources)
{
return acpi_nvs_for_each_region(apei_get_nvs_callback, resources);
}
/*
* IO memory/port resource management mechanism is used to check
* whether memory/port area used by GARs conflicts with normal memory
* or IO memory/port of devices.
*/
int apei_resources_request(struct apei_resources *resources,
const char *desc)
{
struct apei_res *res, *res_bak = NULL;
struct resource *r;
struct apei_resources nvs_resources;
int rc;
rc = apei_resources_sub(resources, &apei_resources_all);
if (rc)
return rc;
/*
* Some firmware uses ACPI NVS region, that has been marked as
* busy, so exclude it from APEI resources to avoid false
* conflict.
*/
apei_resources_init(&nvs_resources);
rc = apei_get_nvs_resources(&nvs_resources);
if (rc)
goto res_fini;
rc = apei_resources_sub(resources, &nvs_resources);
if (rc)
goto res_fini;
rc = -EINVAL;
list_for_each_entry(res, &resources->iomem, list) {
r = request_mem_region(res->start, res->end - res->start,
desc);
if (!r) {
pr_err(APEI_PFX
"Can not request [mem %#010llx-%#010llx] for %s registers\n",
(unsigned long long)res->start,
(unsigned long long)res->end - 1, desc);
res_bak = res;
goto err_unmap_iomem;
}
}
list_for_each_entry(res, &resources->ioport, list) {
r = request_region(res->start, res->end - res->start, desc);
if (!r) {
pr_err(APEI_PFX
"Can not request [io %#06llx-%#06llx] for %s registers\n",
(unsigned long long)res->start,
(unsigned long long)res->end - 1, desc);
res_bak = res;
goto err_unmap_ioport;
}
}
rc = apei_resources_merge(&apei_resources_all, resources);
if (rc) {
pr_err(APEI_PFX "Fail to merge resources!\n");
goto err_unmap_ioport;
}
return 0;
err_unmap_ioport:
list_for_each_entry(res, &resources->ioport, list) {
if (res == res_bak)
break;
release_region(res->start, res->end - res->start);
}
res_bak = NULL;
err_unmap_iomem:
list_for_each_entry(res, &resources->iomem, list) {
if (res == res_bak)
break;
release_mem_region(res->start, res->end - res->start);
}
res_fini:
apei_resources_fini(&nvs_resources);
return rc;
}
EXPORT_SYMBOL_GPL(apei_resources_request);
void apei_resources_release(struct apei_resources *resources)
{
int rc;
struct apei_res *res;
list_for_each_entry(res, &resources->iomem, list)
release_mem_region(res->start, res->end - res->start);
list_for_each_entry(res, &resources->ioport, list)
release_region(res->start, res->end - res->start);
rc = apei_resources_sub(&apei_resources_all, resources);
if (rc)
pr_err(APEI_PFX "Fail to sub resources!\n");
}
EXPORT_SYMBOL_GPL(apei_resources_release);
static int apei_check_gar(struct acpi_generic_address *reg, u64 *paddr,
u32 *access_bit_width)
{
u32 bit_width, bit_offset, access_size_code, space_id;
bit_width = reg->bit_width;
bit_offset = reg->bit_offset;
access_size_code = reg->access_width;
space_id = reg->space_id;
/* Handle possible alignment issues */
memcpy(paddr, &reg->address, sizeof(*paddr));
if (!*paddr) {
pr_warning(FW_BUG APEI_PFX
"Invalid physical address in GAR [0x%llx/%u/%u/%u/%u]\n",
*paddr, bit_width, bit_offset, access_size_code,
space_id);
return -EINVAL;
}
if (access_size_code < 1 || access_size_code > 4) {
pr_warning(FW_BUG APEI_PFX
"Invalid access size code in GAR [0x%llx/%u/%u/%u/%u]\n",
*paddr, bit_width, bit_offset, access_size_code,
space_id);
return -EINVAL;
}
*access_bit_width = 1UL << (access_size_code + 2);
if ((bit_width + bit_offset) > *access_bit_width) {
pr_warning(FW_BUG APEI_PFX
"Invalid bit width + offset in GAR [0x%llx/%u/%u/%u/%u]\n",
*paddr, bit_width, bit_offset, access_size_code,
space_id);
return -EINVAL;
}
if (space_id != ACPI_ADR_SPACE_SYSTEM_MEMORY &&
space_id != ACPI_ADR_SPACE_SYSTEM_IO) {
pr_warning(FW_BUG APEI_PFX
"Invalid address space type in GAR [0x%llx/%u/%u/%u/%u]\n",
*paddr, bit_width, bit_offset, access_size_code,
space_id);
return -EINVAL;
}
return 0;
}
int apei_map_generic_address(struct acpi_generic_address *reg)
{
int rc;
u32 access_bit_width;
u64 address;
rc = apei_check_gar(reg, &address, &access_bit_width);
if (rc)
return rc;
return acpi_os_map_generic_address(reg);
}
EXPORT_SYMBOL_GPL(apei_map_generic_address);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
/* read GAR in interrupt (including NMI) or process context */
int apei_read(u64 *val, struct acpi_generic_address *reg)
{
int rc;
u32 access_bit_width;
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
u64 address;
acpi_status status;
rc = apei_check_gar(reg, &address, &access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (rc)
return rc;
*val = 0;
switch(reg->space_id) {
case ACPI_ADR_SPACE_SYSTEM_MEMORY:
status = acpi_os_read_memory((acpi_physical_address) address,
val, access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (ACPI_FAILURE(status))
return -EIO;
break;
case ACPI_ADR_SPACE_SYSTEM_IO:
status = acpi_os_read_port(address, (u32 *)val,
access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (ACPI_FAILURE(status))
return -EIO;
break;
default:
return -EINVAL;
}
return 0;
}
EXPORT_SYMBOL_GPL(apei_read);
/* write GAR in interrupt (including NMI) or process context */
int apei_write(u64 val, struct acpi_generic_address *reg)
{
int rc;
u32 access_bit_width;
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
u64 address;
acpi_status status;
rc = apei_check_gar(reg, &address, &access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (rc)
return rc;
switch (reg->space_id) {
case ACPI_ADR_SPACE_SYSTEM_MEMORY:
status = acpi_os_write_memory((acpi_physical_address) address,
val, access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (ACPI_FAILURE(status))
return -EIO;
break;
case ACPI_ADR_SPACE_SYSTEM_IO:
status = acpi_os_write_port(address, val, access_bit_width);
ACPI APEI: Convert atomicio routines APEI needs memory access in interrupt context. The obvious choice is acpi_read(), but originally it couldn't be used in interrupt context because it makes temporary mappings with ioremap(). Therefore, we added drivers/acpi/atomicio.c, which provides: acpi_pre_map_gar() -- ioremap in process context acpi_atomic_read() -- memory access in interrupt context acpi_post_unmap_gar() -- iounmap Later we added acpi_os_map_generic_address() (2971852) and enhanced acpi_read() so it works in interrupt context as long as the address has been previously mapped (620242a). Now this sequence: acpi_os_map_generic_address() -- ioremap in process context acpi_read()/apei_read() -- now OK in interrupt context acpi_os_unmap_generic_address() is equivalent to what atomicio.c provides. This patch introduces apei_read() and apei_write(), which currently are functional equivalents of acpi_read() and acpi_write(). This is mainly proactive, to prevent APEI breakages if acpi_read() and acpi_write() are ever augmented to support the 'bit_offset' field of GAS, as APEI's __apei_exec_write_register() precludes splitting up functionality related to 'bit_offset' and APEI's 'mask' (see its APEI_EXEC_PRESERVE_REGISTER block). With apei_read() and apei_write() in place, usages of atomicio routines are converted to apei_read()/apei_write() and existing calls within osl.c and the CA, based on the re-factoring that was done in an earlier patch series - http://marc.info/?l=linux-acpi&m=128769263327206&w=2: acpi_pre_map_gar() --> acpi_os_map_generic_address() acpi_post_unmap_gar() --> acpi_os_unmap_generic_address() acpi_atomic_read() --> apei_read() acpi_atomic_write() --> apei_write() Note that acpi_read() and acpi_write() currently use 'bit_width' for accessing GARs which seems incorrect. 'bit_width' is the size of the register, while 'access_width' is the size of the access the processor must generate on the bus. The 'access_width' may be larger, for example, if the hardware only supports 32-bit or 64-bit reads. I wanted to minimize any possible impacts with this patch series so I did *not* change this behavior. Signed-off-by: Myron Stowe <myron.stowe@redhat.com> Signed-off-by: Len Brown <len.brown@intel.com>
2011-11-08 06:23:41 +07:00
if (ACPI_FAILURE(status))
return -EIO;
break;
default:
return -EINVAL;
}
return 0;
}
EXPORT_SYMBOL_GPL(apei_write);
static int collect_res_callback(struct apei_exec_context *ctx,
struct acpi_whea_header *entry,
void *data)
{
struct apei_resources *resources = data;
struct acpi_generic_address *reg = &entry->register_region;
u8 ins = entry->instruction;
u32 access_bit_width;
u64 paddr;
int rc;
if (!(ctx->ins_table[ins].flags & APEI_EXEC_INS_ACCESS_REGISTER))
return 0;
rc = apei_check_gar(reg, &paddr, &access_bit_width);
if (rc)
return rc;
switch (reg->space_id) {
case ACPI_ADR_SPACE_SYSTEM_MEMORY:
return apei_res_add(&resources->iomem, paddr,
access_bit_width / 8);
case ACPI_ADR_SPACE_SYSTEM_IO:
return apei_res_add(&resources->ioport, paddr,
access_bit_width / 8);
default:
return -EINVAL;
}
}
/*
* Same register may be used by multiple instructions in GARs, so
* resources are collected before requesting.
*/
int apei_exec_collect_resources(struct apei_exec_context *ctx,
struct apei_resources *resources)
{
return apei_exec_for_each_entry(ctx, collect_res_callback,
resources, NULL);
}
EXPORT_SYMBOL_GPL(apei_exec_collect_resources);
struct dentry *apei_get_debugfs_dir(void)
{
static struct dentry *dapei;
if (!dapei)
dapei = debugfs_create_dir("apei", NULL);
return dapei;
}
EXPORT_SYMBOL_GPL(apei_get_debugfs_dir);
int apei_osc_setup(void)
{
static u8 whea_uuid_str[] = "ed855e0c-6c90-47bf-a62a-26de0fc5ad5c";
acpi_handle handle;
u32 capbuf[3];
struct acpi_osc_context context = {
.uuid_str = whea_uuid_str,
.rev = 1,
.cap.length = sizeof(capbuf),
.cap.pointer = capbuf,
};
capbuf[OSC_QUERY_TYPE] = OSC_QUERY_ENABLE;
capbuf[OSC_SUPPORT_TYPE] = 1;
capbuf[OSC_CONTROL_TYPE] = 0;
if (ACPI_FAILURE(acpi_get_handle(NULL, "\\_SB", &handle))
|| ACPI_FAILURE(acpi_run_osc(handle, &context)))
return -EIO;
else {
kfree(context.ret.pointer);
return 0;
}
}
EXPORT_SYMBOL_GPL(apei_osc_setup);