[PATCH] [SCSI] scsilun_to_int should ignore the highest 2 bits

[PATCH] [SCSI] scsilun_to_int should ignore the highest 2 bits

[Jan Vesely]

From: Jan Vesely <jvesely@redhat.com>

The comment says the function does this but it does not.
Reported luns change from weirdly high numbers (like 16640)
to something saner (256), when using flat space addressing.

CC: James Bottomley <JBottomley@Parallels.com>
CC: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Jan Vesely <jvesely@redhat.com>
---
 drivers/scsi/scsi_scan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
index 3e58b22..38dc093 100644
--- a/drivers/scsi/scsi_scan.c
+++ b/drivers/scsi/scsi_scan.c
@@ -1244,7 +1244,7 @@ int scsilun_to_int(struct scsi_lun *scsilun)
 
 	lun = 0;
 	for (i = 0; i < sizeof(lun); i += 2)
-		lun = lun | (((scsilun->scsi_lun[i] << 8) |
+		lun = lun | ((((scsilun->scsi_lun[i] & 0x3f) << 8) |
 			      scsilun->scsi_lun[i + 1]) << (i * 8));
 	return lun;
 }
-- 
1.8.1.4

Re: [PATCH] [SCSI] scsilun_to_int should ignore the highest 2 bits

[Hannes Reinecke]

On 06/06/2013 10:18 AM, Jan Vesely wrote:
> From: Jan Vesely <jvesely@redhat.com>
> 
> The comment says the function does this but it does not.
> Reported luns change from weirdly high numbers (like 16640)
> to something saner (256), when using flat space addressing.
> 
> CC: James Bottomley <JBottomley@Parallels.com>
> CC: Dan Williams <dan.j.williams@intel.com>
> Signed-off-by: Jan Vesely <jvesely@redhat.com>
> ---
>  drivers/scsi/scsi_scan.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
> index 3e58b22..38dc093 100644
> --- a/drivers/scsi/scsi_scan.c
> +++ b/drivers/scsi/scsi_scan.c
> @@ -1244,7 +1244,7 @@ int scsilun_to_int(struct scsi_lun *scsilun)
>  
>  	lun = 0;
>  	for (i = 0; i < sizeof(lun); i += 2)
> -		lun = lun | (((scsilun->scsi_lun[i] << 8) |
> +		lun = lun | ((((scsilun->scsi_lun[i] & 0x3f) << 8) |
>  			      scsilun->scsi_lun[i + 1]) << (i * 8));
>  	return lun;
>  }
> 
Bzzt. It's not that simple.

For SCSI-3 _all_ numbers are valid, and doesn't know of any
addressing scheme. It's only SPC-2 which introduced the addressing
scheme. So at the very least you should be checking the scsi
revision before attempting something like this.

But in general doing a sequential scan past 256 is criminally
dangerous. Any array / device attempting to is in most cases
misconfigured or does not have the correct BLIST flag set.

I know of some older Hitachi and EMC firmware which would pretend to
be SCSI-2, but supporting more than 256 LUNs per host.
Which, of course, it totally bonkers.

I'll be posting my 64-bit LUN patchset, that should fix this issue.

Cheers,

Hannes
-- 
Dr. Hannes Reinecke		      zSeries & Storage
hare@suse.de			      +49 911 74053 688
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg
GF: J. Hawn, J. Guild, F. Imendörffer, HRB 16746 (AG Nürnberg)

Re: [PATCH] [SCSI] scsilun_to_int should ignore the highest 2 bits

[Jan Vesely]

On Thu 06 Jun 2013 10:42:16 CEST, Hannes Reinecke wrote:
> On 06/06/2013 10:18 AM, Jan Vesely wrote:
>> From: Jan Vesely <jvesely@redhat.com>
>>
>> The comment says the function does this but it does not.
>> Reported luns change from weirdly high numbers (like 16640)
>> to something saner (256), when using flat space addressing.
>>
>> CC: James Bottomley <JBottomley@Parallels.com>
>> CC: Dan Williams <dan.j.williams@intel.com>
>> Signed-off-by: Jan Vesely <jvesely@redhat.com>
>> ---
>>  drivers/scsi/scsi_scan.c | 2 +-
>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
>> index 3e58b22..38dc093 100644
>> --- a/drivers/scsi/scsi_scan.c
>> +++ b/drivers/scsi/scsi_scan.c
>> @@ -1244,7 +1244,7 @@ int scsilun_to_int(struct scsi_lun *scsilun)
>>
>>  	lun = 0;
>>  	for (i = 0; i < sizeof(lun); i += 2)
>> -		lun = lun | (((scsilun->scsi_lun[i] << 8) |
>> +		lun = lun | ((((scsilun->scsi_lun[i] & 0x3f) << 8) |
>>  			      scsilun->scsi_lun[i + 1]) << (i * 8));
>>  	return lun;
>>  }
>>
> Bzzt. It's not that simple.
>
> For SCSI-3 _all_ numbers are valid, and doesn't know of any
> addressing scheme. It's only SPC-2 which introduced the addressing
> scheme. So at the very least you should be checking the scsi
> revision before attempting something like this.
>
> But in general doing a sequential scan past 256 is criminally
> dangerous. Any array / device attempting to is in most cases
> misconfigured or does not have the correct BLIST flag set.
>
> I know of some older Hitachi and EMC firmware which would pretend to
> be SCSI-2, but supporting more than 256 LUNs per host.
> Which, of course, it totally bonkers.
>
> I'll be posting my 64-bit LUN patchset, that should fix this issue.
>
> Cheers,
>
> Hannes

thanks for your response.

I'm concerned with iSCSI. it uses SAM2 LUN addressing scheme,
and since I found that comment I did not investigate further.

I'll wait for your lun64 patches,
thanks again
--
Jan Vesely <jvesely@redhat.com>