Skip to contentSkip to conversation

SCSI fixes for 2.6.32-rc8

PATCH In review James Bottomley 7 participants 13 messages Nov 30, 2009 – Dec 9, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 SCSI fixes for 2.6.32-rc8

This is a set of miscellaneous (but small) fixes for SCSI.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Alexey Kuznetsov (1):
fix crash when disconnecting usb storage

James Bottomley (1):
fix async scan add/remove race resulting in an oops

Martin K. Petersen (1):
sd: Return correct error code for DIF

The diffstat is:

 drivers/scsi/hosts.c       |    2 -
 drivers/scsi/scsi_scan.c   |   18 ++----------
 drivers/scsi/scsi_sysfs.c  |   63 +++++++++++++++++++--------------------------
 drivers/scsi/sd_dif.c      |    2 -
 include/scsi/scsi_device.h |    1 
 5 files changed, 34 insertions(+), 52 deletions(-)

And the full patch is attached below.

James

---

drivers/scsi/hosts.c +1 -1
index 5fd2da4..c968cc3 100644
@@ -164,8 +164,8 @@ void scsi_remove_host(struct Scsi_Host *shost)
 			return;
 		}
 	spin_unlock_irqrestore(shost->host_lock, flags);
-	mutex_unlock(&shost->scan_mutex);
 	scsi_forget_host(shost);
+	mutex_unlock(&shost->scan_mutex);
 	scsi_proc_host_rm(shost);
 
 	spin_lock_irqsave(shost->host_lock, flags);
drivers/scsi/scsi_scan.c +4 -14
index 0547a7f..47291bc 100644
@@ -952,16 +952,6 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result,
 	return SCSI_SCAN_LUN_PRESENT;
 }
 
-static inline void scsi_destroy_sdev(struct scsi_device *sdev)
-{
-	scsi_device_set_state(sdev, SDEV_DEL);
-	if (sdev->host->hostt->slave_destroy)
-		sdev->host->hostt->slave_destroy(sdev);
-	transport_destroy_device(&sdev->sdev_gendev);
-	put_device(&sdev->sdev_dev);
-	put_device(&sdev->sdev_gendev);
-}
-
 #ifdef CONFIG_SCSI_LOGGING
 /** 
  * scsi_inq_str - print INQUIRY data from min to max index, strip trailing whitespace
@@ -1139,7 +1129,7 @@ static int scsi_probe_and_add_lun(struct scsi_target *starget,
 			}
 		}
 	} else
-		scsi_destroy_sdev(sdev);
+		__scsi_remove_device(sdev);
  out:
 	return res;
 }
@@ -1500,7 +1490,7 @@ static int scsi_report_lun_scan(struct scsi_target *starget, int bflags,
 		/*
 		 * the sdev we used didn't appear in the report luns scan
 		 */
-		scsi_destroy_sdev(sdev);
+		__scsi_remove_device(sdev);
 	return ret;
 }
 
@@ -1710,7 +1700,7 @@ static void scsi_sysfs_add_devices(struct Scsi_Host *shost)
 	shost_for_each_device(sdev, shost) {
 		if (!scsi_host_scan_allowed(shost) ||
 		    scsi_sysfs_add_sdev(sdev) != 0)
-			scsi_destroy_sdev(sdev);
+			__scsi_remove_device(sdev);
 	}
 }
 
@@ -1943,7 +1933,7 @@ void scsi_free_host_dev(struct scsi_device *sdev)
 {
 	BUG_ON(sdev->id != sdev->host->this_id);
 
-	scsi_destroy_sdev(sdev);
+	__scsi_remove_device(sdev);
 }
 EXPORT_SYMBOL(scsi_free_host_dev);
 
drivers/scsi/scsi_sysfs.c +27 -36

109 lines not on this page · raw message

drivers/scsi/sd_dif.c +1 -1
index 88da977..84be621 100644
@@ -418,7 +418,7 @@ error:
 		  __func__, virt, phys, be32_to_cpu(sdt->ref_tag),
 		  be16_to_cpu(sdt->app_tag));
 
-	return -EIO;
+	return -EILSEQ;
 }
 
 /*
include/scsi/scsi_device.h +1 -0
index 9af48cb..f097ae3 100644
@@ -145,6 +145,7 @@ struct scsi_device {
 	unsigned retry_hwerror:1;	/* Retry HARDWARE_ERROR */
 	unsigned last_sector_bug:1;	/* do not use multisector accesses on
 					   SD_LAST_BUGGY_SECTORS */
+	unsigned is_visible:1;	/* is the device visible in sysfs */
 
 	DECLARE_BITMAP(supported_events, SDEV_EVT_MAXBITS); /* supported events */
 	struct list_head event_list;	/* asserted events */
#1 lore raw
Quoted 17 lines · On Mon, 30 Nov 2009 17:29:54 -0500 James Bottomley <[email protected]>…

This is a set of miscellaneous (but small) fixes for SCSI.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Alexey Kuznetsov (1):
fix crash when disconnecting usb storage

James Bottomley (1):
fix async scan add/remove race resulting in an oops

Martin K. Petersen (1):
sd: Return correct error code for DIF

Quoted 8 lines

The diffstat is:

 drivers/scsi/hosts.c       |    2 -
 drivers/scsi/scsi_scan.c   |   18 ++----------
 drivers/scsi/scsi_sysfs.c  |   63 +++++++++++++++++++--------------------------
 drivers/scsi/sd_dif.c      |    2 -
 include/scsi/scsi_device.h |    1 
 5 files changed, 34 insertions(+), 52 deletions(-)

Can you add the tape driver fix?

http://marc.info/?l=linux-scsi&m=125919508329412&w=2

#2 lore raw
Quoted 22 lines · FUJITA Tomonori
Quoted 19 lines · On Mon, 30 Nov 2009 17:29:54 -0500 James Bottomley <[email protected]>…

This is a set of miscellaneous (but small) fixes for SCSI.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Alexey Kuznetsov (1):
fix crash when disconnecting usb storage

James Bottomley (1):
fix async scan add/remove race resulting in an oops

Martin K. Petersen (1):
sd: Return correct error code for DIF

The diffstat is:

Quoted 8 lines
Quoted 6 lines
 drivers/scsi/hosts.c       |    2 -
 drivers/scsi/scsi_scan.c   |   18 ++----------
 drivers/scsi/scsi_sysfs.c  |   63 +++++++++++++++++++--------------------------
 drivers/scsi/sd_dif.c      |    2 -
 include/scsi/scsi_device.h |    1 
 5 files changed, 34 insertions(+), 52 deletions(-)

Can you add the tape driver fix?

Sure ... can I have an acked-by from Kai?

Thanks,

James

#3 lore raw

If we are doing another small batch of SCSI fixes before .32 I'd like to see Mike Christie's error handling fix go in as well. I just noticed it ended up being queued in scsi-misc and not scsi-rc-fixes. It really is an important bug fix. ad63082626f99651d261ccd8698ce4e997362f7e

Signature · 1 line
Martin K. Petersen	Oracle Linux Engineering
Acked-by
#4 lore raw
In reply to James Bottomley
Quoted 30 lines · James Bottomley
Quoted 29 lines · FUJITA Tomonori
Quoted 25 lines · On Mon, 30 Nov 2009 17:29:54 -0500 James Bottomley <[email protected]>…

This is a set of miscellaneous (but small) fixes for SCSI.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Alexey Kuznetsov (1):
fix crash when disconnecting usb storage

James Bottomley (1):
fix async scan add/remove race resulting in an oops

Martin K. Petersen (1):
sd: Return correct error code for DIF

The diffstat is:

 drivers/scsi/hosts.c       |    2 -
 drivers/scsi/scsi_scan.c   |   18 ++----------
 drivers/scsi/scsi_sysfs.c  |   63 +++++++++++++++++++--------------------------
 drivers/scsi/sd_dif.c      |    2 -
 include/scsi/scsi_device.h |    1 
 5 files changed, 34 insertions(+), 52 deletions(-)
Quoted 4 lines
Quoted 1 line

Can you add the tape driver fix?

Sure ... can I have an acked-by from Kai?

There was already tested-by but here is

Thanks,
Kai

#5 lore raw
In reply to James Bottomley
Quoted 17 lines · James Bottomley

This is a set of miscellaneous (but small) fixes for SCSI.

The patch is available here:

master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6.git

The short changelog is:

Alexey Kuznetsov (1):
fix crash when disconnecting usb storage

James Bottomley (1):
fix async scan add/remove race resulting in an oops

Martin K. Petersen (1):
sd: Return correct error code for DIF

Quoted 8 lines

The diffstat is:

 drivers/scsi/hosts.c       |    2 -
 drivers/scsi/scsi_scan.c   |   18 ++----------
 drivers/scsi/scsi_sysfs.c  |   63 +++++++++++++++++++--------------------------
 drivers/scsi/sd_dif.c      |    2 -
 include/scsi/scsi_device.h |    1 
 5 files changed, 34 insertions(+), 52 deletions(-)

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Thanks,
Kleber

Signature · 2 lines
Kleber S. Souza
IBM Linux Technology Center
#6 lore raw
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI:
Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

---
 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +0 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int ipr_reset_restore_cfg_space(struct

ipr_cmnd *ipr_cmd)

        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
        if (rc != PCIBIOS_SUCCESSFUL) {
Signature · 3 lines
-- 
Kleber S. Souza
IBM Linux Technology Center
#7 lore raw
Quoted 68 lines · On Tue, 01 Dec 2009 12:13:47 -0500 James Bottomley <[email protected]>…
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI: Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

---
 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +1 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int ipr_reset_restore_cfg_space(struct
 ipr_cmnd *ipr_cmd)
        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
Quoted 1 line
        if (rc != PCIBIOS_SUCCESSFUL) {

Rafael may have input here, but it seems like we need a low level save/restore routine that ignores the flag (which is generally used for suspend/resume I think?).

Maybe adding low level _pci_save_state/_pci_restore_state that don't check/set the flags would help?

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#8 lore raw
Quoted 71 lines · Jesse Barnes
Quoted 68 lines · On Tue, 01 Dec 2009 12:13:47 -0500 James Bottomley <[email protected]>…
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI: Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

---
 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +1 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int ipr_reset_restore_cfg_space(struct
 ipr_cmnd *ipr_cmd)
        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
Quoted 5 lines
Quoted 1 line
        if (rc != PCIBIOS_SUCCESSFUL) {

Rafael may have input here, but it seems like we need a low level save/restore routine that ignores the flag (which is generally used for suspend/resume I think?).

There are some other users, but they are only a few.

Quoted 2 lines

Maybe adding low level _pci_save_state/_pci_restore_state that don't check/set the flags would help?

That might work, but how do we know that the state we're going to restore is actually valid at this particular point?

Perhaps we need a version using a separate storage area?

Rafael

#9 lore raw
Quoted 82 lines · On Tue, 1 Dec 2009 22:40:48 +0100 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 73 lines · Jesse Barnes
Quoted 70 lines · On Tue, 01 Dec 2009 12:13:47 -0500 James Bottomley <[email protected]>…
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI: Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

Signed-off-by: Kleber Sacilotto de Souza
<[email protected]> ---

 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +1 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int
 ipr_reset_restore_cfg_space(struct ipr_cmnd *ipr_cmd)
        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
Quoted 5 lines
Quoted 1 line
        if (rc != PCIBIOS_SUCCESSFUL) {

Rafael may have input here, but it seems like we need a low level save/restore routine that ignores the flag (which is generally used for suspend/resume I think?).

There are some other users, but they are only a few.

Quoted 7 lines
Quoted 2 lines

Maybe adding low level _pci_save_state/_pci_restore_state that don't check/set the flags would help?

That might work, but how do we know that the state we're going to restore is actually valid at this particular point?

Perhaps we need a version using a separate storage area?

Yeah, that would probably be best. Let the caller allocate the space and save/restore it all it wants for special cases like error handling.

Signature · 1 line
Jesse Barnes, Intel Open Source Technology Center
#10 lore raw
Quoted 88 lines · Jesse Barnes
Quoted 85 lines · On Tue, 1 Dec 2009 22:40:48 +0100 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 73 lines · Jesse Barnes
Quoted 70 lines · On Tue, 01 Dec 2009 12:13:47 -0500 James Bottomley <[email protected]>…
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI: Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

Signed-off-by: Kleber Sacilotto de Souza
<[email protected]> ---

 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +1 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int
 ipr_reset_restore_cfg_space(struct ipr_cmnd *ipr_cmd)
        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
Quoted 5 lines
Quoted 1 line
        if (rc != PCIBIOS_SUCCESSFUL) {

Rafael may have input here, but it seems like we need a low level save/restore routine that ignores the flag (which is generally used for suspend/resume I think?).

There are some other users, but they are only a few.

Quoted 2 lines

Maybe adding low level _pci_save_state/_pci_restore_state that don't check/set the flags would help?

Quoted 7 lines
Quoted 4 lines

That might work, but how do we know that the state we're going to restore is actually valid at this particular point?

Perhaps we need a version using a separate storage area?

Yeah, that would probably be best. Let the caller allocate the space and save/restore it all it wants for special cases like error handling.

OK, so could I have a resolution for this, please, guys?

Do I just apply the patch to fiddle with the internal state which looks ugly but will fix the bug, or are you going to provide us with the correct interface to use?

James

#11 lore raw
Quoted 94 lines · James Bottomley
Quoted 93 lines · Jesse Barnes
Quoted 89 lines · On Tue, 1 Dec 2009 22:40:48 +0100 "Rafael J. Wysocki" <[email protected]> wrote:
Quoted 73 lines · Jesse Barnes
Quoted 70 lines · On Tue, 01 Dec 2009 12:13:47 -0500 James Bottomley <[email protected]>…
Quoted 2 lines · Kleber Sacilotto de Souza

Can you please add the patch from "[PATCH] ipr: fix EEH recovery" sent to this list?

Adding linux-pci because this hack actually tampers with internal PCI device state, which looks wrong.

The thread is here:

http://marc.info/?l=linux-scsi&m=125918723218627

and the proposed full patch and explanation below.

PCI people, is this correct, or is there a better way to do it?

James

---

Hi,

After commits c82f63e411f1b58427c103bd95af2863b1c96dd1 (PCI: check saved state before restore) and 4b77b0a2ba27d64f58f16d8d4d48d8319dda36ff (PCI: Clear saved_state after the state has been restored) PCI drivers are prevented from restoring the device standard configuration registers twice in a row. These changes introduced a regression on ipr EEH recovery.

The ipr device driver saves the PCI state only during the device probe and restores it on ipr_reset_restore_cfg_space() during IOA resets. This behavior is causing the EEH recovery to fail after the second error detected, since the registers are not being restored.

One possible solution would be saving the registers after restoring them. The problem with this approach is that while recovering from an EEH error if pci_save_state() results in an EEH error, the adapter/slot will be reset, and end up back in ipr_reset_restore_cfg_space(), but it won't have a valid saved state to restore, so pci_restore_state() will fail.

The following patch introduces a workaround for this problem, hacking around the PCI API by setting pdev->state_saved = true before we do the restore. It fixes the EEH regression and prevents that we hit another EEH error during EEH recovery.

Thanks,
Kleber

Signed-off-by: Kleber Sacilotto de Souza
<[email protected]> ---

 drivers/scsi/ipr.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)
drivers/scsi/ipr.c +1 -0
index 76d294f..c3ff9a6 100644
@@ -6516,6 +6516,7 @@ static int
 ipr_reset_restore_cfg_space(struct ipr_cmnd *ipr_cmd)
        int rc;
 
        ENTER;
+       ioa_cfg->pdev->state_saved = true;
        rc = pci_restore_state(ioa_cfg->pdev);
 
Quoted 5 lines
Quoted 1 line
        if (rc != PCIBIOS_SUCCESSFUL) {

Rafael may have input here, but it seems like we need a low level save/restore routine that ignores the flag (which is generally used for suspend/resume I think?).

There are some other users, but they are only a few.

Quoted 2 lines

Maybe adding low level _pci_save_state/_pci_restore_state that don't check/set the flags would help?

That might work, but how do we know that the state we're going to restore is actually valid at this particular point?

Perhaps we need a version using a separate storage area?

Quoted 8 lines
Quoted 2 lines

Yeah, that would probably be best. Let the caller allocate the space and save/restore it all it wants for special cases like error handling.

OK, so could I have a resolution for this, please, guys?

Do I just apply the patch to fiddle with the internal state which looks ugly but will fix the bug, or are you going to provide us with the correct interface to use?

I guess at the moment it's better to apply the workaround first and then remove it when the correct interface is ready.

I really wouldn't like to hurry with reworking pci_save_state(), because that has a potential of infroducing some new nasty bugs if not done with care.

Jesse, what's your opinion?

Rafael

#12 lore raw
Quoted 7 lines · Rafael J. Wysocki

I guess at the moment it's better to apply the workaround first and then remove it when the correct interface is ready.

I really wouldn't like to hurry with reworking pci_save_state(), because that has a potential of infroducing some new nasty bugs if not done with care.

Jesse, what's your opinion?

Yeah, that's fine. I won't have time to hack anything together for a week or two either, and I don't want to hold up a valid fix.

Jesse