mpt2sas: Fix &&/|| confusion in _scsih_sas_device_status_change_event()

PATCH NAK Roel Kluin 3 participants 7 messages Dec 30, 2009 – Jan 13, 2010 SCSI

lore

Share & tools

lore

Roel Kluin

author

2009-12-30 16:55 UTC #0 lore raw

PATCH 1/1 mpt2sas: Fix &&/|| confusion in _scsih_sas_device_status_change_event()

This never evaluates to true.

Signed-off-by Roel Kluin <[email protected]>

---
 drivers/scsi/mpt2sas/mpt2sas_scsih.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

drivers/scsi/mpt2sas/mpt2sas_scsih.c +1 -1

index efabea1..82c5474 100644
@@ -4363,7 +4363,7 @@ _scsih_sas_device_status_change_event(struct MPT2SAS_ADAPTER *ioc,
 #endif
 
 	if (!(event_data->ReasonCode ==
-	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
+	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET ||
 	   event_data->ReasonCode ==
 	    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET))
 		return;

The code beyond this check is intended for either INTERNAL_DEVICE_RESET or CMP_DEVICE_RESET. If the reason code is something else, we will want to return. There are 10 other types of reason codes besides these two. This proposed patch means we return only when the reason code is either INTERNAL_DEVICE_RESET or CMP_INTERNAL_RESET.

Eric Moore

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Roel Kluin
Sent: Wednesday, December 30, 2009 9:59 AM
To: James E.J. Bottomley; [email protected]; Andrew Morton; LKML; Desai, Kashyap
Subject: [PATCH] mpt2sas: Fix &&/|| confusion in _scsih_sas_device_status_change_event()

This never evaluates to true.

Signed-off-by Roel Kluin <[email protected]>

---
 drivers/scsi/mpt2sas/mpt2sas_scsih.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

drivers/scsi/mpt2sas/mpt2sas_scsih.c +1 -1

index efabea1..82c5474 100644
@@ -4363,7 +4363,7 @@ _scsih_sas_device_status_change_event(struct MPT2SAS_ADAPTER *ioc,
 #endif
 
 	if (!(event_data->ReasonCode ==
-	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
+	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET ||
 	   event_data->ReasonCode ==
 	    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET))
 		return;

--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

roel kluin

author

2010-01-07 19:58 UTC #2 lore raw

Quoted 5 lines · Eric

nack
The code beyond this check is intended for either INTERNAL_DEVICE_RESET or CMP_DEVICE_RESET. If the reason code is something else, we will want to return. There are 10 other types of reason codes besides these two. This proposed patch means we return only when the reason code is either INTERNAL_DEVICE_RESET or CMP_INTERNAL_RESET.
Eric Moore

my patch is correct but my changelog was wrong. Sorry for the confusion. It should have been:

Even if the ReasonCode is not INTERNAL_DEVICE_RESET nor CMP_DEVICE_RESET this still evaluates to true.

you can test this with:

#include <stdio.h>

int main()

{
        int d;
        for (d=0;d<4;++d)
                printf("!(%d == 1 && %d == 2) = %d\n", d, d, !(d == 1
&& d == 2));
        for (d=0;d<4;++d)
                printf("!(%d == 1 || %d == 2) = %d\n", d, d, !(d == 1
|| d == 2));
        return 0;
}

Roel

Andrew Morton

2010-01-12 21:54 UTC #3 lore raw

Quoted 5 lines · On Thu, 7 Jan 2010 20:58:10 +0100 roel kluin <[email protected]> wrote:

Quoted 4 lines · Eric
nack
The code beyond this check is intended for either INTERNAL_DEVICE_RESET or CMP_DEVICE_RESET. __ If the reason code is something else, we will want to return. __There are 10 other types of reason codes besides these two. This proposed patch means we return only when the reason code is either INTERNAL_DEVICE_RESET or CMP_INTERNAL_RESET.

Quoted 8 lines

Quoted 1 line
Eric Moore
my patch is correct but my changelog was wrong. Sorry for the confusion. It should have been:
Even if the ReasonCode is not INTERNAL_DEVICE_RESET nor CMP_DEVICE_RESET this still evaluates to true.

yup, the current code is wrong.

	if (!(event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	   event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET))
		return;

ReasonCode cannot equal two different things at the same time.

This is what we want:

	if (event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	    event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET)
		return;

Eric, the fix needs runtime testing please - it enables code which hasn't been executed in a long time, if ever.

In fact the compiler wasn't even emitting any code for the second half of _scsih_sas_device_status_change_event() because it worked out that there was no path to it.

Moore, Eric

2010-01-13 00:23 UTC #4 lore raw

Yes your patch will work. I just have tested it by sending INTERNAL_DEVICE_RESETS events, and the code is getting executed.

Regards,
Eric

-----Original Message-----
From: Andrew Morton [mailto:[email protected]]
Sent: Tuesday, January 12, 2010 2:54 PM
To: roel kluin
Cc: Moore, Eric; James E.J. Bottomley; [email protected]; LKML; Desai, Kashyap
Subject: Re: [PATCH] mpt2sas: Fix &&/|| confusion in _scsih_sas_device_status_change_event()

Quoted 5 lines · On Thu, 7 Jan 2010 20:58:10 +0100 roel kluin <[email protected]> wrote:

Quoted 4 lines · Eric
nack
The code beyond this check is intended for either INTERNAL_DEVICE_RESET or CMP_DEVICE_RESET. __ If the reason code is something else, we will want to return. __There are 10 other types of reason codes besides these two. This proposed patch means we return only when the reason code is either INTERNAL_DEVICE_RESET or CMP_INTERNAL_RESET.

Quoted 8 lines

Quoted 1 line
Eric Moore
my patch is correct but my changelog was wrong. Sorry for the confusion. It should have been:
Even if the ReasonCode is not INTERNAL_DEVICE_RESET nor CMP_DEVICE_RESET this still evaluates to true.

yup, the current code is wrong.

	if (!(event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	   event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET))
		return;

ReasonCode cannot equal two different things at the same time.

This is what we want:

	if (event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	    event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET)
		return;

Eric, the fix needs runtime testing please - it enables code which hasn't been executed in a long time, if ever.

In fact the compiler wasn't even emitting any code for the second half of _scsih_sas_device_status_change_event() because it worked out that there was no path to it.

Andrew Morton

2010-01-13 00:38 UTC #5 lore raw

Quoted 45 lines · Eric

Quoted 45 lines
-----Original Message-----
From: Andrew Morton [mailto:[email protected]]
Sent: Tuesday, January 12, 2010 2:54 PM
To: roel kluin
Cc: Moore, Eric; James E.J. Bottomley; [email protected]; LKML; Desai, Kashyap
Subject: Re: [PATCH] mpt2sas: Fix &&/|| confusion in _scsih_sas_device_status_change_event()
Quoted 5 lines · On Thu, 7 Jan 2010 20:58:10 +0100 roel kluin <[email protected]> wrote:
Quoted 4 lines · Eric
nack
The code beyond this check is intended for either INTERNAL_DEVICE_RESET or CMP_DEVICE_RESET. __ If the reason code is something else, we will want to return. __There are 10 other types of reason codes besides these two. This proposed patch means we return only when the reason code is either INTERNAL_DEVICE_RESET or CMP_INTERNAL_RESET.
Quoted 8 lines
Quoted 1 line
Eric Moore
my patch is correct but my changelog was wrong. Sorry for the confusion. It should have been:
Even if the ReasonCode is not INTERNAL_DEVICE_RESET nor CMP_DEVICE_RESET this still evaluates to true.
yup, the current code is wrong.
	if (!(event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	   event_data->ReasonCode ==
	    MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET))
		return;
ReasonCode cannot equal two different things at the same time.
This is what we want:
	if (event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_INTERNAL_DEVICE_RESET &&
	    event_data->ReasonCode !=
	      MPI2_EVENT_SAS_DEV_STAT_RC_CMP_INTERNAL_DEV_RESET)
		return;
Eric, the fix needs runtime testing please - it enables code which hasn't been executed in a long time, if ever.

Quoted 7 lines

Quoted 4 lines
In fact the compiler wasn't even emitting any code for the second half of _scsih_sas_device_status_change_event() because it worked out that there was no path to it.
Yes your patch will work. I just have tested it by sending INTERNAL_DEVICE_RESETS events, and the code is getting executed.

(top-posting and quoting fixed)

Thanks. Was that an ack to merge?

Moore, Eric

2010-01-13 00:42 UTC #6 lore raw

Quoted 3 lines

Quoted 3 lines
Thanks. Was that an ack to merge?

Yes, please merge in the code.

ack