[patch] mptfusion: prevent some memory corruption

[Dan Carpenter <dan.carpenter@oracle.com>]

These are signed values the come from the user, we put a cap on the
upper bounds but not on the lower bounds.

We use "karg.dataSgeOffset" to calculate "sz".  We verify "sz" and
proceed as if that means that "karg.dataSgeOffset" is correct but this
fails to consider that the "sz" calculations can have integer overflows.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
---
During my QC process, I realized that I sent a similar patch last year,
but never received a response.

http://permalink.gmane.org/gmane.linux.kernel.janitors/32590

Looking at both of them, I guess I prefer today's patch because it is
simpler.

diff --git a/drivers/message/fusion/mptctl.c b/drivers/message/fusion/mptctl.c
index 70bb753..fc73937 100644
--- a/drivers/message/fusion/mptctl.c
+++ b/drivers/message/fusion/mptctl.c
@@ -1859,6 +1859,15 @@ mptctl_do_mpt_command (struct mpt_ioctl_command karg, void __user *mfPtr)
 	}
 	spin_unlock_irqrestore(&ioc->taskmgmt_lock, flags);
 
+	/* Basic sanity checks to prevent underflows or integer overflows */
+	if (karg.maxReplyBytes < 0 ||
+	    karg.dataInSize < 0 ||
+	    karg.dataOutSize < 0 ||
+	    karg.dataSgeOffset < 0 ||
+	    karg.maxSenseBytes < 0 ||
+	    karg.dataSgeOffset > ioc->req_sz / 4)
+		return -EINVAL;
+
 	/* Verify that the final request frame will not be too large.
 	 */
 	sz = karg.dataSgeOffset * 4;