[PATCH] mptctl.c memory leak on failure

[Arnaldo Carvalho de Melo]

Hi,

	Please consider applying this patch. And I got confused by the
kmalloc and the comment: the buffer is used for DMA but the kmalloc doesn't
has GFP_DMA, maybe I'm missing something here, its about time for me to
sleep 8) but if it doesn't needs DMA you should consider using vmalloc and
you'll get rid of the 32 pages limitation, like I did for wanrouter that
used kmalloc but didn't needed DMA, so I've changed it to vmalloc as per
Alan's cluebating^Wsuggestion at the time 8)

- Arnaldo

--- linux-2.4.0-ac3/drivers/message/fusion/mptctl.c	Sat Jan  6 23:13:11 2001
+++ linux-2.4.0-ac3.acme/drivers/message/fusion/mptctl.c	Sun Jan  7 01:02:52 2001
@@ -290,6 +290,7 @@
 	dma_addr_t		 fwbuf_dma;
 	FWDownloadTCSGE_t	*fwVoodoo;
 	SGEAllUnion_t		*fwSgl;
+	int ret;
 
 	dprintk((KERN_INFO "mptctl_do_fwdl called. mptctl_id = %xh.\n", mptctl_id)); //tc
 
@@ -309,20 +310,22 @@
 	}
 //dprintk((KERN_INFO "DbG: fwbuf       = %p\n", fwbuf));
 
+	ret = -EFAULT;
 	if (copy_from_user(fwbuf, ufwbuf, fwlen) < 0) {
 		printk(KERN_ERR "%s@%d::_ioctl_fwdl - "
 				"Unable to copy f/w buffer @ %p\n",
 				__FILE__, __LINE__, (void*)ufwbuf);
-		return -EFAULT;
+		goto out;
 	}
 
 	/* Perform Firmware download.
 	 */
 
+	ret = -ENXIO; /* (-6) No such device or address */
 	if ((ioc = mpt_verify_adapter(ioc, &iocp)) < 0) {
 		printk("%s@%d::_ioctl_fwdl - ioc%d not found!\n",
 				__FILE__, __LINE__, ioc);
-		return -ENXIO;			  /* (-6) No such device or address */
+		goto out;
 	}
 
 	mf = MptGetMsgFrame(mptctl_id, ioc);
@@ -373,8 +376,9 @@
 					   "In ReplyMsg loop - iteration %d\n",
 					   foo)); //tc
 			}
+			ret = -ETIME;
 			if (++foo > 60000000)
-				return -ETIME;
+				goto out;
 			mb();
 			schedule();
 			barrier();
@@ -402,6 +406,9 @@
 		printk(KERN_WARNING MYNAM ": (bad VooDoo):\n");
 		return -ENOMSG;
 	}
+	return 0;
+out:	kfree(fwbuf);
+	return ret;
 }
 
 /*=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=*/