Skip to contentSkip to conversation

mpt2sas: Fixed Big Indian Issues on 32 bit PPC

SCSI Dan Carpenter 2 participants 2 messages Feb 20, 2012 SCSI

lore
Share & tools
lore
#0 lore raw

Hello Kashyap, Desai,

The patch c97951ec46d4: "[SCSI] mpt2sas: Fixed Big Indian Issues on 32 bit PPC" from Jun 14, 2011, leads to the following static checker warning:

drivers/scsi/mpt2sas/mpt2sas_base.c:3352 _base_get_port_facts()
warn: are we memsetting too much? 6 vs 28

We changed the size of certain data structures.

-       Mpi2IOCFactsReply_t facts;
-       Mpi2PortFactsReply_t *pfacts;
+       struct mpt2sas_facts facts;
+       struct mpt2sas_port_facts *pfacts;

But it's not done consistently. We still memset it as if it were the original size.

	memset(facts, 0, sizeof(Mpi2IOCFactsReply_t));

And we still allocate the original size data structure:

4270 ioc->pfacts = kcalloc(ioc->facts.NumberOfPorts,
4271 sizeof(Mpi2PortFactsReply_t), GFP_KERNEL);

There are a couple other warnings as well. drivers/scsi/mpt2sas/mpt2sas_base.c:3394 _base_get_ioc_facts(24)
warn: are we memsetting too much? 60 vs 64
drivers/scsi/mpt2sas/mpt2sas_base.c:4270 mpt2sas_base_attach(56)
warn: double check that we're allocating correct size: 6 vs 28

Obviously the concern with a too big memset is that we're corrupting the next element in the array or we're corrupting past the end of the array.

regards,
dan carpenter

#1 lore raw

Dan - I fixed the issue about a couple weeks ago in the in-house drivers. From _base_get_ioc_facts, we should be memset'ing the "struct mpt2sas_facts" instead of "Mpi2IOCFactsReply_t ", because the its four bytes larger, the ioc->pfacts is getting set to NULL. Nagalaksmi will pick it up this fix and post upstream in a future patch.

Thankyou,
Eric

________________________________________
From: [email protected] [[email protected]] On Behalf Of Dan Carpenter [[email protected]]
Sent: Monday, February 20, 2012 7:05 AM
To: Desai, Kashyap
Cc: [email protected]
Subject: re: [SCSI] mpt2sas: Fixed Big Indian Issues on 32 bit PPC

Hello Kashyap, Desai,

The patch c97951ec46d4: "[SCSI] mpt2sas: Fixed Big Indian Issues on 32 bit PPC" from Jun 14, 2011, leads to the following static checker warning:

drivers/scsi/mpt2sas/mpt2sas_base.c:3352 _base_get_port_facts()
warn: are we memsetting too much? 6 vs 28

We changed the size of certain data structures.

-       Mpi2IOCFactsReply_t facts;
-       Mpi2PortFactsReply_t *pfacts;
+       struct mpt2sas_facts facts;
+       struct mpt2sas_port_facts *pfacts;

But it's not done consistently. We still memset it as if it were the original size.

        memset(facts, 0, sizeof(Mpi2IOCFactsReply_t));

And we still allocate the original size data structure:

4270 ioc->pfacts = kcalloc(ioc->facts.NumberOfPorts,
4271 sizeof(Mpi2PortFactsReply_t), GFP_KERNEL);

There are a couple other warnings as well. drivers/scsi/mpt2sas/mpt2sas_base.c:3394 _base_get_ioc_facts(24)
warn: are we memsetting too much? 60 vs 64
drivers/scsi/mpt2sas/mpt2sas_base.c:4270 mpt2sas_base_attach(56)
warn: double check that we're allocating correct size: 6 vs 28

Obviously the concern with a too big memset is that we're corrupting the next element in the array or we're corrupting past the end of the array.

regards,
dan carpenter

--
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