[PATCH] scsi: mpt3sas: Fix out-of-bounds compiler warning

* [PATCH] scsi: mpt3sas: Fix out-of-bounds compiler warning
@ 2022-05-31 20:09 Helge Deller
  2022-06-08  2:27 ` Martin K. Petersen
  0 siblings, 1 reply; 2+ messages in thread
From: Helge Deller @ 2022-05-31 20:09 UTC (permalink / raw)
  To: MPT-FusionLinux.pdl, linux-scsi, Martin K. Petersen,
	Sathya Prakash, Sreekanth Reddy, James Bottomley
  Cc: linux-parisc

I'm facing this warning when building for the parisc64 architecture:

drivers/scsi/mpt3sas/mpt3sas_base.c: In function ‘_base_make_ioc_operational’:
drivers/scsi/mpt3sas/mpt3sas_base.c:5396:40: warning: array subscript ‘Mpi2SasIOUnitPage1_t {aka struct _MPI2_CONFIG_PAGE_SASIOUNIT_1}[0]’ is partly outside array bounds of ‘unsigned char[20]’ [-Warray-bounds]
 5396 |             (le16_to_cpu(sas_iounit_pg1->SASWideMaxQueueDepth)) ?
drivers/scsi/mpt3sas/mpt3sas_base.c:5382:26: note: referencing an object of size 20 allocated by ‘kzalloc’
 5382 |         sas_iounit_pg1 = kzalloc(sz, GFP_KERNEL);
      |                          ^~~~~~~~~~~~~~~~~~~~~~~

The problem is, that only 20 bytes are allocated with kmalloc(), which
is sufficient to hold the bytes which are needed.
Nevertheless, gcc complains because the whole Mpi2SasIOUnitPage1_t
struct is 32 bytes in size and thus doesn't fit into those 20 bytes.

This patch simply allocates all 32 bytes (instead of 20) and thus avoids the
warning. There is no functional change introduced by this patch.

While touching the code I cleaned up to calculation of max_wideport_qd,
max_narrowport_qd and max_sata_qd to make it easier readable.

Test sucessfully tested on a HP C8000 PA-RISC workstation with 64-bit
kernel.

Signed-off-by: Helge Deller <deller@gmx.de>

diff --git a/drivers/scsi/mpt3sas/mpt3sas_base.c b/drivers/scsi/mpt3sas/mpt3sas_base.c
index 538d2c0cd971..aa142052ebe4 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_base.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_base.c
@@ -5368,6 +5368,7 @@ static int _base_assign_fw_reported_qd(struct MPT3SAS_ADAPTER *ioc)
 	Mpi2ConfigReply_t mpi_reply;
 	Mpi2SasIOUnitPage1_t *sas_iounit_pg1 = NULL;
 	Mpi26PCIeIOUnitPage1_t pcie_iounit_pg1;
+	u16 depth;
 	int sz;
 	int rc = 0;

@@ -5379,7 +5380,7 @@ static int _base_assign_fw_reported_qd(struct MPT3SAS_ADAPTER *ioc)
 		goto out;
 	/* sas iounit page 1 */
 	sz = offsetof(Mpi2SasIOUnitPage1_t, PhyData);
-	sas_iounit_pg1 = kzalloc(sz, GFP_KERNEL);
+	sas_iounit_pg1 = kzalloc(sizeof(Mpi2SasIOUnitPage1_t), GFP_KERNEL);
 	if (!sas_iounit_pg1) {
 		pr_err("%s: failure at %s:%d/%s()!\n",
 		    ioc->name, __FILE__, __LINE__, __func__);
@@ -5392,16 +5393,16 @@ static int _base_assign_fw_reported_qd(struct MPT3SAS_ADAPTER *ioc)
 		    ioc->name, __FILE__, __LINE__, __func__);
 		goto out;
 	}
-	ioc->max_wideport_qd =
-	    (le16_to_cpu(sas_iounit_pg1->SASWideMaxQueueDepth)) ?
-	    le16_to_cpu(sas_iounit_pg1->SASWideMaxQueueDepth) :
-	    MPT3SAS_SAS_QUEUE_DEPTH;
-	ioc->max_narrowport_qd =
-	    (le16_to_cpu(sas_iounit_pg1->SASNarrowMaxQueueDepth)) ?
-	    le16_to_cpu(sas_iounit_pg1->SASNarrowMaxQueueDepth) :
-	    MPT3SAS_SAS_QUEUE_DEPTH;
-	ioc->max_sata_qd = (sas_iounit_pg1->SATAMaxQDepth) ?
-	    sas_iounit_pg1->SATAMaxQDepth : MPT3SAS_SATA_QUEUE_DEPTH;
+
+	depth = le16_to_cpu(sas_iounit_pg1->SASWideMaxQueueDepth);
+	ioc->max_wideport_qd = (depth ? depth : MPT3SAS_SAS_QUEUE_DEPTH);
+
+	depth = le16_to_cpu(sas_iounit_pg1->SASNarrowMaxQueueDepth);
+	ioc->max_narrowport_qd = (depth ? depth : MPT3SAS_SAS_QUEUE_DEPTH);
+
+	depth = sas_iounit_pg1->SATAMaxQDepth;
+	ioc->max_sata_qd = (depth ? depth : MPT3SAS_SATA_QUEUE_DEPTH);
+
 	/* pcie iounit page 1 */
 	rc = mpt3sas_config_get_pcie_iounit_pg1(ioc, &mpi_reply,
 	    &pcie_iounit_pg1, sizeof(Mpi26PCIeIOUnitPage1_t));

^ permalink raw reply related	[flat|nested] 2+ messages in thread