[Precise, Trusty][PATCH 1/2] scsi: arcmsr: Buffer overflow in arcmsr_iop_message_xfer()

[Luis Henriques]

From: Dan Carpenter <dan.carpenter at oracle.com>

We need to put an upper bound on "user_len" so the memcpy() doesn't
overflow.

Cc: <stable at vger.kernel.org>
Reported-by: Marco Grassi <marco.gra at gmail.com>
Signed-off-by: Dan Carpenter <dan.carpenter at oracle.com>
Reviewed-by: Tomas Henzl <thenzl at redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen at oracle.com>
CVE-2016-7425
(backported from commit 7bc2b55a5c030685b399bb65b6baa9ccc3d1f167)
[ luis:
  - use '1032' instead of ARCMSR_API_DATA_BUFLEN
  - adjust context ]
Signed-off-by: Luis Henriques <luis.henriques at canonical.com>
---
 drivers/scsi/arcmsr/arcmsr_hba.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/scsi/arcmsr/arcmsr_hba.c b/drivers/scsi/arcmsr/arcmsr_hba.c
index f980600f78a8..67f505ccdcab 100644
--- a/drivers/scsi/arcmsr/arcmsr_hba.c
+++ b/drivers/scsi/arcmsr/arcmsr_hba.c
@@ -1803,7 +1803,8 @@ static int arcmsr_iop_message_xfer(struct AdapterControlBlock *acb,
 
 	case ARCMSR_MESSAGE_WRITE_WQBUFFER: {
 		unsigned char *ver_addr;
-		int32_t my_empty_len, user_len, wqbuf_firstindex, wqbuf_lastindex;
+		int32_t my_empty_len, wqbuf_firstindex, wqbuf_lastindex;
+		uint32_t user_len;
 		uint8_t *pQbuffer, *ptmpuserbuffer;
 
 		ver_addr = kmalloc(1032, GFP_ATOMIC);
@@ -1820,6 +1821,11 @@ static int arcmsr_iop_message_xfer(struct AdapterControlBlock *acb,
 		}
 		ptmpuserbuffer = ver_addr;
 		user_len = pcmdmessagefld->cmdmessage.Length;
+		if (user_len > 1032) {
+			retvalue = ARCMSR_MESSAGE_FAIL;
+			kfree(ver_addr);
+			goto message_out;
+		}
 		memcpy(ptmpuserbuffer, pcmdmessagefld->messagedatabuffer, user_len);
 		wqbuf_lastindex = acb->wqbuf_lastindex;
 		wqbuf_firstindex = acb->wqbuf_firstindex;