2019-05-27 13:55:05 +07:00
|
|
|
// SPDX-License-Identifier: GPL-2.0-or-later
|
2005-04-17 05:20:36 +07:00
|
|
|
/*
|
2007-05-24 00:28:01 +07:00
|
|
|
* Copyright IBM Corporation 2001, 2005, 2006
|
|
|
|
* Copyright Dave Engebretsen & Todd Inglett 2001
|
|
|
|
* Copyright Linas Vepstas 2005, 2006
|
2012-02-28 03:03:51 +07:00
|
|
|
* Copyright 2001-2012 IBM Corporation.
|
2005-11-04 07:47:50 +07:00
|
|
|
*
|
2007-05-24 00:28:01 +07:00
|
|
|
* Please address comments and feedback to Linas Vepstas <linas@austin.ibm.com>
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
|
|
|
|
2005-11-04 07:50:10 +07:00
|
|
|
#include <linux/delay.h>
|
2012-02-28 03:03:51 +07:00
|
|
|
#include <linux/sched.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/list.h>
|
|
|
|
#include <linux/pci.h>
|
2014-07-15 14:00:56 +07:00
|
|
|
#include <linux/iommu.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/proc_fs.h>
|
|
|
|
#include <linux/rbtree.h>
|
2014-02-12 14:24:56 +07:00
|
|
|
#include <linux/reboot.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <linux/seq_file.h>
|
|
|
|
#include <linux/spinlock.h>
|
2011-05-27 21:46:24 +07:00
|
|
|
#include <linux/export.h>
|
2007-12-21 11:52:07 +07:00
|
|
|
#include <linux/of.h>
|
|
|
|
|
2011-07-27 06:09:06 +07:00
|
|
|
#include <linux/atomic.h>
|
2017-02-10 08:04:56 +07:00
|
|
|
#include <asm/debugfs.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <asm/eeh.h>
|
2005-11-04 07:50:04 +07:00
|
|
|
#include <asm/eeh_event.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <asm/io.h>
|
2014-06-10 08:41:56 +07:00
|
|
|
#include <asm/iommu.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <asm/machdep.h>
|
2005-11-04 07:50:04 +07:00
|
|
|
#include <asm/ppc-pci.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
#include <asm/rtas.h>
|
2017-07-27 13:24:53 +07:00
|
|
|
#include <asm/pte-walk.h>
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
|
|
|
|
/** Overview:
|
2016-02-16 19:06:05 +07:00
|
|
|
* EEH, or "Enhanced Error Handling" is a PCI bridge technology for
|
2005-04-17 05:20:36 +07:00
|
|
|
* dealing with PCI bus errors that can't be dealt with within the
|
|
|
|
* usual PCI framework, except by check-stopping the CPU. Systems
|
|
|
|
* that are designed for high-availability/reliability cannot afford
|
|
|
|
* to crash due to a "mere" PCI error, thus the need for EEH.
|
|
|
|
* An EEH-capable bridge operates by converting a detected error
|
|
|
|
* into a "slot freeze", taking the PCI adapter off-line, making
|
|
|
|
* the slot behave, from the OS'es point of view, as if the slot
|
|
|
|
* were "empty": all reads return 0xff's and all writes are silently
|
|
|
|
* ignored. EEH slot isolation events can be triggered by parity
|
|
|
|
* errors on the address or data busses (e.g. during posted writes),
|
2005-11-04 07:47:50 +07:00
|
|
|
* which in turn might be caused by low voltage on the bus, dust,
|
|
|
|
* vibration, humidity, radioactivity or plain-old failed hardware.
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
|
|
|
* Note, however, that one of the leading causes of EEH slot
|
|
|
|
* freeze events are buggy device drivers, buggy device microcode,
|
|
|
|
* or buggy device hardware. This is because any attempt by the
|
|
|
|
* device to bus-master data to a memory address that is not
|
|
|
|
* assigned to the device will trigger a slot freeze. (The idea
|
|
|
|
* is to prevent devices-gone-wild from corrupting system memory).
|
|
|
|
* Buggy hardware/drivers will have a miserable time co-existing
|
|
|
|
* with EEH.
|
|
|
|
*
|
|
|
|
* Ideally, a PCI device driver, when suspecting that an isolation
|
2011-03-31 08:57:33 +07:00
|
|
|
* event has occurred (e.g. by reading 0xff's), will then ask EEH
|
2005-04-17 05:20:36 +07:00
|
|
|
* whether this is the case, and then take appropriate steps to
|
|
|
|
* reset the PCI slot, the PCI device, and then resume operations.
|
|
|
|
* However, until that day, the checking is done here, with the
|
|
|
|
* eeh_check_failure() routine embedded in the MMIO macros. If
|
|
|
|
* the slot is found to be isolated, an "EEH Event" is synthesized
|
|
|
|
* and sent out for processing.
|
|
|
|
*/
|
|
|
|
|
2005-11-04 07:49:31 +07:00
|
|
|
/* If a device driver keeps reading an MMIO register in an interrupt
|
2008-07-21 23:40:17 +07:00
|
|
|
* handler after a slot isolation event, it might be broken.
|
|
|
|
* This sets the threshold for how many read attempts we allow
|
|
|
|
* before printing an error message.
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
2007-03-20 02:53:22 +07:00
|
|
|
#define EEH_MAX_FAILS 2100000
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2007-05-09 23:38:11 +07:00
|
|
|
/* Time to wait for a PCI slot to report status, in milliseconds */
|
2013-11-26 05:27:54 +07:00
|
|
|
#define PCI_BUS_RESET_WAIT_MSEC (5*60*1000)
|
2007-03-20 02:58:07 +07:00
|
|
|
|
2012-09-08 05:44:21 +07:00
|
|
|
/*
|
2014-04-24 15:00:17 +07:00
|
|
|
* EEH probe mode support, which is part of the flags,
|
|
|
|
* is to support multiple platforms for EEH. Some platforms
|
|
|
|
* like pSeries do PCI emunation based on device tree.
|
|
|
|
* However, other platforms like powernv probe PCI devices
|
|
|
|
* from hardware. The flag is used to distinguish that.
|
|
|
|
* In addition, struct eeh_ops::probe would be invoked for
|
|
|
|
* particular OF node or PCI device so that the corresponding
|
|
|
|
* PE would be created there.
|
2012-09-08 05:44:21 +07:00
|
|
|
*/
|
2014-04-24 15:00:17 +07:00
|
|
|
int eeh_subsystem_flags;
|
|
|
|
EXPORT_SYMBOL(eeh_subsystem_flags);
|
|
|
|
|
2014-12-11 10:28:56 +07:00
|
|
|
/*
|
|
|
|
* EEH allowed maximal frozen times. If one particular PE's
|
|
|
|
* frozen count in last hour exceeds this limit, the PE will
|
|
|
|
* be forced to be offline permanently.
|
|
|
|
*/
|
2019-02-15 07:48:11 +07:00
|
|
|
u32 eeh_max_freezes = 5;
|
2014-12-11 10:28:56 +07:00
|
|
|
|
2019-02-15 07:48:16 +07:00
|
|
|
/*
|
|
|
|
* Controls whether a recovery event should be scheduled when an
|
|
|
|
* isolated device is discovered. This is only really useful for
|
|
|
|
* debugging problems with the EEH core.
|
|
|
|
*/
|
|
|
|
bool eeh_debugfs_no_recover;
|
|
|
|
|
2014-04-24 15:00:17 +07:00
|
|
|
/* Platform dependent EEH operations */
|
|
|
|
struct eeh_ops *eeh_ops = NULL;
|
2012-09-08 05:44:21 +07:00
|
|
|
|
2005-11-04 07:49:23 +07:00
|
|
|
/* Lock to avoid races due to multiple reports of an error */
|
2013-06-20 12:21:03 +07:00
|
|
|
DEFINE_RAW_SPINLOCK(confirm_error_lock);
|
2016-09-28 11:34:54 +07:00
|
|
|
EXPORT_SYMBOL_GPL(confirm_error_lock);
|
2005-11-04 07:49:23 +07:00
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/* Lock to protect passed flags */
|
|
|
|
static DEFINE_MUTEX(eeh_dev_mutex);
|
|
|
|
|
2007-05-09 23:38:11 +07:00
|
|
|
/* Buffer for reporting pci register dumps. Its here in BSS, and
|
|
|
|
* not dynamically alloced, so that it ends up in RMO where RTAS
|
|
|
|
* can access it.
|
|
|
|
*/
|
2014-09-30 09:39:08 +07:00
|
|
|
#define EEH_PCI_REGS_LOG_LEN 8192
|
2007-05-09 06:35:32 +07:00
|
|
|
static unsigned char pci_regs_buf[EEH_PCI_REGS_LOG_LEN];
|
|
|
|
|
2012-02-29 22:47:45 +07:00
|
|
|
/*
|
|
|
|
* The struct is used to maintain the EEH global statistic
|
|
|
|
* information. Besides, the EEH global statistics will be
|
|
|
|
* exported to user space through procfs
|
|
|
|
*/
|
|
|
|
struct eeh_stats {
|
|
|
|
u64 no_device; /* PCI device not found */
|
|
|
|
u64 no_dn; /* OF node not found */
|
|
|
|
u64 no_cfg_addr; /* Config address not found */
|
|
|
|
u64 ignored_check; /* EEH check skipped */
|
|
|
|
u64 total_mmio_ffs; /* Total EEH checks */
|
|
|
|
u64 false_positives; /* Unnecessary EEH checks */
|
|
|
|
u64 slot_resets; /* PE reset */
|
|
|
|
};
|
|
|
|
|
|
|
|
static struct eeh_stats eeh_stats;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2014-04-24 15:00:18 +07:00
|
|
|
static int __init eeh_setup(char *str)
|
|
|
|
{
|
|
|
|
if (!strcmp(str, "off"))
|
2014-07-17 11:41:38 +07:00
|
|
|
eeh_add_flag(EEH_FORCE_DISABLED);
|
2014-11-22 17:58:09 +07:00
|
|
|
else if (!strcmp(str, "early_log"))
|
|
|
|
eeh_add_flag(EEH_EARLY_DUMP_LOG);
|
2014-04-24 15:00:18 +07:00
|
|
|
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
__setup("eeh=", eeh_setup);
|
|
|
|
|
2019-08-16 11:48:10 +07:00
|
|
|
void eeh_show_enabled(void)
|
|
|
|
{
|
|
|
|
if (eeh_has_flag(EEH_FORCE_DISABLED))
|
|
|
|
pr_info("EEH: Recovery disabled by kernel parameter.\n");
|
|
|
|
else if (eeh_has_flag(EEH_ENABLED))
|
|
|
|
pr_info("EEH: Capable adapter found: recovery enabled.\n");
|
|
|
|
else
|
|
|
|
pr_info("EEH: No capable adapters found: recovery disabled.\n");
|
|
|
|
}
|
|
|
|
|
2014-09-30 09:39:08 +07:00
|
|
|
/*
|
|
|
|
* This routine captures assorted PCI configuration space data
|
|
|
|
* for the indicated PCI device, and puts them into a buffer
|
|
|
|
* for RTAS error logging.
|
2007-05-09 06:35:32 +07:00
|
|
|
*/
|
2014-09-30 09:39:08 +07:00
|
|
|
static size_t eeh_dump_dev_log(struct eeh_dev *edev, char *buf, size_t len)
|
2007-05-09 06:35:32 +07:00
|
|
|
{
|
2015-03-17 12:15:07 +07:00
|
|
|
struct pci_dn *pdn = eeh_dev_to_pdn(edev);
|
2007-05-09 06:35:32 +07:00
|
|
|
u32 cfg;
|
2007-05-09 06:36:21 +07:00
|
|
|
int cap, i;
|
2014-07-17 11:41:40 +07:00
|
|
|
int n = 0, l = 0;
|
|
|
|
char buffer[128];
|
2007-05-09 06:35:32 +07:00
|
|
|
|
2018-09-12 08:23:20 +07:00
|
|
|
if (!pdn) {
|
|
|
|
pr_warn("EEH: Note: No error log for absent device.\n");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-07-23 00:05:29 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "%04x:%02x:%02x.%01x\n",
|
2017-08-29 14:34:01 +07:00
|
|
|
pdn->phb->global_number, pdn->busno,
|
2015-03-17 12:15:07 +07:00
|
|
|
PCI_SLOT(pdn->devfn), PCI_FUNC(pdn->devfn));
|
2016-07-23 00:05:29 +07:00
|
|
|
pr_warn("EEH: of node=%04x:%02x:%02x.%01x\n",
|
2017-08-29 14:34:01 +07:00
|
|
|
pdn->phb->global_number, pdn->busno,
|
2015-03-17 12:15:07 +07:00
|
|
|
PCI_SLOT(pdn->devfn), PCI_FUNC(pdn->devfn));
|
2007-05-09 06:36:21 +07:00
|
|
|
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, PCI_VENDOR_ID, 4, &cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "dev/vend:%08x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: PCI device/vendor: %08x\n", cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, PCI_COMMAND, 4, &cfg);
|
2007-05-09 06:35:32 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "cmd/stat:%x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: PCI cmd/status register: %08x\n", cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
|
2007-07-27 05:35:40 +07:00
|
|
|
/* Gather bridge-specific registers */
|
2014-04-24 15:00:16 +07:00
|
|
|
if (edev->mode & EEH_DEV_BRIDGE) {
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, PCI_SEC_STATUS, 2, &cfg);
|
2007-07-27 05:35:40 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "sec stat:%x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: Bridge secondary status: %04x\n", cfg);
|
2007-07-27 05:35:40 +07:00
|
|
|
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, PCI_BRIDGE_CONTROL, 2, &cfg);
|
2007-07-27 05:35:40 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "brdg ctl:%x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: Bridge control: %04x\n", cfg);
|
2007-07-27 05:35:40 +07:00
|
|
|
}
|
|
|
|
|
2007-05-09 06:36:21 +07:00
|
|
|
/* Dump out the PCI-X command and status regs */
|
2014-04-24 15:00:16 +07:00
|
|
|
cap = edev->pcix_cap;
|
2007-05-09 06:36:21 +07:00
|
|
|
if (cap) {
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, cap, 4, &cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "pcix-cmd:%x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: PCI-X cmd: %08x\n", cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, cap+4, 4, &cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "pcix-stat:%x\n", cfg);
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: PCI-X status: %08x\n", cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
}
|
|
|
|
|
2014-04-24 15:00:16 +07:00
|
|
|
/* If PCI-E capable, dump PCI-E cap 10 */
|
|
|
|
cap = edev->pcie_cap;
|
|
|
|
if (cap) {
|
2007-05-09 06:36:21 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "pci-e cap10:\n");
|
2014-04-24 15:00:15 +07:00
|
|
|
pr_warn("EEH: PCI-E capabilities and status follow:\n");
|
2007-05-09 06:36:21 +07:00
|
|
|
|
|
|
|
for (i=0; i<=8; i++) {
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, cap+4*i, 4, &cfg);
|
2007-05-09 06:36:21 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "%02x:%x\n", 4*i, cfg);
|
2014-07-17 11:41:40 +07:00
|
|
|
|
|
|
|
if ((i % 4) == 0) {
|
|
|
|
if (i != 0)
|
|
|
|
pr_warn("%s\n", buffer);
|
|
|
|
|
|
|
|
l = scnprintf(buffer, sizeof(buffer),
|
|
|
|
"EEH: PCI-E %02x: %08x ",
|
|
|
|
4*i, cfg);
|
|
|
|
} else {
|
|
|
|
l += scnprintf(buffer+l, sizeof(buffer)-l,
|
|
|
|
"%08x ", cfg);
|
|
|
|
}
|
|
|
|
|
2007-05-09 06:36:21 +07:00
|
|
|
}
|
2014-07-17 11:41:40 +07:00
|
|
|
|
|
|
|
pr_warn("%s\n", buffer);
|
2014-04-24 15:00:16 +07:00
|
|
|
}
|
2007-05-09 06:36:21 +07:00
|
|
|
|
2014-04-24 15:00:16 +07:00
|
|
|
/* If AER capable, dump it */
|
|
|
|
cap = edev->aer_cap;
|
|
|
|
if (cap) {
|
|
|
|
n += scnprintf(buf+n, len-n, "pci-e AER:\n");
|
|
|
|
pr_warn("EEH: PCI-E AER capability register set follows:\n");
|
|
|
|
|
2014-07-17 11:41:40 +07:00
|
|
|
for (i=0; i<=13; i++) {
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, cap+4*i, 4, &cfg);
|
2014-04-24 15:00:16 +07:00
|
|
|
n += scnprintf(buf+n, len-n, "%02x:%x\n", 4*i, cfg);
|
2014-07-17 11:41:40 +07:00
|
|
|
|
|
|
|
if ((i % 4) == 0) {
|
|
|
|
if (i != 0)
|
|
|
|
pr_warn("%s\n", buffer);
|
|
|
|
|
|
|
|
l = scnprintf(buffer, sizeof(buffer),
|
|
|
|
"EEH: PCI-E AER %02x: %08x ",
|
|
|
|
4*i, cfg);
|
|
|
|
} else {
|
|
|
|
l += scnprintf(buffer+l, sizeof(buffer)-l,
|
|
|
|
"%08x ", cfg);
|
|
|
|
}
|
2007-05-09 06:36:21 +07:00
|
|
|
}
|
2014-07-17 11:41:40 +07:00
|
|
|
|
|
|
|
pr_warn("%s\n", buffer);
|
2007-05-09 06:36:21 +07:00
|
|
|
}
|
2007-07-27 05:35:40 +07:00
|
|
|
|
2007-05-09 06:35:32 +07:00
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
2018-05-25 10:11:32 +07:00
|
|
|
static void *eeh_dump_pe_log(struct eeh_pe *pe, void *flag)
|
2014-09-30 09:39:08 +07:00
|
|
|
{
|
|
|
|
struct eeh_dev *edev, *tmp;
|
|
|
|
size_t *plen = flag;
|
|
|
|
|
|
|
|
eeh_pe_for_each_dev(pe, edev, tmp)
|
|
|
|
*plen += eeh_dump_dev_log(edev, pci_regs_buf + *plen,
|
|
|
|
EEH_PCI_REGS_LOG_LEN - *plen);
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
2012-02-28 03:03:51 +07:00
|
|
|
/**
|
|
|
|
* eeh_slot_error_detail - Generate combined log including driver log and error log
|
2012-09-08 05:44:16 +07:00
|
|
|
* @pe: EEH PE
|
2012-02-28 03:03:51 +07:00
|
|
|
* @severity: temporary or permanent error log
|
|
|
|
*
|
|
|
|
* This routine should be called to generate the combined log, which
|
|
|
|
* is comprised of driver log and error log. The driver log is figured
|
|
|
|
* out from the config space of the corresponding PCI device, while
|
|
|
|
* the error log is fetched through platform dependent function call.
|
|
|
|
*/
|
2012-09-08 05:44:16 +07:00
|
|
|
void eeh_slot_error_detail(struct eeh_pe *pe, int severity)
|
2007-05-09 06:35:32 +07:00
|
|
|
{
|
|
|
|
size_t loglen = 0;
|
|
|
|
|
2013-06-27 12:46:42 +07:00
|
|
|
/*
|
|
|
|
* When the PHB is fenced or dead, it's pointless to collect
|
|
|
|
* the data from PCI config space because it should return
|
|
|
|
* 0xFF's. For ER, we still retrieve the data from the PCI
|
|
|
|
* config space.
|
2014-04-24 15:00:14 +07:00
|
|
|
*
|
|
|
|
* For pHyp, we have to enable IO for log retrieval. Otherwise,
|
|
|
|
* 0xFF's is always returned from PCI config space.
|
2017-01-06 06:39:49 +07:00
|
|
|
*
|
|
|
|
* When the @severity is EEH_LOG_PERM, the PE is going to be
|
|
|
|
* removed. Prior to that, the drivers for devices included in
|
|
|
|
* the PE will be closed. The drivers rely on working IO path
|
|
|
|
* to bring the devices to quiet state. Otherwise, PCI traffic
|
|
|
|
* from those devices after they are removed is like to cause
|
|
|
|
* another unexpected EEH error.
|
2013-06-27 12:46:42 +07:00
|
|
|
*/
|
2014-04-24 15:00:07 +07:00
|
|
|
if (!(pe->type & EEH_PE_PHB)) {
|
2017-01-06 06:39:49 +07:00
|
|
|
if (eeh_has_flag(EEH_ENABLE_IO_FOR_LOG) ||
|
|
|
|
severity == EEH_LOG_PERM)
|
2014-04-24 15:00:14 +07:00
|
|
|
eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
|
2015-08-28 08:57:00 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The config space of some PCI devices can't be accessed
|
|
|
|
* when their PEs are in frozen state. Otherwise, fenced
|
|
|
|
* PHB might be seen. Those PEs are identified with flag
|
|
|
|
* EEH_PE_CFG_RESTRICTED, indicating EEH_PE_CFG_BLOCKED
|
|
|
|
* is set automatically when the PE is put to EEH_PE_ISOLATED.
|
|
|
|
*
|
|
|
|
* Restoring BARs possibly triggers PCI config access in
|
|
|
|
* (OPAL) firmware and then causes fenced PHB. If the
|
|
|
|
* PCI config is blocked with flag EEH_PE_CFG_BLOCKED, it's
|
|
|
|
* pointless to restore BARs and dump config space.
|
|
|
|
*/
|
2013-06-27 12:46:42 +07:00
|
|
|
eeh_ops->configure_bridge(pe);
|
2015-08-28 08:57:00 +07:00
|
|
|
if (!(pe->state & EEH_PE_CFG_BLOCKED)) {
|
|
|
|
eeh_pe_restore_bars(pe);
|
2013-06-27 12:46:42 +07:00
|
|
|
|
2015-08-28 08:57:00 +07:00
|
|
|
pci_regs_buf[0] = 0;
|
|
|
|
eeh_pe_traverse(pe, eeh_dump_pe_log, &loglen);
|
|
|
|
}
|
2013-06-27 12:46:42 +07:00
|
|
|
}
|
2012-09-08 05:44:16 +07:00
|
|
|
|
|
|
|
eeh_ops->get_log(pe, severity, pci_regs_buf, loglen);
|
2007-05-09 06:35:32 +07:00
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/**
|
2012-02-28 03:03:51 +07:00
|
|
|
* eeh_token_to_phys - Convert EEH address token to phys address
|
|
|
|
* @token: I/O token, should be address in the form 0xA....
|
|
|
|
*
|
|
|
|
* This routine should be called to convert virtual I/O address
|
|
|
|
* to physical one.
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
|
|
|
static inline unsigned long eeh_token_to_phys(unsigned long token)
|
|
|
|
{
|
|
|
|
pte_t *ptep;
|
|
|
|
unsigned long pa;
|
2013-06-20 16:00:18 +07:00
|
|
|
int hugepage_shift;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2013-06-20 16:00:18 +07:00
|
|
|
/*
|
2015-03-30 12:11:03 +07:00
|
|
|
* We won't find hugepages here(this is iomem). Hence we are not
|
|
|
|
* worried about _PAGE_SPLITTING/collapse. Also we will not hit
|
|
|
|
* page table free, because of init_mm.
|
2013-06-20 16:00:18 +07:00
|
|
|
*/
|
2017-07-27 13:24:53 +07:00
|
|
|
ptep = find_init_mm_pte(token, &hugepage_shift);
|
2005-04-17 05:20:36 +07:00
|
|
|
if (!ptep)
|
|
|
|
return token;
|
|
|
|
|
2019-07-10 22:05:17 +07:00
|
|
|
pa = pte_pfn(*ptep);
|
|
|
|
|
|
|
|
/* On radix we can do hugepage mappings for io, so handle that */
|
|
|
|
if (hugepage_shift) {
|
|
|
|
pa <<= hugepage_shift;
|
|
|
|
pa |= token & ((1ul << hugepage_shift) - 1);
|
|
|
|
} else {
|
|
|
|
pa <<= PAGE_SHIFT;
|
|
|
|
pa |= token & (PAGE_SIZE - 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
return pa;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2013-06-20 12:21:16 +07:00
|
|
|
/*
|
|
|
|
* On PowerNV platform, we might already have fenced PHB there.
|
|
|
|
* For that case, it's meaningless to recover frozen PE. Intead,
|
|
|
|
* We have to handle fenced PHB firstly.
|
|
|
|
*/
|
|
|
|
static int eeh_phb_check_failure(struct eeh_pe *pe)
|
|
|
|
{
|
|
|
|
struct eeh_pe *phb_pe;
|
|
|
|
unsigned long flags;
|
|
|
|
int ret;
|
|
|
|
|
2014-07-17 11:41:38 +07:00
|
|
|
if (!eeh_has_flag(EEH_PROBE_MODE_DEV))
|
2013-06-20 12:21:16 +07:00
|
|
|
return -EPERM;
|
|
|
|
|
|
|
|
/* Find the PHB PE */
|
|
|
|
phb_pe = eeh_phb_pe_get(pe->phb);
|
|
|
|
if (!phb_pe) {
|
2016-11-16 10:02:15 +07:00
|
|
|
pr_warn("%s Can't find PE for PHB#%x\n",
|
2014-07-17 11:41:41 +07:00
|
|
|
__func__, pe->phb->global_number);
|
2013-06-20 12:21:16 +07:00
|
|
|
return -EEXIST;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* If the PHB has been in problematic state */
|
|
|
|
eeh_serialize_lock(&flags);
|
2014-04-24 15:00:07 +07:00
|
|
|
if (phb_pe->state & EEH_PE_ISOLATED) {
|
2013-06-20 12:21:16 +07:00
|
|
|
ret = 0;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Check PHB state */
|
|
|
|
ret = eeh_ops->get_state(phb_pe, NULL);
|
|
|
|
if ((ret < 0) ||
|
2018-03-19 09:49:23 +07:00
|
|
|
(ret == EEH_STATE_NOT_SUPPORT) || eeh_state_active(ret)) {
|
2013-06-20 12:21:16 +07:00
|
|
|
ret = 0;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Isolate the PHB and send event */
|
2018-09-12 08:23:31 +07:00
|
|
|
eeh_pe_mark_isolated(phb_pe);
|
2013-06-20 12:21:16 +07:00
|
|
|
eeh_serialize_unlock(flags);
|
|
|
|
|
2019-09-03 17:15:56 +07:00
|
|
|
pr_debug("EEH: PHB#%x failure detected, location: %s\n",
|
2014-06-11 15:26:44 +07:00
|
|
|
phb_pe->phb->global_number, eeh_pe_loc_get(phb_pe));
|
2013-09-06 08:00:05 +07:00
|
|
|
eeh_send_failure_event(phb_pe);
|
2013-06-20 12:21:16 +07:00
|
|
|
return 1;
|
|
|
|
out:
|
|
|
|
eeh_serialize_unlock(flags);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/**
|
2012-09-08 05:44:22 +07:00
|
|
|
* eeh_dev_check_failure - Check if all 1's data is due to EEH slot freeze
|
|
|
|
* @edev: eeh device
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
|
|
|
* Check for an EEH failure for the given device node. Call this
|
|
|
|
* routine if the result of a read was all 0xff's and you want to
|
|
|
|
* find out if this is due to an EEH slot freeze. This routine
|
|
|
|
* will query firmware for the EEH status.
|
|
|
|
*
|
|
|
|
* Returns 0 if there has not been an EEH error; otherwise returns
|
2005-11-04 07:47:50 +07:00
|
|
|
* a non-zero value and queues up a slot isolation event notification.
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
|
|
|
* It is safe to call this routine in an interrupt context.
|
|
|
|
*/
|
2012-09-08 05:44:22 +07:00
|
|
|
int eeh_dev_check_failure(struct eeh_dev *edev)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
unsigned long flags;
|
2017-08-29 14:34:03 +07:00
|
|
|
struct device_node *dn;
|
2012-09-08 05:44:22 +07:00
|
|
|
struct pci_dev *dev;
|
2019-09-03 17:15:56 +07:00
|
|
|
struct eeh_pe *pe, *parent_pe;
|
2005-11-04 07:49:23 +07:00
|
|
|
int rc = 0;
|
2015-03-17 12:15:08 +07:00
|
|
|
const char *location = NULL;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.total_mmio_ffs++;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2014-02-12 14:24:55 +07:00
|
|
|
if (!eeh_enabled())
|
2005-04-17 05:20:36 +07:00
|
|
|
return 0;
|
|
|
|
|
2012-09-08 05:44:22 +07:00
|
|
|
if (!edev) {
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.no_dn++;
|
2005-04-17 05:20:36 +07:00
|
|
|
return 0;
|
2005-11-04 07:48:52 +07:00
|
|
|
}
|
2012-09-08 05:44:22 +07:00
|
|
|
dev = eeh_dev_to_pci_dev(edev);
|
2014-09-17 09:48:26 +07:00
|
|
|
pe = eeh_dev_to_pe(edev);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* Access to IO BARs might get this far and still not want checking. */
|
2012-09-08 05:44:13 +07:00
|
|
|
if (!pe) {
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.ignored_check++;
|
2019-08-16 11:48:13 +07:00
|
|
|
eeh_edev_dbg(edev, "Ignored check\n");
|
2005-04-17 05:20:36 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-09-08 05:44:13 +07:00
|
|
|
if (!pe->addr && !pe->config_addr) {
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.no_cfg_addr++;
|
2005-04-17 05:20:36 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-06-20 12:21:16 +07:00
|
|
|
/*
|
|
|
|
* On PowerNV platform, we might already have fenced PHB
|
|
|
|
* there and we need take care of that firstly.
|
|
|
|
*/
|
|
|
|
ret = eeh_phb_check_failure(pe);
|
|
|
|
if (ret > 0)
|
|
|
|
return ret;
|
|
|
|
|
2014-06-10 08:41:55 +07:00
|
|
|
/*
|
|
|
|
* If the PE isn't owned by us, we shouldn't check the
|
|
|
|
* state. Instead, let the owner handle it if the PE has
|
|
|
|
* been frozen.
|
|
|
|
*/
|
|
|
|
if (eeh_pe_passed(pe))
|
|
|
|
return 0;
|
|
|
|
|
2005-11-04 07:49:23 +07:00
|
|
|
/* If we already have a pending isolation event for this
|
|
|
|
* slot, we know it's bad already, we don't need to check.
|
|
|
|
* Do this checking under a lock; as multiple PCI devices
|
|
|
|
* in one slot might report errors simultaneously, and we
|
|
|
|
* only want one error recovery routine running.
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
2013-06-20 12:21:03 +07:00
|
|
|
eeh_serialize_lock(&flags);
|
2005-11-04 07:49:23 +07:00
|
|
|
rc = 1;
|
2012-09-08 05:44:13 +07:00
|
|
|
if (pe->state & EEH_PE_ISOLATED) {
|
|
|
|
pe->check_count++;
|
|
|
|
if (pe->check_count % EEH_MAX_FAILS == 0) {
|
2017-08-29 14:34:03 +07:00
|
|
|
dn = pci_device_to_OF_node(dev);
|
|
|
|
if (dn)
|
|
|
|
location = of_get_property(dn, "ibm,loc-code",
|
|
|
|
NULL);
|
2019-08-16 11:48:13 +07:00
|
|
|
eeh_edev_err(edev, "%d reads ignored for recovering device at location=%s driver=%s\n",
|
2015-03-17 12:15:08 +07:00
|
|
|
pe->check_count,
|
|
|
|
location ? location : "unknown",
|
2019-08-16 11:48:13 +07:00
|
|
|
eeh_driver_name(dev));
|
|
|
|
eeh_edev_err(edev, "Might be infinite loop in %s driver\n",
|
2012-01-11 16:09:58 +07:00
|
|
|
eeh_driver_name(dev));
|
2005-11-04 07:49:31 +07:00
|
|
|
dump_stack();
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
2005-11-04 07:49:23 +07:00
|
|
|
goto dn_unlock;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Now test for an EEH failure. This is VERY expensive.
|
|
|
|
* Note that the eeh_config_addr may be a parent device
|
|
|
|
* in the case of a device behind a bridge, or it may be
|
|
|
|
* function zero of a multi-function device.
|
|
|
|
* In any case they must share a common PHB.
|
|
|
|
*/
|
2012-09-08 05:44:13 +07:00
|
|
|
ret = eeh_ops->get_state(pe, NULL);
|
2005-11-04 07:49:15 +07:00
|
|
|
|
2007-03-20 02:51:00 +07:00
|
|
|
/* Note that config-io to empty slots may fail;
|
2012-02-28 03:03:51 +07:00
|
|
|
* they are empty when they don't have children.
|
2012-02-28 03:03:57 +07:00
|
|
|
* We will punt with the following conditions: Failure to get
|
|
|
|
* PE's state, EEH not support and Permanently unavailable
|
|
|
|
* state, PE is in good state.
|
2012-02-28 03:03:51 +07:00
|
|
|
*/
|
2012-02-28 03:03:57 +07:00
|
|
|
if ((ret < 0) ||
|
2018-03-19 09:49:23 +07:00
|
|
|
(ret == EEH_STATE_NOT_SUPPORT) || eeh_state_active(ret)) {
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.false_positives++;
|
2012-09-08 05:44:13 +07:00
|
|
|
pe->false_positives++;
|
2005-11-04 07:49:23 +07:00
|
|
|
rc = 0;
|
|
|
|
goto dn_unlock;
|
2005-11-04 07:49:15 +07:00
|
|
|
}
|
|
|
|
|
2014-05-05 06:29:03 +07:00
|
|
|
/*
|
|
|
|
* It should be corner case that the parent PE has been
|
|
|
|
* put into frozen state as well. We should take care
|
|
|
|
* that at first.
|
|
|
|
*/
|
|
|
|
parent_pe = pe->parent;
|
|
|
|
while (parent_pe) {
|
|
|
|
/* Hit the ceiling ? */
|
|
|
|
if (parent_pe->type & EEH_PE_PHB)
|
|
|
|
break;
|
|
|
|
|
|
|
|
/* Frozen parent PE ? */
|
|
|
|
ret = eeh_ops->get_state(parent_pe, NULL);
|
2018-05-25 10:11:33 +07:00
|
|
|
if (ret > 0 && !eeh_state_active(ret)) {
|
2014-05-05 06:29:03 +07:00
|
|
|
pe = parent_pe;
|
2018-05-25 10:11:33 +07:00
|
|
|
pr_err("EEH: Failure of PHB#%x-PE#%x will be handled at parent PHB#%x-PE#%x.\n",
|
|
|
|
pe->phb->global_number, pe->addr,
|
|
|
|
pe->phb->global_number, parent_pe->addr);
|
|
|
|
}
|
2014-05-05 06:29:03 +07:00
|
|
|
|
|
|
|
/* Next parent level */
|
|
|
|
parent_pe = parent_pe->parent;
|
|
|
|
}
|
|
|
|
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.slot_resets++;
|
2013-06-20 12:20:51 +07:00
|
|
|
|
2005-11-04 07:49:23 +07:00
|
|
|
/* Avoid repeated reports of this failure, including problems
|
|
|
|
* with other functions on this device, and functions under
|
2012-02-28 03:03:51 +07:00
|
|
|
* bridges.
|
|
|
|
*/
|
2018-09-12 08:23:31 +07:00
|
|
|
eeh_pe_mark_isolated(pe);
|
2013-06-20 12:21:03 +07:00
|
|
|
eeh_serialize_unlock(flags);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* Most EEH events are due to device driver bugs. Having
|
|
|
|
* a stack trace will help the device-driver authors figure
|
2012-02-28 03:03:51 +07:00
|
|
|
* out what happened. So print that out.
|
|
|
|
*/
|
2019-09-03 17:15:56 +07:00
|
|
|
pr_debug("EEH: %s: Frozen PHB#%x-PE#%x detected\n",
|
|
|
|
__func__, pe->phb->global_number, pe->addr);
|
2013-09-06 08:00:05 +07:00
|
|
|
eeh_send_failure_event(pe);
|
|
|
|
|
2005-11-04 07:49:23 +07:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
dn_unlock:
|
2013-06-20 12:21:03 +07:00
|
|
|
eeh_serialize_unlock(flags);
|
2005-11-04 07:49:23 +07:00
|
|
|
return rc;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2012-09-08 05:44:22 +07:00
|
|
|
EXPORT_SYMBOL_GPL(eeh_dev_check_failure);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/**
|
2012-02-28 03:03:51 +07:00
|
|
|
* eeh_check_failure - Check if all 1's data is due to EEH slot freeze
|
2014-09-30 09:38:50 +07:00
|
|
|
* @token: I/O address
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
2014-09-30 09:38:50 +07:00
|
|
|
* Check for an EEH failure at the given I/O address. Call this
|
2005-04-17 05:20:36 +07:00
|
|
|
* routine if the result of a read was all 0xff's and you want to
|
2014-09-30 09:38:50 +07:00
|
|
|
* find out if this is due to an EEH slot freeze event. This routine
|
2005-04-17 05:20:36 +07:00
|
|
|
* will query firmware for the EEH status.
|
|
|
|
*
|
|
|
|
* Note this routine is safe to call in an interrupt context.
|
|
|
|
*/
|
2014-09-30 09:38:50 +07:00
|
|
|
int eeh_check_failure(const volatile void __iomem *token)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
|
|
|
unsigned long addr;
|
2012-09-08 05:44:22 +07:00
|
|
|
struct eeh_dev *edev;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* Finding the phys addr + pci device; this is pretty quick. */
|
|
|
|
addr = eeh_token_to_phys((unsigned long __force) token);
|
2012-09-08 05:44:23 +07:00
|
|
|
edev = eeh_addr_cache_get_dev(addr);
|
2012-09-08 05:44:22 +07:00
|
|
|
if (!edev) {
|
2012-02-29 22:47:45 +07:00
|
|
|
eeh_stats.no_device++;
|
2014-09-30 09:38:50 +07:00
|
|
|
return 0;
|
2005-11-04 07:48:52 +07:00
|
|
|
}
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2014-09-30 09:38:50 +07:00
|
|
|
return eeh_dev_check_failure(edev);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(eeh_check_failure);
|
|
|
|
|
2005-11-04 07:50:10 +07:00
|
|
|
|
2006-09-16 06:57:42 +07:00
|
|
|
/**
|
2012-02-28 03:03:52 +07:00
|
|
|
* eeh_pci_enable - Enable MMIO or DMA transfers for this slot
|
2012-09-08 05:44:16 +07:00
|
|
|
* @pe: EEH PE
|
2012-02-28 03:03:51 +07:00
|
|
|
*
|
|
|
|
* This routine should be called to reenable frozen MMIO or DMA
|
|
|
|
* so that it would work correctly again. It's useful while doing
|
|
|
|
* recovery or log collection on the indicated device.
|
2006-09-16 06:57:42 +07:00
|
|
|
*/
|
2012-09-08 05:44:16 +07:00
|
|
|
int eeh_pci_enable(struct eeh_pe *pe, int function)
|
2006-09-16 06:57:42 +07:00
|
|
|
{
|
2014-09-30 09:39:00 +07:00
|
|
|
int active_flag, rc;
|
2014-04-24 15:00:14 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* pHyp doesn't allow to enable IO or DMA on unfrozen PE.
|
|
|
|
* Also, it's pointless to enable them on unfrozen PE. So
|
2014-09-30 09:39:00 +07:00
|
|
|
* we have to check before enabling IO or DMA.
|
2014-04-24 15:00:14 +07:00
|
|
|
*/
|
2014-09-30 09:39:00 +07:00
|
|
|
switch (function) {
|
|
|
|
case EEH_OPT_THAW_MMIO:
|
2015-10-08 10:58:55 +07:00
|
|
|
active_flag = EEH_STATE_MMIO_ACTIVE | EEH_STATE_MMIO_ENABLED;
|
2014-09-30 09:39:00 +07:00
|
|
|
break;
|
|
|
|
case EEH_OPT_THAW_DMA:
|
|
|
|
active_flag = EEH_STATE_DMA_ACTIVE;
|
|
|
|
break;
|
|
|
|
case EEH_OPT_DISABLE:
|
|
|
|
case EEH_OPT_ENABLE:
|
|
|
|
case EEH_OPT_FREEZE_PE:
|
|
|
|
active_flag = 0;
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
pr_warn("%s: Invalid function %d\n",
|
|
|
|
__func__, function);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if IO or DMA has been enabled before
|
|
|
|
* enabling them.
|
|
|
|
*/
|
|
|
|
if (active_flag) {
|
2014-04-24 15:00:14 +07:00
|
|
|
rc = eeh_ops->get_state(pe, NULL);
|
|
|
|
if (rc < 0)
|
|
|
|
return rc;
|
|
|
|
|
2014-09-30 09:39:00 +07:00
|
|
|
/* Needn't enable it at all */
|
|
|
|
if (rc == EEH_STATE_NOT_SUPPORT)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* It's already enabled */
|
|
|
|
if (rc & active_flag)
|
2014-04-24 15:00:14 +07:00
|
|
|
return 0;
|
|
|
|
}
|
2006-09-16 06:57:42 +07:00
|
|
|
|
2014-09-30 09:39:00 +07:00
|
|
|
|
|
|
|
/* Issue the request */
|
2012-09-08 05:44:16 +07:00
|
|
|
rc = eeh_ops->set_option(pe, function);
|
2006-09-16 06:57:42 +07:00
|
|
|
if (rc)
|
2014-04-24 15:00:14 +07:00
|
|
|
pr_warn("%s: Unexpected state change %d on "
|
2016-11-16 10:02:15 +07:00
|
|
|
"PHB#%x-PE#%x, err=%d\n",
|
2014-04-24 15:00:14 +07:00
|
|
|
__func__, function, pe->phb->global_number,
|
|
|
|
pe->addr, rc);
|
2006-09-16 06:57:42 +07:00
|
|
|
|
2014-09-30 09:39:00 +07:00
|
|
|
/* Check if the request is finished successfully */
|
|
|
|
if (active_flag) {
|
2018-09-12 08:23:32 +07:00
|
|
|
rc = eeh_wait_state(pe, PCI_BUS_RESET_WAIT_MSEC);
|
2015-10-23 13:19:46 +07:00
|
|
|
if (rc < 0)
|
2014-09-30 09:39:00 +07:00
|
|
|
return rc;
|
2014-04-24 15:00:14 +07:00
|
|
|
|
2014-09-30 09:39:00 +07:00
|
|
|
if (rc & active_flag)
|
|
|
|
return 0;
|
2014-04-24 15:00:14 +07:00
|
|
|
|
2014-09-30 09:39:00 +07:00
|
|
|
return -EIO;
|
|
|
|
}
|
2007-03-20 02:59:59 +07:00
|
|
|
|
2006-09-16 06:57:42 +07:00
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2019-08-16 11:48:15 +07:00
|
|
|
static void eeh_disable_and_save_dev_state(struct eeh_dev *edev,
|
2018-05-25 10:11:32 +07:00
|
|
|
void *userdata)
|
2015-02-11 06:20:49 +07:00
|
|
|
{
|
|
|
|
struct pci_dev *pdev = eeh_dev_to_pci_dev(edev);
|
|
|
|
struct pci_dev *dev = userdata;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The caller should have disabled and saved the
|
|
|
|
* state for the specified device
|
|
|
|
*/
|
|
|
|
if (!pdev || pdev == dev)
|
2019-08-16 11:48:15 +07:00
|
|
|
return;
|
2015-02-11 06:20:49 +07:00
|
|
|
|
|
|
|
/* Ensure we have D0 power state */
|
|
|
|
pci_set_power_state(pdev, PCI_D0);
|
|
|
|
|
|
|
|
/* Save device state */
|
|
|
|
pci_save_state(pdev);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Disable device to avoid any DMA traffic and
|
|
|
|
* interrupt from the device
|
|
|
|
*/
|
|
|
|
pci_write_config_word(pdev, PCI_COMMAND, PCI_COMMAND_INTX_DISABLE);
|
|
|
|
}
|
|
|
|
|
2019-08-16 11:48:15 +07:00
|
|
|
static void eeh_restore_dev_state(struct eeh_dev *edev, void *userdata)
|
2015-02-11 06:20:49 +07:00
|
|
|
{
|
2015-03-17 12:15:07 +07:00
|
|
|
struct pci_dn *pdn = eeh_dev_to_pdn(edev);
|
2015-02-11 06:20:49 +07:00
|
|
|
struct pci_dev *pdev = eeh_dev_to_pci_dev(edev);
|
|
|
|
struct pci_dev *dev = userdata;
|
|
|
|
|
|
|
|
if (!pdev)
|
2019-08-16 11:48:15 +07:00
|
|
|
return;
|
2015-02-11 06:20:49 +07:00
|
|
|
|
|
|
|
/* Apply customization from firmware */
|
2015-03-17 12:15:07 +07:00
|
|
|
if (pdn && eeh_ops->restore_config)
|
|
|
|
eeh_ops->restore_config(pdn);
|
2015-02-11 06:20:49 +07:00
|
|
|
|
|
|
|
/* The caller should restore state for the specified device */
|
|
|
|
if (pdev != dev)
|
2015-06-03 11:52:59 +07:00
|
|
|
pci_restore_state(pdev);
|
2015-02-11 06:20:49 +07:00
|
|
|
}
|
|
|
|
|
2018-01-05 23:45:46 +07:00
|
|
|
int eeh_restore_vf_config(struct pci_dn *pdn)
|
|
|
|
{
|
|
|
|
struct eeh_dev *edev = pdn_to_eeh_dev(pdn);
|
|
|
|
u32 devctl, cmd, cap2, aer_capctl;
|
|
|
|
int old_mps;
|
|
|
|
|
|
|
|
if (edev->pcie_cap) {
|
|
|
|
/* Restore MPS */
|
|
|
|
old_mps = (ffs(pdn->mps) - 8) << 5;
|
|
|
|
eeh_ops->read_config(pdn, edev->pcie_cap + PCI_EXP_DEVCTL,
|
|
|
|
2, &devctl);
|
|
|
|
devctl &= ~PCI_EXP_DEVCTL_PAYLOAD;
|
|
|
|
devctl |= old_mps;
|
|
|
|
eeh_ops->write_config(pdn, edev->pcie_cap + PCI_EXP_DEVCTL,
|
|
|
|
2, devctl);
|
|
|
|
|
2018-02-07 00:59:40 +07:00
|
|
|
/* Disable Completion Timeout if possible */
|
2018-01-05 23:45:46 +07:00
|
|
|
eeh_ops->read_config(pdn, edev->pcie_cap + PCI_EXP_DEVCAP2,
|
|
|
|
4, &cap2);
|
2018-02-07 00:59:40 +07:00
|
|
|
if (cap2 & PCI_EXP_DEVCAP2_COMP_TMOUT_DIS) {
|
2018-01-05 23:45:46 +07:00
|
|
|
eeh_ops->read_config(pdn,
|
|
|
|
edev->pcie_cap + PCI_EXP_DEVCTL2,
|
|
|
|
4, &cap2);
|
2018-02-07 00:59:40 +07:00
|
|
|
cap2 |= PCI_EXP_DEVCTL2_COMP_TMOUT_DIS;
|
2018-01-05 23:45:46 +07:00
|
|
|
eeh_ops->write_config(pdn,
|
|
|
|
edev->pcie_cap + PCI_EXP_DEVCTL2,
|
|
|
|
4, cap2);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Enable SERR and parity checking */
|
|
|
|
eeh_ops->read_config(pdn, PCI_COMMAND, 2, &cmd);
|
|
|
|
cmd |= (PCI_COMMAND_PARITY | PCI_COMMAND_SERR);
|
|
|
|
eeh_ops->write_config(pdn, PCI_COMMAND, 2, cmd);
|
|
|
|
|
|
|
|
/* Enable report various errors */
|
|
|
|
if (edev->pcie_cap) {
|
|
|
|
eeh_ops->read_config(pdn, edev->pcie_cap + PCI_EXP_DEVCTL,
|
|
|
|
2, &devctl);
|
|
|
|
devctl &= ~PCI_EXP_DEVCTL_CERE;
|
|
|
|
devctl |= (PCI_EXP_DEVCTL_NFERE |
|
|
|
|
PCI_EXP_DEVCTL_FERE |
|
|
|
|
PCI_EXP_DEVCTL_URRE);
|
|
|
|
eeh_ops->write_config(pdn, edev->pcie_cap + PCI_EXP_DEVCTL,
|
|
|
|
2, devctl);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Enable ECRC generation and check */
|
|
|
|
if (edev->pcie_cap && edev->aer_cap) {
|
|
|
|
eeh_ops->read_config(pdn, edev->aer_cap + PCI_ERR_CAP,
|
|
|
|
4, &aer_capctl);
|
|
|
|
aer_capctl |= (PCI_ERR_CAP_ECRC_GENE | PCI_ERR_CAP_ECRC_CHKE);
|
|
|
|
eeh_ops->write_config(pdn, edev->aer_cap + PCI_ERR_CAP,
|
|
|
|
4, aer_capctl);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-05-08 05:04:05 +07:00
|
|
|
/**
|
2016-02-08 10:39:19 +07:00
|
|
|
* pcibios_set_pcie_reset_state - Set PCI-E reset state
|
2012-02-28 03:03:51 +07:00
|
|
|
* @dev: pci device struct
|
|
|
|
* @state: reset state to enter
|
2007-05-08 05:04:05 +07:00
|
|
|
*
|
|
|
|
* Return value:
|
|
|
|
* 0 if success
|
2012-02-28 03:03:51 +07:00
|
|
|
*/
|
2007-05-08 05:04:05 +07:00
|
|
|
int pcibios_set_pcie_reset_state(struct pci_dev *dev, enum pcie_reset_state state)
|
|
|
|
{
|
2012-09-08 05:44:17 +07:00
|
|
|
struct eeh_dev *edev = pci_dev_to_eeh_dev(dev);
|
2014-09-17 09:48:26 +07:00
|
|
|
struct eeh_pe *pe = eeh_dev_to_pe(edev);
|
2012-09-08 05:44:17 +07:00
|
|
|
|
|
|
|
if (!pe) {
|
|
|
|
pr_err("%s: No PE found on PCI device %s\n",
|
|
|
|
__func__, pci_name(dev));
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
2007-05-08 05:04:05 +07:00
|
|
|
|
|
|
|
switch (state) {
|
|
|
|
case pcie_deassert_reset:
|
2012-09-08 05:44:17 +07:00
|
|
|
eeh_ops->reset(pe, EEH_RESET_DEACTIVATE);
|
2018-11-29 10:16:38 +07:00
|
|
|
eeh_unfreeze_pe(pe);
|
2016-03-04 06:53:09 +07:00
|
|
|
if (!(pe->type & EEH_PE_VF))
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, true);
|
2015-02-11 06:20:49 +07:00
|
|
|
eeh_pe_dev_traverse(pe, eeh_restore_dev_state, dev);
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_ISOLATED, true);
|
2007-05-08 05:04:05 +07:00
|
|
|
break;
|
|
|
|
case pcie_hot_reset:
|
2018-09-12 08:23:31 +07:00
|
|
|
eeh_pe_mark_isolated(pe);
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, true);
|
2015-02-11 06:20:49 +07:00
|
|
|
eeh_ops->set_option(pe, EEH_OPT_FREEZE_PE);
|
|
|
|
eeh_pe_dev_traverse(pe, eeh_disable_and_save_dev_state, dev);
|
2016-03-04 06:53:09 +07:00
|
|
|
if (!(pe->type & EEH_PE_VF))
|
|
|
|
eeh_pe_state_mark(pe, EEH_PE_CFG_BLOCKED);
|
2012-09-08 05:44:17 +07:00
|
|
|
eeh_ops->reset(pe, EEH_RESET_HOT);
|
2007-05-08 05:04:05 +07:00
|
|
|
break;
|
|
|
|
case pcie_warm_reset:
|
2018-09-12 08:23:31 +07:00
|
|
|
eeh_pe_mark_isolated(pe);
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, true);
|
2015-02-11 06:20:49 +07:00
|
|
|
eeh_ops->set_option(pe, EEH_OPT_FREEZE_PE);
|
|
|
|
eeh_pe_dev_traverse(pe, eeh_disable_and_save_dev_state, dev);
|
2016-03-04 06:53:09 +07:00
|
|
|
if (!(pe->type & EEH_PE_VF))
|
|
|
|
eeh_pe_state_mark(pe, EEH_PE_CFG_BLOCKED);
|
2012-09-08 05:44:17 +07:00
|
|
|
eeh_ops->reset(pe, EEH_RESET_FUNDAMENTAL);
|
2007-05-08 05:04:05 +07:00
|
|
|
break;
|
|
|
|
default:
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_ISOLATED | EEH_PE_CFG_BLOCKED, true);
|
2007-05-08 05:04:05 +07:00
|
|
|
return -EINVAL;
|
|
|
|
};
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-09-16 06:56:35 +07:00
|
|
|
/**
|
2012-09-08 05:44:17 +07:00
|
|
|
* eeh_set_pe_freset - Check the required reset for the indicated device
|
|
|
|
* @data: EEH device
|
|
|
|
* @flag: return value
|
2012-02-28 03:03:51 +07:00
|
|
|
*
|
|
|
|
* Each device might have its preferred reset type: fundamental or
|
|
|
|
* hot reset. The routine is used to collected the information for
|
|
|
|
* the indicated device and its children so that the bunch of the
|
|
|
|
* devices could be reset properly.
|
|
|
|
*/
|
2019-08-16 11:48:15 +07:00
|
|
|
static void eeh_set_dev_freset(struct eeh_dev *edev, void *flag)
|
2012-02-28 03:03:51 +07:00
|
|
|
{
|
|
|
|
struct pci_dev *dev;
|
2012-09-08 05:44:17 +07:00
|
|
|
unsigned int *freset = (unsigned int *)flag;
|
2005-11-04 07:50:10 +07:00
|
|
|
|
2012-09-08 05:44:17 +07:00
|
|
|
dev = eeh_dev_to_pci_dev(edev);
|
2012-02-28 03:03:51 +07:00
|
|
|
if (dev)
|
|
|
|
*freset |= dev->needs_freset;
|
|
|
|
}
|
|
|
|
|
2018-11-29 10:16:41 +07:00
|
|
|
static void eeh_pe_refreeze_passed(struct eeh_pe *root)
|
|
|
|
{
|
|
|
|
struct eeh_pe *pe;
|
|
|
|
int state;
|
|
|
|
|
|
|
|
eeh_for_each_pe(root, pe) {
|
|
|
|
if (eeh_pe_passed(pe)) {
|
|
|
|
state = eeh_ops->get_state(pe, NULL);
|
|
|
|
if (state &
|
|
|
|
(EEH_STATE_MMIO_ACTIVE | EEH_STATE_MMIO_ENABLED)) {
|
|
|
|
pr_info("EEH: Passed-through PE PHB#%x-PE#%x was thawed by reset, re-freezing for safety.\n",
|
|
|
|
pe->phb->global_number, pe->addr);
|
|
|
|
eeh_pe_set_option(pe, EEH_OPT_FREEZE_PE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2012-02-28 03:03:51 +07:00
|
|
|
/**
|
2016-11-17 12:07:47 +07:00
|
|
|
* eeh_pe_reset_full - Complete a full reset process on the indicated PE
|
2012-09-08 05:44:17 +07:00
|
|
|
* @pe: EEH PE
|
2012-02-28 03:03:51 +07:00
|
|
|
*
|
2016-11-17 12:07:47 +07:00
|
|
|
* This function executes a full reset procedure on a PE, including setting
|
|
|
|
* the appropriate flags, performing a fundamental or hot reset, and then
|
|
|
|
* deactivating the reset status. It is designed to be used within the EEH
|
|
|
|
* subsystem, as opposed to eeh_pe_reset which is exported to drivers and
|
|
|
|
* only performs a single operation at a time.
|
|
|
|
*
|
|
|
|
* This function will attempt to reset a PE three times before failing.
|
2012-02-28 03:03:51 +07:00
|
|
|
*/
|
2018-11-29 10:16:41 +07:00
|
|
|
int eeh_pe_reset_full(struct eeh_pe *pe, bool include_passed)
|
2005-11-04 07:50:10 +07:00
|
|
|
{
|
2016-11-17 12:07:47 +07:00
|
|
|
int reset_state = (EEH_PE_RESET | EEH_PE_CFG_BLOCKED);
|
|
|
|
int type = EEH_RESET_HOT;
|
2011-04-22 16:59:47 +07:00
|
|
|
unsigned int freset = 0;
|
2018-11-29 10:16:42 +07:00
|
|
|
int i, state = 0, ret;
|
2009-07-31 05:42:39 +07:00
|
|
|
|
2016-11-17 12:07:47 +07:00
|
|
|
/*
|
|
|
|
* Determine the type of reset to perform - hot or fundamental.
|
|
|
|
* Hot reset is the default operation, unless any device under the
|
|
|
|
* PE requires a fundamental reset.
|
2013-06-20 12:20:51 +07:00
|
|
|
*/
|
2012-09-08 05:44:17 +07:00
|
|
|
eeh_pe_dev_traverse(pe, eeh_set_dev_freset, &freset);
|
2011-04-22 16:59:47 +07:00
|
|
|
|
|
|
|
if (freset)
|
2016-11-17 12:07:47 +07:00
|
|
|
type = EEH_RESET_FUNDAMENTAL;
|
2006-09-22 06:25:56 +07:00
|
|
|
|
2016-11-17 12:07:47 +07:00
|
|
|
/* Mark the PE as in reset state and block config space accesses */
|
|
|
|
eeh_pe_state_mark(pe, reset_state);
|
2014-11-14 06:47:29 +07:00
|
|
|
|
2016-11-17 12:07:47 +07:00
|
|
|
/* Make three attempts at resetting the bus */
|
2014-11-14 06:47:28 +07:00
|
|
|
for (i = 0; i < 3; i++) {
|
2018-11-29 10:16:41 +07:00
|
|
|
ret = eeh_pe_reset(pe, type, include_passed);
|
2018-11-29 10:16:42 +07:00
|
|
|
if (!ret)
|
|
|
|
ret = eeh_pe_reset(pe, EEH_RESET_DEACTIVATE,
|
|
|
|
include_passed);
|
|
|
|
if (ret) {
|
|
|
|
ret = -EIO;
|
|
|
|
pr_warn("EEH: Failure %d resetting PHB#%x-PE#%x (attempt %d)\n\n",
|
|
|
|
state, pe->phb->global_number, pe->addr, i + 1);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (i)
|
|
|
|
pr_warn("EEH: PHB#%x-PE#%x: Successful reset (attempt %d)\n",
|
|
|
|
pe->phb->global_number, pe->addr, i + 1);
|
2016-11-17 12:07:47 +07:00
|
|
|
|
|
|
|
/* Wait until the PE is in a functioning state */
|
2018-09-12 08:23:32 +07:00
|
|
|
state = eeh_wait_state(pe, PCI_BUS_RESET_WAIT_MSEC);
|
2014-11-14 06:47:28 +07:00
|
|
|
if (state < 0) {
|
2018-11-29 10:16:42 +07:00
|
|
|
pr_warn("EEH: Unrecoverable slot failure on PHB#%x-PE#%x",
|
|
|
|
pe->phb->global_number, pe->addr);
|
2014-11-14 06:47:28 +07:00
|
|
|
ret = -ENOTRECOVERABLE;
|
2016-11-17 12:07:47 +07:00
|
|
|
break;
|
2006-09-22 06:25:56 +07:00
|
|
|
}
|
2018-09-12 08:23:32 +07:00
|
|
|
if (eeh_state_active(state))
|
|
|
|
break;
|
2018-11-29 10:16:42 +07:00
|
|
|
else
|
|
|
|
pr_warn("EEH: PHB#%x-PE#%x: Slot inactive after reset: 0x%x (attempt %d)\n",
|
|
|
|
pe->phb->global_number, pe->addr, state, i + 1);
|
2005-11-04 07:50:10 +07:00
|
|
|
}
|
2005-11-04 07:54:54 +07:00
|
|
|
|
2018-11-29 10:16:41 +07:00
|
|
|
/* Resetting the PE may have unfrozen child PEs. If those PEs have been
|
|
|
|
* (potentially) passed through to a guest, re-freeze them:
|
|
|
|
*/
|
|
|
|
if (!include_passed)
|
|
|
|
eeh_pe_refreeze_passed(pe);
|
|
|
|
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, reset_state, true);
|
2014-11-14 06:47:28 +07:00
|
|
|
return ret;
|
2005-11-04 07:50:10 +07:00
|
|
|
}
|
|
|
|
|
2005-11-04 07:50:17 +07:00
|
|
|
/**
|
2012-02-28 03:03:51 +07:00
|
|
|
* eeh_save_bars - Save device bars
|
2012-02-28 03:04:07 +07:00
|
|
|
* @edev: PCI device associated EEH device
|
2005-11-04 07:50:17 +07:00
|
|
|
*
|
|
|
|
* Save the values of the device bars. Unlike the restore
|
|
|
|
* routine, this routine is *not* recursive. This is because
|
2011-02-25 03:10:18 +07:00
|
|
|
* PCI devices are added individually; but, for the restore,
|
2005-11-04 07:50:17 +07:00
|
|
|
* an entire slot is reset at a time.
|
|
|
|
*/
|
2012-09-08 05:44:21 +07:00
|
|
|
void eeh_save_bars(struct eeh_dev *edev)
|
2005-11-04 07:50:17 +07:00
|
|
|
{
|
2015-03-17 12:15:07 +07:00
|
|
|
struct pci_dn *pdn;
|
2005-11-04 07:50:17 +07:00
|
|
|
int i;
|
|
|
|
|
2015-03-17 12:15:07 +07:00
|
|
|
pdn = eeh_dev_to_pdn(edev);
|
|
|
|
if (!pdn)
|
2005-11-04 07:50:17 +07:00
|
|
|
return;
|
2013-06-20 12:20:51 +07:00
|
|
|
|
2005-11-04 07:50:17 +07:00
|
|
|
for (i = 0; i < 16; i++)
|
2015-03-17 12:15:07 +07:00
|
|
|
eeh_ops->read_config(pdn, i * 4, 4, &edev->config_space[i]);
|
2013-11-12 13:49:21 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* For PCI bridges including root port, we need enable bus
|
|
|
|
* master explicitly. Otherwise, it can't fetch IODA table
|
|
|
|
* entries correctly. So we cache the bit in advance so that
|
|
|
|
* we can restore it after reset, either PHB range or PE range.
|
|
|
|
*/
|
|
|
|
if (edev->mode & EEH_DEV_BRIDGE)
|
|
|
|
edev->config_space[1] |= PCI_COMMAND_MASTER;
|
2005-11-04 07:50:17 +07:00
|
|
|
}
|
|
|
|
|
2012-02-28 03:03:53 +07:00
|
|
|
/**
|
|
|
|
* eeh_ops_register - Register platform dependent EEH operations
|
|
|
|
* @ops: platform dependent EEH operations
|
|
|
|
*
|
|
|
|
* Register the platform dependent EEH operation callback
|
|
|
|
* functions. The platform should call this function before
|
|
|
|
* any other EEH operations.
|
|
|
|
*/
|
|
|
|
int __init eeh_ops_register(struct eeh_ops *ops)
|
|
|
|
{
|
|
|
|
if (!ops->name) {
|
2014-07-17 11:41:41 +07:00
|
|
|
pr_warn("%s: Invalid EEH ops name for %p\n",
|
2012-02-28 03:03:53 +07:00
|
|
|
__func__, ops);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (eeh_ops && eeh_ops != ops) {
|
2014-07-17 11:41:41 +07:00
|
|
|
pr_warn("%s: EEH ops of platform %s already existing (%s)\n",
|
2012-02-28 03:03:53 +07:00
|
|
|
__func__, eeh_ops->name, ops->name);
|
|
|
|
return -EEXIST;
|
|
|
|
}
|
|
|
|
|
|
|
|
eeh_ops = ops;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* eeh_ops_unregister - Unreigster platform dependent EEH operations
|
|
|
|
* @name: name of EEH platform operations
|
|
|
|
*
|
|
|
|
* Unregister the platform dependent EEH operation callback
|
|
|
|
* functions.
|
|
|
|
*/
|
|
|
|
int __exit eeh_ops_unregister(const char *name)
|
|
|
|
{
|
|
|
|
if (!name || !strlen(name)) {
|
2014-07-17 11:41:41 +07:00
|
|
|
pr_warn("%s: Invalid EEH ops name\n",
|
2012-02-28 03:03:53 +07:00
|
|
|
__func__);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (eeh_ops && !strcmp(eeh_ops->name, name)) {
|
|
|
|
eeh_ops = NULL;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
return -EEXIST;
|
|
|
|
}
|
|
|
|
|
2014-02-12 14:24:56 +07:00
|
|
|
static int eeh_reboot_notifier(struct notifier_block *nb,
|
|
|
|
unsigned long action, void *unused)
|
|
|
|
{
|
2014-07-17 11:41:38 +07:00
|
|
|
eeh_clear_flag(EEH_ENABLED);
|
2014-02-12 14:24:56 +07:00
|
|
|
return NOTIFY_DONE;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct notifier_block eeh_reboot_nb = {
|
|
|
|
.notifier_call = eeh_reboot_notifier,
|
|
|
|
};
|
|
|
|
|
2012-02-28 03:03:51 +07:00
|
|
|
/**
|
|
|
|
* eeh_init - EEH initialization
|
|
|
|
*
|
2005-04-17 05:20:36 +07:00
|
|
|
* Initialize EEH by trying to enable it for all of the adapters in the system.
|
|
|
|
* As a side effect we can determine here if eeh is supported at all.
|
|
|
|
* Note that we leave EEH on so failed config cycles won't cause a machine
|
|
|
|
* check. If a user turns off EEH for a particular adapter they are really
|
|
|
|
* telling Linux to ignore errors. Some hardware (e.g. POWER5) won't
|
|
|
|
* grant access to a slot if EEH isn't enabled, and so we always enable
|
|
|
|
* EEH for all slots/all devices.
|
|
|
|
*
|
|
|
|
* The eeh-force-off option disables EEH checking globally, for all slots.
|
|
|
|
* Even if force-off is set, the EEH hardware is still enabled, so that
|
|
|
|
* newer systems can boot.
|
|
|
|
*/
|
2017-09-07 13:35:44 +07:00
|
|
|
static int eeh_init(void)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2012-03-21 04:30:29 +07:00
|
|
|
struct pci_controller *hose, *tmp;
|
2013-06-20 12:20:56 +07:00
|
|
|
int ret = 0;
|
|
|
|
|
2014-02-12 14:24:56 +07:00
|
|
|
/* Register reboot notifier */
|
|
|
|
ret = register_reboot_notifier(&eeh_reboot_nb);
|
|
|
|
if (ret) {
|
|
|
|
pr_warn("%s: Failed to register notifier (%d)\n",
|
|
|
|
__func__, ret);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2012-02-28 03:03:54 +07:00
|
|
|
/* call platform initialization function */
|
|
|
|
if (!eeh_ops) {
|
2014-07-17 11:41:41 +07:00
|
|
|
pr_warn("%s: Platform EEH operation not found\n",
|
2012-02-28 03:03:54 +07:00
|
|
|
__func__);
|
2012-09-08 05:44:02 +07:00
|
|
|
return -EEXIST;
|
2014-11-25 23:10:06 +07:00
|
|
|
} else if ((ret = eeh_ops->init()))
|
2012-09-08 05:44:02 +07:00
|
|
|
return ret;
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2017-09-07 13:35:40 +07:00
|
|
|
/* Initialize PHB PEs */
|
|
|
|
list_for_each_entry_safe(hose, tmp, &hose_list, list_node)
|
|
|
|
eeh_dev_phb_init_dynamic(hose);
|
|
|
|
|
2019-08-16 11:48:08 +07:00
|
|
|
eeh_addr_cache_init();
|
|
|
|
|
2013-06-20 12:21:00 +07:00
|
|
|
/* Initialize EEH event */
|
powerpc/eeh: Cleanup EEH_POSTPONED_PROBE
Currently a flag, EEH_POSTPONED_PROBE, is used to prevent an incorrect
message "EEH: No capable adapters found" from being displayed during
the boot of powernv systems.
It is necessary because, on powernv, the call to eeh_probe_devices()
made from eeh_init() is too early and EEH can't yet be enabled. A
second call is made later from eeh_pnv_post_init(), which succeeds.
(On pseries, the first call succeeds because PCI devices are set up
early enough and no second call is made.)
This can be simplified by moving the early call to eeh_probe_devices()
from eeh_init() (where it's seen by both platforms) to
pSeries_final_fixup(), so that each platform only calls
eeh_probe_devices() once, at a point where it can succeed.
This is slightly later in the boot sequence, but but still early
enough and it is now in the same place in the sequence for both
platforms (the pcibios_fixup hook).
The display of the message can be cleaned up as well, by moving it
into eeh_probe_devices().
Signed-off-by: Sam Bobroff <sbobroff@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
2018-09-12 08:23:23 +07:00
|
|
|
return eeh_event_init();
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2012-09-08 05:44:02 +07:00
|
|
|
core_initcall_sync(eeh_init);
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/**
|
2015-03-17 12:15:08 +07:00
|
|
|
* eeh_add_device_early - Enable EEH for the indicated device node
|
2015-03-17 12:15:06 +07:00
|
|
|
* @pdn: PCI device node for which to set up EEH
|
2005-04-17 05:20:36 +07:00
|
|
|
*
|
|
|
|
* This routine must be used to perform EEH initialization for PCI
|
|
|
|
* devices that were added after system boot (e.g. hotplug, dlpar).
|
|
|
|
* This routine must be called before any i/o is performed to the
|
|
|
|
* adapter (inluding any config-space i/o).
|
|
|
|
* Whether this actually enables EEH or not for this device depends
|
|
|
|
* on the CEC architecture, type of the device, on earlier boot
|
|
|
|
* command-line arguments & etc.
|
|
|
|
*/
|
2015-03-17 12:15:06 +07:00
|
|
|
void eeh_add_device_early(struct pci_dn *pdn)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2017-08-29 14:34:01 +07:00
|
|
|
struct pci_controller *phb = pdn ? pdn->phb : NULL;
|
2015-03-17 12:15:06 +07:00
|
|
|
struct eeh_dev *edev = pdn_to_eeh_dev(pdn);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2016-04-12 02:17:22 +07:00
|
|
|
if (!edev)
|
2005-04-17 05:20:36 +07:00
|
|
|
return;
|
2005-11-04 07:54:23 +07:00
|
|
|
|
2015-05-01 06:22:15 +07:00
|
|
|
if (!eeh_has_flag(EEH_PROBE_MODE_DEVTREE))
|
|
|
|
return;
|
|
|
|
|
2005-11-04 07:54:23 +07:00
|
|
|
/* USB Bus children of PCI devices will not have BUID's */
|
2015-03-17 12:15:06 +07:00
|
|
|
if (NULL == phb ||
|
|
|
|
(eeh_has_flag(EEH_PROBE_MODE_DEVTREE) && 0 == phb->buid))
|
2005-04-17 05:20:36 +07:00
|
|
|
return;
|
|
|
|
|
2015-03-17 12:15:06 +07:00
|
|
|
eeh_ops->probe(pdn, NULL);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2012-02-28 03:03:51 +07:00
|
|
|
/**
|
|
|
|
* eeh_add_device_tree_early - Enable EEH for the indicated device
|
2015-03-17 12:15:06 +07:00
|
|
|
* @pdn: PCI device node
|
2012-02-28 03:03:51 +07:00
|
|
|
*
|
|
|
|
* This routine must be used to perform EEH initialization for the
|
|
|
|
* indicated PCI device that was added after system boot (e.g.
|
|
|
|
* hotplug, dlpar).
|
|
|
|
*/
|
2015-03-17 12:15:06 +07:00
|
|
|
void eeh_add_device_tree_early(struct pci_dn *pdn)
|
2005-11-04 07:51:31 +07:00
|
|
|
{
|
2015-03-17 12:15:06 +07:00
|
|
|
struct pci_dn *n;
|
|
|
|
|
|
|
|
if (!pdn)
|
|
|
|
return;
|
2007-12-21 11:52:07 +07:00
|
|
|
|
2015-03-17 12:15:06 +07:00
|
|
|
list_for_each_entry(n, &pdn->child_list, list)
|
|
|
|
eeh_add_device_tree_early(n);
|
|
|
|
eeh_add_device_early(pdn);
|
2005-11-04 07:51:31 +07:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_add_device_tree_early);
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/**
|
2012-02-28 03:03:51 +07:00
|
|
|
* eeh_add_device_late - Perform EEH initialization for the indicated pci device
|
2005-04-17 05:20:36 +07:00
|
|
|
* @dev: pci device for which to set up EEH
|
|
|
|
*
|
|
|
|
* This routine must be used to complete EEH initialization for PCI
|
|
|
|
* devices that were added after system boot (e.g. hotplug, dlpar).
|
|
|
|
*/
|
2013-07-24 09:24:52 +07:00
|
|
|
void eeh_add_device_late(struct pci_dev *dev)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2015-03-17 12:15:08 +07:00
|
|
|
struct pci_dn *pdn;
|
2012-02-28 03:04:07 +07:00
|
|
|
struct eeh_dev *edev;
|
2005-11-04 07:48:45 +07:00
|
|
|
|
2019-08-16 11:48:09 +07:00
|
|
|
if (!dev)
|
2005-04-17 05:20:36 +07:00
|
|
|
return;
|
|
|
|
|
2015-03-17 12:15:08 +07:00
|
|
|
pdn = pci_get_pdn_by_devfn(dev->bus, dev->devfn);
|
|
|
|
edev = pdn_to_eeh_dev(pdn);
|
2019-08-16 11:48:13 +07:00
|
|
|
eeh_edev_dbg(edev, "Adding device\n");
|
2012-02-28 03:04:07 +07:00
|
|
|
if (edev->pdev == dev) {
|
2019-08-16 11:48:13 +07:00
|
|
|
eeh_edev_dbg(edev, "Device already referenced!\n");
|
2008-10-28 02:48:41 +07:00
|
|
|
return;
|
|
|
|
}
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The EEH cache might not be removed correctly because of
|
|
|
|
* unbalanced kref to the device during unplug time, which
|
|
|
|
* relies on pcibios_release_device(). So we have to remove
|
|
|
|
* that here explicitly.
|
|
|
|
*/
|
|
|
|
if (edev->pdev) {
|
|
|
|
eeh_rmv_from_parent_pe(edev);
|
|
|
|
eeh_addr_cache_rmv_dev(edev->pdev);
|
|
|
|
eeh_sysfs_remove_device(edev->pdev);
|
2013-07-24 09:25:01 +07:00
|
|
|
edev->mode &= ~EEH_DEV_SYSFS;
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
|
2014-01-12 13:13:45 +07:00
|
|
|
/*
|
|
|
|
* We definitely should have the PCI device removed
|
|
|
|
* though it wasn't correctly. So we needn't call
|
|
|
|
* into error handler afterwards.
|
|
|
|
*/
|
|
|
|
edev->mode |= EEH_DEV_NO_HANDLER;
|
|
|
|
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
edev->pdev = NULL;
|
|
|
|
dev->dev.archdata.edev = NULL;
|
|
|
|
}
|
2008-10-28 02:48:41 +07:00
|
|
|
|
2015-08-14 13:03:19 +07:00
|
|
|
if (eeh_has_flag(EEH_PROBE_MODE_DEV))
|
|
|
|
eeh_ops->probe(pdn, NULL);
|
|
|
|
|
2012-02-28 03:04:07 +07:00
|
|
|
edev->pdev = dev;
|
|
|
|
dev->dev.archdata.edev = edev;
|
2005-11-04 07:48:45 +07:00
|
|
|
|
2012-09-08 05:44:23 +07:00
|
|
|
eeh_addr_cache_insert_dev(dev);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
2006-04-01 01:04:52 +07:00
|
|
|
|
2012-02-28 03:03:51 +07:00
|
|
|
/**
|
|
|
|
* eeh_add_device_tree_late - Perform EEH initialization for the indicated PCI bus
|
|
|
|
* @bus: PCI bus
|
|
|
|
*
|
|
|
|
* This routine must be used to perform EEH initialization for PCI
|
|
|
|
* devices which are attached to the indicated PCI bus. The PCI bus
|
|
|
|
* is added after system boot through hotplug or dlpar.
|
|
|
|
*/
|
2006-04-01 01:04:52 +07:00
|
|
|
void eeh_add_device_tree_late(struct pci_bus *bus)
|
|
|
|
{
|
|
|
|
struct pci_dev *dev;
|
|
|
|
|
2019-08-16 11:48:09 +07:00
|
|
|
if (eeh_has_flag(EEH_FORCE_DISABLED))
|
|
|
|
return;
|
2006-04-01 01:04:52 +07:00
|
|
|
list_for_each_entry(dev, &bus->devices, bus_list) {
|
2013-06-20 12:20:51 +07:00
|
|
|
eeh_add_device_late(dev);
|
|
|
|
if (dev->hdr_type == PCI_HEADER_TYPE_BRIDGE) {
|
|
|
|
struct pci_bus *subbus = dev->subordinate;
|
|
|
|
if (subbus)
|
|
|
|
eeh_add_device_tree_late(subbus);
|
|
|
|
}
|
2006-04-01 01:04:52 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_add_device_tree_late);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
2012-12-28 16:13:19 +07:00
|
|
|
/**
|
|
|
|
* eeh_add_sysfs_files - Add EEH sysfs files for the indicated PCI bus
|
|
|
|
* @bus: PCI bus
|
|
|
|
*
|
|
|
|
* This routine must be used to add EEH sysfs files for PCI
|
|
|
|
* devices which are attached to the indicated PCI bus. The PCI bus
|
|
|
|
* is added after system boot through hotplug or dlpar.
|
|
|
|
*/
|
|
|
|
void eeh_add_sysfs_files(struct pci_bus *bus)
|
|
|
|
{
|
|
|
|
struct pci_dev *dev;
|
|
|
|
|
|
|
|
list_for_each_entry(dev, &bus->devices, bus_list) {
|
|
|
|
eeh_sysfs_add_device(dev);
|
|
|
|
if (dev->hdr_type == PCI_HEADER_TYPE_BRIDGE) {
|
|
|
|
struct pci_bus *subbus = dev->subordinate;
|
|
|
|
if (subbus)
|
|
|
|
eeh_add_sysfs_files(subbus);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_add_sysfs_files);
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
/**
|
2012-02-28 03:03:51 +07:00
|
|
|
* eeh_remove_device - Undo EEH setup for the indicated pci device
|
2005-04-17 05:20:36 +07:00
|
|
|
* @dev: pci device to be removed
|
|
|
|
*
|
2006-04-01 01:04:52 +07:00
|
|
|
* This routine should be called when a device is removed from
|
|
|
|
* a running system (e.g. by hotplug or dlpar). It unregisters
|
|
|
|
* the PCI device from the EEH subsystem. I/O errors affecting
|
|
|
|
* this device will no longer be detected after this call; thus,
|
|
|
|
* i/o errors affecting this slot may leave this device unusable.
|
2005-04-17 05:20:36 +07:00
|
|
|
*/
|
2013-07-24 09:24:55 +07:00
|
|
|
void eeh_remove_device(struct pci_dev *dev)
|
2005-04-17 05:20:36 +07:00
|
|
|
{
|
2012-02-28 03:04:07 +07:00
|
|
|
struct eeh_dev *edev;
|
|
|
|
|
2014-02-12 14:24:55 +07:00
|
|
|
if (!dev || !eeh_enabled())
|
2005-04-17 05:20:36 +07:00
|
|
|
return;
|
2012-02-28 03:04:07 +07:00
|
|
|
edev = pci_dev_to_eeh_dev(dev);
|
2005-04-17 05:20:36 +07:00
|
|
|
|
|
|
|
/* Unregister the device with the EEH/PCI address search system */
|
2019-08-16 11:48:13 +07:00
|
|
|
dev_dbg(&dev->dev, "EEH: Removing device\n");
|
2005-11-04 07:48:45 +07:00
|
|
|
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
if (!edev || !edev->pdev || !edev->pe) {
|
2019-08-16 11:48:13 +07:00
|
|
|
dev_dbg(&dev->dev, "EEH: Device not referenced!\n");
|
2008-10-28 02:48:41 +07:00
|
|
|
return;
|
2006-04-07 03:41:41 +07:00
|
|
|
}
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* During the hotplug for EEH error recovery, we need the EEH
|
|
|
|
* device attached to the parent PE in order for BAR restore
|
|
|
|
* a bit later. So we keep it for BAR restore and remove it
|
|
|
|
* from the parent PE during the BAR resotre.
|
|
|
|
*/
|
2012-02-28 03:04:07 +07:00
|
|
|
edev->pdev = NULL;
|
2016-03-04 06:53:11 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The flag "in_error" is used to trace EEH devices for VFs
|
|
|
|
* in error state or not. It's set in eeh_report_error(). If
|
|
|
|
* it's not set, eeh_report_{reset,resume}() won't be called
|
|
|
|
* for the VF EEH device.
|
|
|
|
*/
|
|
|
|
edev->in_error = false;
|
2012-02-28 03:04:07 +07:00
|
|
|
dev->dev.archdata.edev = NULL;
|
powerpc/eeh: Use partial hotplug for EEH unaware drivers
When EEH error happens to one specific PE, some devices with drivers
supporting EEH won't except hotplug on the device. However, there
might have other deivces without driver, or with driver without EEH
support. For the case, we need do partial hotplug in order to make
sure that the PE becomes absolutely quite during reset. Otherise,
the PE reset might fail and leads to failure of error recovery.
The current code doesn't handle that 'mixed' case properly, it either
uses the error callbacks to the drivers, or tries hotplug, but doesn't
handle a PE (EEH domain) composed of a combination of the two.
The patch intends to support so-called "partial" hotplug for EEH:
Before we do reset, we stop and remove those PCI devices without
EEH sensitive driver. The corresponding EEH devices are not detached
from its PE, but with special flag. After the reset is done, those
EEH devices with the special flag will be scanned one by one.
Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2013-07-24 09:24:58 +07:00
|
|
|
if (!(edev->pe->state & EEH_PE_KEEP))
|
|
|
|
eeh_rmv_from_parent_pe(edev);
|
|
|
|
else
|
|
|
|
edev->mode |= EEH_DEV_DISCONNECTED;
|
2008-10-28 02:48:41 +07:00
|
|
|
|
2014-01-12 13:13:45 +07:00
|
|
|
/*
|
|
|
|
* We're removing from the PCI subsystem, that means
|
|
|
|
* the PCI device driver can't support EEH or not
|
|
|
|
* well. So we rely on hotplug completely to do recovery
|
|
|
|
* for the specific PCI device.
|
|
|
|
*/
|
|
|
|
edev->mode |= EEH_DEV_NO_HANDLER;
|
|
|
|
|
2012-09-08 05:44:23 +07:00
|
|
|
eeh_addr_cache_rmv_dev(dev);
|
2008-10-28 02:48:41 +07:00
|
|
|
eeh_sysfs_remove_device(dev);
|
2013-07-24 09:25:01 +07:00
|
|
|
edev->mode &= ~EEH_DEV_SYSFS;
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
2018-11-29 10:16:38 +07:00
|
|
|
int eeh_unfreeze_pe(struct eeh_pe *pe)
|
2014-09-30 09:39:01 +07:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
|
|
|
|
if (ret) {
|
|
|
|
pr_warn("%s: Failure %d enabling IO on PHB#%x-PE#%x\n",
|
|
|
|
__func__, ret, pe->phb->global_number, pe->addr);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = eeh_pci_enable(pe, EEH_OPT_THAW_DMA);
|
|
|
|
if (ret) {
|
|
|
|
pr_warn("%s: Failure %d enabling DMA on PHB#%x-PE#%x\n",
|
|
|
|
__func__, ret, pe->phb->global_number, pe->addr);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2014-09-30 09:39:07 +07:00
|
|
|
|
|
|
|
static struct pci_device_id eeh_reset_ids[] = {
|
|
|
|
{ PCI_DEVICE(0x19a2, 0x0710) }, /* Emulex, BE */
|
|
|
|
{ PCI_DEVICE(0x10df, 0xe220) }, /* Emulex, Lancer */
|
2014-11-14 06:47:30 +07:00
|
|
|
{ PCI_DEVICE(0x14e4, 0x1657) }, /* Broadcom BCM5719 */
|
2014-09-30 09:39:07 +07:00
|
|
|
{ 0 }
|
|
|
|
};
|
|
|
|
|
|
|
|
static int eeh_pe_change_owner(struct eeh_pe *pe)
|
|
|
|
{
|
|
|
|
struct eeh_dev *edev, *tmp;
|
|
|
|
struct pci_dev *pdev;
|
|
|
|
struct pci_device_id *id;
|
2018-03-19 09:49:23 +07:00
|
|
|
int ret;
|
2014-09-30 09:39:07 +07:00
|
|
|
|
|
|
|
/* Check PE state */
|
|
|
|
ret = eeh_ops->get_state(pe, NULL);
|
|
|
|
if (ret < 0 || ret == EEH_STATE_NOT_SUPPORT)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* Unfrozen PE, nothing to do */
|
2018-03-19 09:49:23 +07:00
|
|
|
if (eeh_state_active(ret))
|
2014-09-30 09:39:07 +07:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* Frozen PE, check if it needs PE level reset */
|
|
|
|
eeh_pe_for_each_dev(pe, edev, tmp) {
|
|
|
|
pdev = eeh_dev_to_pci_dev(edev);
|
|
|
|
if (!pdev)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
for (id = &eeh_reset_ids[0]; id->vendor != 0; id++) {
|
|
|
|
if (id->vendor != PCI_ANY_ID &&
|
|
|
|
id->vendor != pdev->vendor)
|
|
|
|
continue;
|
|
|
|
if (id->device != PCI_ANY_ID &&
|
|
|
|
id->device != pdev->device)
|
|
|
|
continue;
|
|
|
|
if (id->subvendor != PCI_ANY_ID &&
|
|
|
|
id->subvendor != pdev->subsystem_vendor)
|
|
|
|
continue;
|
|
|
|
if (id->subdevice != PCI_ANY_ID &&
|
|
|
|
id->subdevice != pdev->subsystem_device)
|
|
|
|
continue;
|
|
|
|
|
2016-04-27 08:14:53 +07:00
|
|
|
return eeh_pe_reset_and_recover(pe);
|
2014-09-30 09:39:07 +07:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-11-29 10:16:38 +07:00
|
|
|
ret = eeh_unfreeze_pe(pe);
|
|
|
|
if (!ret)
|
2018-11-29 10:16:39 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_ISOLATED, true);
|
2018-11-29 10:16:38 +07:00
|
|
|
return ret;
|
2014-09-30 09:39:07 +07:00
|
|
|
}
|
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/**
|
|
|
|
* eeh_dev_open - Increase count of pass through devices for PE
|
|
|
|
* @pdev: PCI device
|
|
|
|
*
|
|
|
|
* Increase count of passed through devices for the indicated
|
|
|
|
* PE. In the result, the EEH errors detected on the PE won't be
|
|
|
|
* reported. The PE owner will be responsible for detection
|
|
|
|
* and recovery.
|
|
|
|
*/
|
|
|
|
int eeh_dev_open(struct pci_dev *pdev)
|
|
|
|
{
|
|
|
|
struct eeh_dev *edev;
|
2014-09-30 09:38:54 +07:00
|
|
|
int ret = -ENODEV;
|
2014-06-10 08:41:56 +07:00
|
|
|
|
|
|
|
mutex_lock(&eeh_dev_mutex);
|
|
|
|
|
|
|
|
/* No PCI device ? */
|
|
|
|
if (!pdev)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* No EEH device or PE ? */
|
|
|
|
edev = pci_dev_to_eeh_dev(pdev);
|
|
|
|
if (!edev || !edev->pe)
|
|
|
|
goto out;
|
|
|
|
|
2014-09-30 09:38:54 +07:00
|
|
|
/*
|
|
|
|
* The PE might have been put into frozen state, but we
|
|
|
|
* didn't detect that yet. The passed through PCI devices
|
|
|
|
* in frozen PE won't work properly. Clear the frozen state
|
|
|
|
* in advance.
|
|
|
|
*/
|
2014-09-30 09:39:07 +07:00
|
|
|
ret = eeh_pe_change_owner(edev->pe);
|
2014-09-30 09:39:01 +07:00
|
|
|
if (ret)
|
|
|
|
goto out;
|
2014-09-30 09:38:54 +07:00
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/* Increase PE's pass through count */
|
|
|
|
atomic_inc(&edev->pe->pass_dev_cnt);
|
|
|
|
mutex_unlock(&eeh_dev_mutex);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
out:
|
|
|
|
mutex_unlock(&eeh_dev_mutex);
|
2014-09-30 09:38:54 +07:00
|
|
|
return ret;
|
2014-06-10 08:41:56 +07:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_dev_open);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* eeh_dev_release - Decrease count of pass through devices for PE
|
|
|
|
* @pdev: PCI device
|
|
|
|
*
|
|
|
|
* Decrease count of pass through devices for the indicated PE. If
|
|
|
|
* there is no passed through device in PE, the EEH errors detected
|
|
|
|
* on the PE will be reported and handled as usual.
|
|
|
|
*/
|
|
|
|
void eeh_dev_release(struct pci_dev *pdev)
|
|
|
|
{
|
|
|
|
struct eeh_dev *edev;
|
|
|
|
|
|
|
|
mutex_lock(&eeh_dev_mutex);
|
|
|
|
|
|
|
|
/* No PCI device ? */
|
|
|
|
if (!pdev)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* No EEH device ? */
|
|
|
|
edev = pci_dev_to_eeh_dev(pdev);
|
|
|
|
if (!edev || !edev->pe || !eeh_pe_passed(edev->pe))
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* Decrease PE's pass through count */
|
2015-08-27 12:58:27 +07:00
|
|
|
WARN_ON(atomic_dec_if_positive(&edev->pe->pass_dev_cnt) < 0);
|
2014-09-30 09:39:07 +07:00
|
|
|
eeh_pe_change_owner(edev->pe);
|
2014-06-10 08:41:56 +07:00
|
|
|
out:
|
|
|
|
mutex_unlock(&eeh_dev_mutex);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(eeh_dev_release);
|
|
|
|
|
2014-08-05 15:52:59 +07:00
|
|
|
#ifdef CONFIG_IOMMU_API
|
|
|
|
|
2014-07-15 14:00:56 +07:00
|
|
|
static int dev_has_iommu_table(struct device *dev, void *data)
|
|
|
|
{
|
|
|
|
struct pci_dev *pdev = to_pci_dev(dev);
|
|
|
|
struct pci_dev **ppdev = data;
|
|
|
|
|
|
|
|
if (!dev)
|
|
|
|
return 0;
|
|
|
|
|
2018-11-30 20:23:19 +07:00
|
|
|
if (device_iommu_mapped(dev)) {
|
2014-07-15 14:00:56 +07:00
|
|
|
*ppdev = pdev;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/**
|
|
|
|
* eeh_iommu_group_to_pe - Convert IOMMU group to EEH PE
|
|
|
|
* @group: IOMMU group
|
|
|
|
*
|
|
|
|
* The routine is called to convert IOMMU group to EEH PE.
|
|
|
|
*/
|
|
|
|
struct eeh_pe *eeh_iommu_group_to_pe(struct iommu_group *group)
|
|
|
|
{
|
|
|
|
struct pci_dev *pdev = NULL;
|
|
|
|
struct eeh_dev *edev;
|
2014-07-15 14:00:56 +07:00
|
|
|
int ret;
|
2014-06-10 08:41:56 +07:00
|
|
|
|
|
|
|
/* No IOMMU group ? */
|
|
|
|
if (!group)
|
|
|
|
return NULL;
|
|
|
|
|
2014-07-15 14:00:56 +07:00
|
|
|
ret = iommu_group_for_each_dev(group, &pdev, dev_has_iommu_table);
|
|
|
|
if (!ret || !pdev)
|
2014-06-10 08:41:56 +07:00
|
|
|
return NULL;
|
|
|
|
|
|
|
|
/* No EEH device or PE ? */
|
|
|
|
edev = pci_dev_to_eeh_dev(pdev);
|
|
|
|
if (!edev || !edev->pe)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
return edev->pe;
|
|
|
|
}
|
2014-08-07 09:47:16 +07:00
|
|
|
EXPORT_SYMBOL_GPL(eeh_iommu_group_to_pe);
|
2014-06-10 08:41:56 +07:00
|
|
|
|
2014-08-05 15:52:59 +07:00
|
|
|
#endif /* CONFIG_IOMMU_API */
|
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/**
|
|
|
|
* eeh_pe_set_option - Set options for the indicated PE
|
|
|
|
* @pe: EEH PE
|
|
|
|
* @option: requested option
|
|
|
|
*
|
|
|
|
* The routine is called to enable or disable EEH functionality
|
|
|
|
* on the indicated PE, to enable IO or DMA for the frozen PE.
|
|
|
|
*/
|
|
|
|
int eeh_pe_set_option(struct eeh_pe *pe, int option)
|
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
/* Invalid PE ? */
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* EEH functionality could possibly be disabled, just
|
|
|
|
* return error for the case. And the EEH functinality
|
|
|
|
* isn't expected to be disabled on one specific PE.
|
|
|
|
*/
|
|
|
|
switch (option) {
|
|
|
|
case EEH_OPT_ENABLE:
|
2014-09-30 09:39:01 +07:00
|
|
|
if (eeh_enabled()) {
|
2014-09-30 09:39:07 +07:00
|
|
|
ret = eeh_pe_change_owner(pe);
|
2014-06-10 08:41:56 +07:00
|
|
|
break;
|
2014-09-30 09:39:01 +07:00
|
|
|
}
|
2014-06-10 08:41:56 +07:00
|
|
|
ret = -EIO;
|
|
|
|
break;
|
|
|
|
case EEH_OPT_DISABLE:
|
|
|
|
break;
|
|
|
|
case EEH_OPT_THAW_MMIO:
|
|
|
|
case EEH_OPT_THAW_DMA:
|
2016-09-28 11:34:53 +07:00
|
|
|
case EEH_OPT_FREEZE_PE:
|
2014-06-10 08:41:56 +07:00
|
|
|
if (!eeh_ops || !eeh_ops->set_option) {
|
|
|
|
ret = -ENOENT;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2014-09-30 09:39:01 +07:00
|
|
|
ret = eeh_pci_enable(pe, option);
|
2014-06-10 08:41:56 +07:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
pr_debug("%s: Option %d out of range (%d, %d)\n",
|
|
|
|
__func__, option, EEH_OPT_DISABLE, EEH_OPT_THAW_DMA);
|
|
|
|
ret = -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_pe_set_option);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* eeh_pe_get_state - Retrieve PE's state
|
|
|
|
* @pe: EEH PE
|
|
|
|
*
|
|
|
|
* Retrieve the PE's state, which includes 3 aspects: enabled
|
|
|
|
* DMA, enabled IO and asserted reset.
|
|
|
|
*/
|
|
|
|
int eeh_pe_get_state(struct eeh_pe *pe)
|
|
|
|
{
|
|
|
|
int result, ret = 0;
|
|
|
|
bool rst_active, dma_en, mmio_en;
|
|
|
|
|
|
|
|
/* Existing PE ? */
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
if (!eeh_ops || !eeh_ops->get_state)
|
|
|
|
return -ENOENT;
|
|
|
|
|
2016-03-04 06:53:14 +07:00
|
|
|
/*
|
|
|
|
* If the parent PE is owned by the host kernel and is undergoing
|
|
|
|
* error recovery, we should return the PE state as temporarily
|
|
|
|
* unavailable so that the error recovery on the guest is suspended
|
|
|
|
* until the recovery completes on the host.
|
|
|
|
*/
|
|
|
|
if (pe->parent &&
|
|
|
|
!(pe->state & EEH_PE_REMOVED) &&
|
|
|
|
(pe->parent->state & (EEH_PE_ISOLATED | EEH_PE_RECOVERING)))
|
|
|
|
return EEH_PE_STATE_UNAVAIL;
|
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
result = eeh_ops->get_state(pe, NULL);
|
|
|
|
rst_active = !!(result & EEH_STATE_RESET_ACTIVE);
|
|
|
|
dma_en = !!(result & EEH_STATE_DMA_ENABLED);
|
|
|
|
mmio_en = !!(result & EEH_STATE_MMIO_ENABLED);
|
|
|
|
|
|
|
|
if (rst_active)
|
|
|
|
ret = EEH_PE_STATE_RESET;
|
|
|
|
else if (dma_en && mmio_en)
|
|
|
|
ret = EEH_PE_STATE_NORMAL;
|
|
|
|
else if (!dma_en && !mmio_en)
|
|
|
|
ret = EEH_PE_STATE_STOPPED_IO_DMA;
|
|
|
|
else if (!dma_en && mmio_en)
|
|
|
|
ret = EEH_PE_STATE_STOPPED_DMA;
|
|
|
|
else
|
|
|
|
ret = EEH_PE_STATE_UNAVAIL;
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_pe_get_state);
|
|
|
|
|
2018-11-29 10:16:41 +07:00
|
|
|
static int eeh_pe_reenable_devices(struct eeh_pe *pe, bool include_passed)
|
2014-09-30 09:38:53 +07:00
|
|
|
{
|
|
|
|
struct eeh_dev *edev, *tmp;
|
|
|
|
struct pci_dev *pdev;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
eeh_pe_restore_bars(pe);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Reenable PCI devices as the devices passed
|
|
|
|
* through are always enabled before the reset.
|
|
|
|
*/
|
|
|
|
eeh_pe_for_each_dev(pe, edev, tmp) {
|
|
|
|
pdev = eeh_dev_to_pci_dev(edev);
|
|
|
|
if (!pdev)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
ret = pci_reenable_device(pdev);
|
|
|
|
if (ret) {
|
|
|
|
pr_warn("%s: Failure %d reenabling %s\n",
|
|
|
|
__func__, ret, pci_name(pdev));
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* The PE is still in frozen state */
|
2018-11-29 10:16:41 +07:00
|
|
|
if (include_passed || !eeh_pe_passed(pe)) {
|
|
|
|
ret = eeh_unfreeze_pe(pe);
|
|
|
|
} else
|
|
|
|
pr_info("EEH: Note: Leaving passthrough PHB#%x-PE#%x frozen.\n",
|
|
|
|
pe->phb->global_number, pe->addr);
|
2018-11-29 10:16:38 +07:00
|
|
|
if (!ret)
|
2018-11-29 10:16:41 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_ISOLATED, include_passed);
|
2018-11-29 10:16:38 +07:00
|
|
|
return ret;
|
2014-09-30 09:38:53 +07:00
|
|
|
}
|
|
|
|
|
2016-11-17 12:07:47 +07:00
|
|
|
|
2014-06-10 08:41:56 +07:00
|
|
|
/**
|
|
|
|
* eeh_pe_reset - Issue PE reset according to specified type
|
|
|
|
* @pe: EEH PE
|
|
|
|
* @option: reset type
|
|
|
|
*
|
|
|
|
* The routine is called to reset the specified PE with the
|
|
|
|
* indicated type, either fundamental reset or hot reset.
|
|
|
|
* PE reset is the most important part for error recovery.
|
|
|
|
*/
|
2018-11-29 10:16:41 +07:00
|
|
|
int eeh_pe_reset(struct eeh_pe *pe, int option, bool include_passed)
|
2014-06-10 08:41:56 +07:00
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
/* Invalid PE ? */
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
if (!eeh_ops || !eeh_ops->set_option || !eeh_ops->reset)
|
|
|
|
return -ENOENT;
|
|
|
|
|
|
|
|
switch (option) {
|
|
|
|
case EEH_RESET_DEACTIVATE:
|
|
|
|
ret = eeh_ops->reset(pe, option);
|
2018-11-29 10:16:41 +07:00
|
|
|
eeh_pe_state_clear(pe, EEH_PE_CFG_BLOCKED, include_passed);
|
2014-06-10 08:41:56 +07:00
|
|
|
if (ret)
|
|
|
|
break;
|
|
|
|
|
2018-11-29 10:16:41 +07:00
|
|
|
ret = eeh_pe_reenable_devices(pe, include_passed);
|
2014-06-10 08:41:56 +07:00
|
|
|
break;
|
|
|
|
case EEH_RESET_HOT:
|
|
|
|
case EEH_RESET_FUNDAMENTAL:
|
2014-09-30 09:38:52 +07:00
|
|
|
/*
|
|
|
|
* Proactively freeze the PE to drop all MMIO access
|
|
|
|
* during reset, which should be banned as it's always
|
|
|
|
* cause recursive EEH error.
|
|
|
|
*/
|
|
|
|
eeh_ops->set_option(pe, EEH_OPT_FREEZE_PE);
|
|
|
|
|
2014-10-01 14:07:50 +07:00
|
|
|
eeh_pe_state_mark(pe, EEH_PE_CFG_BLOCKED);
|
2014-06-10 08:41:56 +07:00
|
|
|
ret = eeh_ops->reset(pe, option);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
pr_debug("%s: Unsupported option %d\n",
|
|
|
|
__func__, option);
|
|
|
|
ret = -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_pe_reset);
|
|
|
|
|
|
|
|
/**
|
|
|
|
* eeh_pe_configure - Configure PCI bridges after PE reset
|
|
|
|
* @pe: EEH PE
|
|
|
|
*
|
|
|
|
* The routine is called to restore the PCI config space for
|
|
|
|
* those PCI devices, especially PCI bridges affected by PE
|
|
|
|
* reset issued previously.
|
|
|
|
*/
|
|
|
|
int eeh_pe_configure(struct eeh_pe *pe)
|
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
/* Invalid PE ? */
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_pe_configure);
|
|
|
|
|
2015-03-26 12:42:08 +07:00
|
|
|
/**
|
|
|
|
* eeh_pe_inject_err - Injecting the specified PCI error to the indicated PE
|
|
|
|
* @pe: the indicated PE
|
|
|
|
* @type: error type
|
|
|
|
* @function: error function
|
|
|
|
* @addr: address
|
|
|
|
* @mask: address mask
|
|
|
|
*
|
|
|
|
* The routine is called to inject the specified PCI error, which
|
|
|
|
* is determined by @type and @function, to the indicated PE for
|
|
|
|
* testing purpose.
|
|
|
|
*/
|
|
|
|
int eeh_pe_inject_err(struct eeh_pe *pe, int type, int func,
|
|
|
|
unsigned long addr, unsigned long mask)
|
|
|
|
{
|
|
|
|
/* Invalid PE ? */
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
/* Unsupported operation ? */
|
|
|
|
if (!eeh_ops || !eeh_ops->err_inject)
|
|
|
|
return -ENOENT;
|
|
|
|
|
|
|
|
/* Check on PCI error type */
|
|
|
|
if (type != EEH_ERR_TYPE_32 && type != EEH_ERR_TYPE_64)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
/* Check on PCI error function */
|
|
|
|
if (func < EEH_ERR_FUNC_MIN || func > EEH_ERR_FUNC_MAX)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
return eeh_ops->err_inject(pe, type, func, addr, mask);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(eeh_pe_inject_err);
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
static int proc_eeh_show(struct seq_file *m, void *v)
|
|
|
|
{
|
2014-02-12 14:24:55 +07:00
|
|
|
if (!eeh_enabled()) {
|
2005-04-17 05:20:36 +07:00
|
|
|
seq_printf(m, "EEH Subsystem is globally disabled\n");
|
2012-02-29 22:47:45 +07:00
|
|
|
seq_printf(m, "eeh_total_mmio_ffs=%llu\n", eeh_stats.total_mmio_ffs);
|
2005-04-17 05:20:36 +07:00
|
|
|
} else {
|
|
|
|
seq_printf(m, "EEH Subsystem is enabled\n");
|
2005-11-04 07:48:52 +07:00
|
|
|
seq_printf(m,
|
2012-02-29 22:47:45 +07:00
|
|
|
"no device=%llu\n"
|
|
|
|
"no device node=%llu\n"
|
|
|
|
"no config address=%llu\n"
|
|
|
|
"check not wanted=%llu\n"
|
|
|
|
"eeh_total_mmio_ffs=%llu\n"
|
|
|
|
"eeh_false_positives=%llu\n"
|
|
|
|
"eeh_slot_resets=%llu\n",
|
|
|
|
eeh_stats.no_device,
|
|
|
|
eeh_stats.no_dn,
|
|
|
|
eeh_stats.no_cfg_addr,
|
|
|
|
eeh_stats.ignored_check,
|
|
|
|
eeh_stats.total_mmio_ffs,
|
|
|
|
eeh_stats.false_positives,
|
|
|
|
eeh_stats.slot_resets);
|
2005-04-17 05:20:36 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-04-24 15:00:18 +07:00
|
|
|
#ifdef CONFIG_DEBUG_FS
|
|
|
|
static int eeh_enable_dbgfs_set(void *data, u64 val)
|
|
|
|
{
|
|
|
|
if (val)
|
2014-07-17 11:41:38 +07:00
|
|
|
eeh_clear_flag(EEH_FORCE_DISABLED);
|
2014-04-24 15:00:18 +07:00
|
|
|
else
|
2014-07-17 11:41:38 +07:00
|
|
|
eeh_add_flag(EEH_FORCE_DISABLED);
|
2014-04-24 15:00:18 +07:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int eeh_enable_dbgfs_get(void *data, u64 *val)
|
|
|
|
{
|
|
|
|
if (eeh_enabled())
|
|
|
|
*val = 0x1ul;
|
|
|
|
else
|
|
|
|
*val = 0x0ul;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-12-20 09:42:51 +07:00
|
|
|
DEFINE_DEBUGFS_ATTRIBUTE(eeh_enable_dbgfs_ops, eeh_enable_dbgfs_get,
|
|
|
|
eeh_enable_dbgfs_set, "0x%llx\n");
|
2019-02-15 07:48:17 +07:00
|
|
|
|
|
|
|
static ssize_t eeh_force_recover_write(struct file *filp,
|
|
|
|
const char __user *user_buf,
|
|
|
|
size_t count, loff_t *ppos)
|
|
|
|
{
|
|
|
|
struct pci_controller *hose;
|
|
|
|
uint32_t phbid, pe_no;
|
|
|
|
struct eeh_pe *pe;
|
|
|
|
char buf[20];
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
ret = simple_write_to_buffer(buf, sizeof(buf), ppos, user_buf, count);
|
|
|
|
if (!ret)
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* When PE is NULL the event is a "special" event. Rather than
|
|
|
|
* recovering a specific PE it forces the EEH core to scan for failed
|
|
|
|
* PHBs and recovers each. This needs to be done before any device
|
|
|
|
* recoveries can occur.
|
|
|
|
*/
|
|
|
|
if (!strncmp(buf, "hwcheck", 7)) {
|
|
|
|
__eeh_send_failure_event(NULL);
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = sscanf(buf, "%x:%x", &phbid, &pe_no);
|
|
|
|
if (ret != 2)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
hose = pci_find_controller_for_domain(phbid);
|
|
|
|
if (!hose)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
/* Retrieve PE */
|
|
|
|
pe = eeh_pe_get(hose, pe_no, 0);
|
|
|
|
if (!pe)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We don't do any state checking here since the detection
|
|
|
|
* process is async to the recovery process. The recovery
|
|
|
|
* thread *should* not break even if we schedule a recovery
|
|
|
|
* from an odd state (e.g. PE removed, or recovery of a
|
|
|
|
* non-isolated PE)
|
|
|
|
*/
|
|
|
|
__eeh_send_failure_event(pe);
|
|
|
|
|
|
|
|
return ret < 0 ? ret : count;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const struct file_operations eeh_force_recover_fops = {
|
|
|
|
.open = simple_open,
|
|
|
|
.llseek = no_llseek,
|
|
|
|
.write = eeh_force_recover_write,
|
|
|
|
};
|
2019-09-03 17:16:03 +07:00
|
|
|
|
|
|
|
static ssize_t eeh_debugfs_dev_usage(struct file *filp,
|
|
|
|
char __user *user_buf,
|
|
|
|
size_t count, loff_t *ppos)
|
|
|
|
{
|
|
|
|
static const char usage[] = "input format: <domain>:<bus>:<dev>.<fn>\n";
|
|
|
|
|
|
|
|
return simple_read_from_buffer(user_buf, count, ppos,
|
|
|
|
usage, sizeof(usage) - 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t eeh_dev_check_write(struct file *filp,
|
|
|
|
const char __user *user_buf,
|
|
|
|
size_t count, loff_t *ppos)
|
|
|
|
{
|
|
|
|
uint32_t domain, bus, dev, fn;
|
|
|
|
struct pci_dev *pdev;
|
|
|
|
struct eeh_dev *edev;
|
|
|
|
char buf[20];
|
|
|
|
int ret;
|
|
|
|
|
2019-09-03 17:16:04 +07:00
|
|
|
memset(buf, 0, sizeof(buf));
|
|
|
|
ret = simple_write_to_buffer(buf, sizeof(buf)-1, ppos, user_buf, count);
|
2019-09-03 17:16:03 +07:00
|
|
|
if (!ret)
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
ret = sscanf(buf, "%x:%x:%x.%x", &domain, &bus, &dev, &fn);
|
|
|
|
if (ret != 4) {
|
|
|
|
pr_err("%s: expected 4 args, got %d\n", __func__, ret);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
pdev = pci_get_domain_bus_and_slot(domain, bus, (dev << 3) | fn);
|
|
|
|
if (!pdev)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
edev = pci_dev_to_eeh_dev(pdev);
|
|
|
|
if (!edev) {
|
|
|
|
pci_err(pdev, "No eeh_dev for this device!\n");
|
|
|
|
pci_dev_put(pdev);
|
|
|
|
return -ENODEV;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = eeh_dev_check_failure(edev);
|
|
|
|
pci_info(pdev, "eeh_dev_check_failure(%04x:%02x:%02x.%01x) = %d\n",
|
|
|
|
domain, bus, dev, fn, ret);
|
|
|
|
|
|
|
|
pci_dev_put(pdev);
|
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const struct file_operations eeh_dev_check_fops = {
|
|
|
|
.open = simple_open,
|
|
|
|
.llseek = no_llseek,
|
|
|
|
.write = eeh_dev_check_write,
|
|
|
|
.read = eeh_debugfs_dev_usage,
|
|
|
|
};
|
|
|
|
|
2019-09-03 17:16:04 +07:00
|
|
|
static int eeh_debugfs_break_device(struct pci_dev *pdev)
|
|
|
|
{
|
|
|
|
struct resource *bar = NULL;
|
|
|
|
void __iomem *mapped;
|
|
|
|
u16 old, bit;
|
|
|
|
int i, pos;
|
|
|
|
|
|
|
|
/* Do we have an MMIO BAR to disable? */
|
|
|
|
for (i = 0; i <= PCI_STD_RESOURCE_END; i++) {
|
|
|
|
struct resource *r = &pdev->resource[i];
|
|
|
|
|
|
|
|
if (!r->flags || !r->start)
|
|
|
|
continue;
|
|
|
|
if (r->flags & IORESOURCE_IO)
|
|
|
|
continue;
|
|
|
|
if (r->flags & IORESOURCE_UNSET)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
bar = r;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!bar) {
|
|
|
|
pci_err(pdev, "Unable to find Memory BAR to cause EEH with\n");
|
|
|
|
return -ENXIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
pci_err(pdev, "Going to break: %pR\n", bar);
|
|
|
|
|
|
|
|
if (pdev->is_virtfn) {
|
|
|
|
#ifndef CONFIG_IOV
|
|
|
|
return -ENXIO;
|
|
|
|
#else
|
|
|
|
/*
|
|
|
|
* VFs don't have a per-function COMMAND register, so the best
|
|
|
|
* we can do is clear the Memory Space Enable bit in the PF's
|
|
|
|
* SRIOV control reg.
|
|
|
|
*
|
|
|
|
* Unfortunately, this requires that we have a PF (i.e doesn't
|
|
|
|
* work for a passed-through VF) and it has the potential side
|
|
|
|
* effect of also causing an EEH on every other VF under the
|
|
|
|
* PF. Oh well.
|
|
|
|
*/
|
|
|
|
pdev = pdev->physfn;
|
|
|
|
if (!pdev)
|
|
|
|
return -ENXIO; /* passed through VFs have no PF */
|
|
|
|
|
|
|
|
pos = pci_find_ext_capability(pdev, PCI_EXT_CAP_ID_SRIOV);
|
|
|
|
pos += PCI_SRIOV_CTRL;
|
|
|
|
bit = PCI_SRIOV_CTRL_MSE;
|
|
|
|
#endif /* !CONFIG_IOV */
|
|
|
|
} else {
|
|
|
|
bit = PCI_COMMAND_MEMORY;
|
|
|
|
pos = PCI_COMMAND;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Process here is:
|
|
|
|
*
|
|
|
|
* 1. Disable Memory space.
|
|
|
|
*
|
|
|
|
* 2. Perform an MMIO to the device. This should result in an error
|
|
|
|
* (CA / UR) being raised by the device which results in an EEH
|
|
|
|
* PE freeze. Using the in_8() accessor skips the eeh detection hook
|
|
|
|
* so the freeze hook so the EEH Detection machinery won't be
|
|
|
|
* triggered here. This is to match the usual behaviour of EEH
|
|
|
|
* where the HW will asyncronously freeze a PE and it's up to
|
|
|
|
* the kernel to notice and deal with it.
|
|
|
|
*
|
|
|
|
* 3. Turn Memory space back on. This is more important for VFs
|
|
|
|
* since recovery will probably fail if we don't. For normal
|
|
|
|
* the COMMAND register is reset as a part of re-initialising
|
|
|
|
* the device.
|
|
|
|
*
|
|
|
|
* Breaking stuff is the point so who cares if it's racy ;)
|
|
|
|
*/
|
|
|
|
pci_read_config_word(pdev, pos, &old);
|
|
|
|
|
|
|
|
mapped = ioremap(bar->start, PAGE_SIZE);
|
|
|
|
if (!mapped) {
|
|
|
|
pci_err(pdev, "Unable to map MMIO BAR %pR\n", bar);
|
|
|
|
return -ENXIO;
|
|
|
|
}
|
|
|
|
|
|
|
|
pci_write_config_word(pdev, pos, old & ~bit);
|
|
|
|
in_8(mapped);
|
|
|
|
pci_write_config_word(pdev, pos, old);
|
|
|
|
|
|
|
|
iounmap(mapped);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t eeh_dev_break_write(struct file *filp,
|
|
|
|
const char __user *user_buf,
|
|
|
|
size_t count, loff_t *ppos)
|
|
|
|
{
|
|
|
|
uint32_t domain, bus, dev, fn;
|
|
|
|
struct pci_dev *pdev;
|
|
|
|
char buf[20];
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
memset(buf, 0, sizeof(buf));
|
|
|
|
ret = simple_write_to_buffer(buf, sizeof(buf)-1, ppos, user_buf, count);
|
|
|
|
if (!ret)
|
|
|
|
return -EFAULT;
|
|
|
|
|
|
|
|
ret = sscanf(buf, "%x:%x:%x.%x", &domain, &bus, &dev, &fn);
|
|
|
|
if (ret != 4) {
|
|
|
|
pr_err("%s: expected 4 args, got %d\n", __func__, ret);
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
pdev = pci_get_domain_bus_and_slot(domain, bus, (dev << 3) | fn);
|
|
|
|
if (!pdev)
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
ret = eeh_debugfs_break_device(pdev);
|
|
|
|
pci_dev_put(pdev);
|
|
|
|
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
static const struct file_operations eeh_dev_break_fops = {
|
|
|
|
.open = simple_open,
|
|
|
|
.llseek = no_llseek,
|
|
|
|
.write = eeh_dev_break_write,
|
|
|
|
.read = eeh_debugfs_dev_usage,
|
|
|
|
};
|
|
|
|
|
2014-04-24 15:00:18 +07:00
|
|
|
#endif
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
static int __init eeh_init_proc(void)
|
|
|
|
{
|
2014-04-24 15:00:18 +07:00
|
|
|
if (machine_is(pseries) || machine_is(powernv)) {
|
2018-05-15 20:57:23 +07:00
|
|
|
proc_create_single("powerpc/eeh", 0, NULL, proc_eeh_show);
|
2014-04-24 15:00:18 +07:00
|
|
|
#ifdef CONFIG_DEBUG_FS
|
2018-12-20 09:42:51 +07:00
|
|
|
debugfs_create_file_unsafe("eeh_enable", 0600,
|
|
|
|
powerpc_debugfs_root, NULL,
|
|
|
|
&eeh_enable_dbgfs_ops);
|
2019-02-15 07:48:11 +07:00
|
|
|
debugfs_create_u32("eeh_max_freezes", 0600,
|
|
|
|
powerpc_debugfs_root, &eeh_max_freezes);
|
2019-02-15 07:48:16 +07:00
|
|
|
debugfs_create_bool("eeh_disable_recovery", 0600,
|
|
|
|
powerpc_debugfs_root,
|
|
|
|
&eeh_debugfs_no_recover);
|
2019-09-03 17:16:03 +07:00
|
|
|
debugfs_create_file_unsafe("eeh_dev_check", 0600,
|
|
|
|
powerpc_debugfs_root, NULL,
|
|
|
|
&eeh_dev_check_fops);
|
2019-09-03 17:16:04 +07:00
|
|
|
debugfs_create_file_unsafe("eeh_dev_break", 0600,
|
|
|
|
powerpc_debugfs_root, NULL,
|
|
|
|
&eeh_dev_break_fops);
|
2019-02-15 07:48:17 +07:00
|
|
|
debugfs_create_file_unsafe("eeh_force_recover", 0600,
|
|
|
|
powerpc_debugfs_root, NULL,
|
|
|
|
&eeh_force_recover_fops);
|
2019-02-15 07:48:13 +07:00
|
|
|
eeh_cache_debugfs_init();
|
2014-04-24 15:00:18 +07:00
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2005-04-17 05:20:36 +07:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
__initcall(eeh_init_proc);
|