linuxppc-dev.lists.ozlabs.org archive mirror
 help / color / mirror / Atom feed
* [PATCH v3 0/5] EEH improvement
@ 2014-02-25  7:28 Gavin Shan
  2014-02-25  7:28 ` [PATCH 1/5] powerpc/eeh: Remove EEH_PE_PHB_DEAD Gavin Shan
                   ` (4 more replies)
  0 siblings, 5 replies; 7+ messages in thread
From: Gavin Shan @ 2014-02-25  7:28 UTC (permalink / raw)
  To: linuxppc-dev; +Cc: Gavin Shan

The series of patches intends to improve reliability of EEH on PowerNV
platform. First all, we have had multiple duplicate states (flags) for
PHB and PE, so we remove those duplicate states to simplify the code.
Besides, we had corrupted PHB diag-data for case of frozen PE. In order
to solve the problem, we introduce eeh_ops->event() and notifications
are sent from EEH core to (PowerNV) platform on creating or destroying
PE instance so that we can allocate or free PHB diag-data backend. Then
we cache the PHB diag-data on the first call to eeh_ops->get_state()
and dump it afterwards, which helps to get correct PHB diag-data.

With the patchset applied, we never dump PHB diag-data for INF errors.
Instead, we just maintain statistics in /proc/powerpc/eeh_inf_err. Also,
we changed the PHB diag-data dump format for a bit to have multiple
fields per line and omits the line with all zero'd fields as Ben suggested.

v2 -> v3:
	* We don't cache the PHB diag-data, instead we just grab and
	  dump PHB diag-data on the first catch-up to avoid broken
	  PHB diag-data.
v1 -> v2:
	* Amending commit logs
	* Support eeh_ops->event() and maintain PHB diag-data on basis
	  of PE instance
	* When dumping PHB diag-data, to replace "-" with "00000000" and
	  omit the line if the fields of it are all zeros.

---

arch/powerpc/include/asm/eeh.h            |    1 -
arch/powerpc/kernel/eeh.c                 |   10 +---
arch/powerpc/kernel/eeh_driver.c          |   10 ++--
arch/powerpc/platforms/powernv/eeh-ioda.c |  137 ++++++++++++++++++++--------------------------
arch/powerpc/platforms/powernv/pci.c      |  228 ++++++++++++++++++++++++++++++++++++++++++---------------------------
arch/powerpc/platforms/powernv/pci.h      |    8 +--
6 files changed, 195 insertions(+), 199 deletions(-)

Thanks,
Gavin

^ permalink raw reply	[flat|nested] 7+ messages in thread
* [PATCH 1/5] powerpc/eeh: Remove EEH_PE_PHB_DEAD
@ 2014-02-21 11:53 Gavin Shan
  2014-02-21 11:53 ` [PATCH 2/5] powerpc/powernv: Remove PNV_EEH_STATE_REMOVED Gavin Shan
  0 siblings, 1 reply; 7+ messages in thread
From: Gavin Shan @ 2014-02-21 11:53 UTC (permalink / raw)
  To: linuxppc-dev; +Cc: Gavin Shan

The PE state (for eeh_pe instance) EEH_PE_PHB_DEAD is duplicate to
EEH_PE_ISOLATED. Originally, those PHB (PHB PE) with EEH_PE_PHB_DEAD
would be removed from the system. However, it's safe to replace
that with EEH_PE_ISOLATED.

The patch also clear EEH_PE_RECOVERING after fenced PHB has been handled,
either failure or success. It makes the PHB PE state consistent with:

	PHB functions normally		  NONE
	PHB has been removed		  EEH_PE_ISOLATED
	PHB fenced, recovery in progress  EEH_PE_ISOLATED | RECOVERING

Signed-off-by: Gavin Shan <shangw@linux.vnet.ibm.com>
---
 arch/powerpc/include/asm/eeh.h   |    1 -
 arch/powerpc/kernel/eeh.c        |   10 ++--------
 arch/powerpc/kernel/eeh_driver.c |   10 +++++-----
 3 files changed, 7 insertions(+), 14 deletions(-)

diff --git a/arch/powerpc/include/asm/eeh.h b/arch/powerpc/include/asm/eeh.h
index d4dd41f..a61b06f 100644
--- a/arch/powerpc/include/asm/eeh.h
+++ b/arch/powerpc/include/asm/eeh.h
@@ -53,7 +53,6 @@ struct device_node;
 
 #define EEH_PE_ISOLATED		(1 << 0)	/* Isolated PE		*/
 #define EEH_PE_RECOVERING	(1 << 1)	/* Recovering PE	*/
-#define EEH_PE_PHB_DEAD		(1 << 2)	/* Dead PHB		*/
 
 #define EEH_PE_KEEP		(1 << 8)	/* Keep PE on hotplug	*/
 
diff --git a/arch/powerpc/kernel/eeh.c b/arch/powerpc/kernel/eeh.c
index e7b76a6..f167676 100644
--- a/arch/powerpc/kernel/eeh.c
+++ b/arch/powerpc/kernel/eeh.c
@@ -232,7 +232,6 @@ void eeh_slot_error_detail(struct eeh_pe *pe, int severity)
 {
 	size_t loglen = 0;
 	struct eeh_dev *edev, *tmp;
-	bool valid_cfg_log = true;
 
 	/*
 	 * When the PHB is fenced or dead, it's pointless to collect
@@ -240,12 +239,7 @@ void eeh_slot_error_detail(struct eeh_pe *pe, int severity)
 	 * 0xFF's. For ER, we still retrieve the data from the PCI
 	 * config space.
 	 */
-	if (eeh_probe_mode_dev() &&
-	    (pe->type & EEH_PE_PHB) &&
-	    (pe->state & (EEH_PE_ISOLATED | EEH_PE_PHB_DEAD)))
-		valid_cfg_log = false;
-
-	if (valid_cfg_log) {
+	if (!(pe->type & EEH_PE_PHB)) {
 		eeh_pci_enable(pe, EEH_OPT_THAW_MMIO);
 		eeh_ops->configure_bridge(pe);
 		eeh_pe_restore_bars(pe);
@@ -309,7 +303,7 @@ static int eeh_phb_check_failure(struct eeh_pe *pe)
 
 	/* If the PHB has been in problematic state */
 	eeh_serialize_lock(&flags);
-	if (phb_pe->state & (EEH_PE_ISOLATED | EEH_PE_PHB_DEAD)) {
+	if (phb_pe->state & EEH_PE_ISOLATED) {
 		ret = 0;
 		goto out;
 	}
diff --git a/arch/powerpc/kernel/eeh_driver.c b/arch/powerpc/kernel/eeh_driver.c
index 7bb30dc..5f50f9c 100644
--- a/arch/powerpc/kernel/eeh_driver.c
+++ b/arch/powerpc/kernel/eeh_driver.c
@@ -661,8 +661,7 @@ static void eeh_handle_special_event(void)
 				phb_pe = eeh_phb_pe_get(hose);
 				if (!phb_pe) continue;
 
-				eeh_pe_state_mark(phb_pe,
-					EEH_PE_ISOLATED | EEH_PE_PHB_DEAD);
+				eeh_pe_state_mark(phb_pe, EEH_PE_ISOLATED);
 			}
 
 			eeh_serialize_unlock(flags);
@@ -678,8 +677,7 @@ static void eeh_handle_special_event(void)
 			eeh_remove_event(pe);
 
 			if (rc == EEH_NEXT_ERR_DEAD_PHB)
-				eeh_pe_state_mark(pe,
-					EEH_PE_ISOLATED | EEH_PE_PHB_DEAD);
+				eeh_pe_state_mark(pe, EEH_PE_ISOLATED);
 			else
 				eeh_pe_state_mark(pe,
 					EEH_PE_ISOLATED | EEH_PE_RECOVERING);
@@ -703,12 +701,14 @@ static void eeh_handle_special_event(void)
 		if (rc == EEH_NEXT_ERR_FROZEN_PE ||
 		    rc == EEH_NEXT_ERR_FENCED_PHB) {
 			eeh_handle_normal_event(pe);
+			eeh_pe_state_clear(pe, EEH_PE_RECOVERING);
 		} else {
 			pci_lock_rescan_remove();
 			list_for_each_entry(hose, &hose_list, list_node) {
 				phb_pe = eeh_phb_pe_get(hose);
 				if (!phb_pe ||
-				    !(phb_pe->state & EEH_PE_PHB_DEAD))
+				    !(phb_pe->state & EEH_PE_ISOLATED) ||
+				    (phb_pe->state & EEH_PE_RECOVERING))
 					continue;
 
 				/* Notify all devices to be down */
-- 
1.7.10.4

^ permalink raw reply related	[flat|nested] 7+ messages in thread
end of thread, other threads:[~2014-02-25  7:28 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2014-02-25  7:28 [PATCH v3 0/5] EEH improvement Gavin Shan
2014-02-25  7:28 ` [PATCH 1/5] powerpc/eeh: Remove EEH_PE_PHB_DEAD Gavin Shan
2014-02-25  7:28 ` [PATCH 2/5] powerpc/powernv: Remove PNV_EEH_STATE_REMOVED Gavin Shan
2014-02-25  7:28 ` [PATCH 3/5] powerpc/powernv: Move PNV_EEH_STATE_ENABLED around Gavin Shan
2014-02-25  7:28 ` [PATCH 4/5] powerpc/powernv: Dump PHB diag-data immediately Gavin Shan
2014-02-25  7:28 ` [PATCH 5/5] powerpc/powernv: Refactor PHB diag-data dump Gavin Shan
  -- strict thread matches above, loose matches on Subject: below --
2014-02-21 11:53 [PATCH 1/5] powerpc/eeh: Remove EEH_PE_PHB_DEAD Gavin Shan
2014-02-21 11:53 ` [PATCH 2/5] powerpc/powernv: Remove PNV_EEH_STATE_REMOVED Gavin Shan

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).