From mboxrd@z Thu Jan  1 00:00:00 1970
From: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Subject: Re: [PATCH v2] vfio: Fix overflow while assigning vfio
 BAR region offset and size
Date: Tue, 7 Jul 2015 13:26:25 +0530
Message-ID: <20150707075623.GA28061@scalar.blr.asicdesigners.com>
References: <61d1c9174f7f9159d4005dd2cea16c7719cec964.1434462470.git.rahul.lakkireddy@chelsio.com>
 <2eeeff646a329cca169033f541de3e96cec3a27a.1435067129.git.rahul.lakkireddy@chelsio.com>
 <4201419.gIoUKGuTjB@xps13>
 <CAD+H993L17VVuf91kE8p_ev0TtrZxh7J_yhe+bE1kjUa_fM8Wg@mail.gmail.com>
 <C6ECDF3AB251BE4894318F4E4512369780CCD04C@IRSMSX109.ger.corp.intel.com>
 <CAD+H990F7G2XntE9gzgZ=rCj-74srT2+i01gDqAd1-w6dJHv7A@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: dev <dev@dpdk.org>, Felix Marti <felix@chelsio.com>,
 Nirranjan Kirubaharan <nirranjan@chelsio.com>,
 Kumar Sanghvi <kumaras@chelsio.com>
To: Alejandro Lucero <alejandro.lucero@netronome.com>
Return-path: <dev-bounces@dpdk.org>
Received: from stargate3.asicdesigners.com (stargate.chelsio.com
 [67.207.112.58]) by dpdk.org (Postfix) with ESMTP id 1EB2FE72
 for <dev@dpdk.org>; Tue,  7 Jul 2015 09:56:37 +0200 (CEST)
Content-Disposition: inline
In-Reply-To: <CAD+H990F7G2XntE9gzgZ=rCj-74srT2+i01gDqAd1-w6dJHv7A@mail.gmail.com>
List-Id: patches and discussions about DPDK <dev.dpdk.org>
List-Unsubscribe: <http://dpdk.org/ml/options/dev>,
 <mailto:dev-request@dpdk.org?subject=unsubscribe>
List-Archive: <http://dpdk.org/ml/archives/dev/>
List-Post: <mailto:dev@dpdk.org>
List-Help: <mailto:dev-request@dpdk.org?subject=help>
List-Subscribe: <http://dpdk.org/ml/listinfo/dev>,
 <mailto:dev-request@dpdk.org?subject=subscribe>
Errors-To: dev-bounces@dpdk.org
Sender: "dev" <dev-bounces@dpdk.org>

Hi Alejandro,

On Mon, Jul 06, 2015 at 16:45:01 +0100, Alejandro Lucero wrote:
> Hi all,
> 
> From the kernel VFIO maintainer:
> 
> "I suppose in the short term, mmap should not be advertised as available
> on 32bit hosts.  Thanks,"
> 
> So, as VFIO support for 32bit systems is broken, DPDK should not configure
> VFIO in that case.
> 

Thank you very much for the clarification.

> 
> If we need to support 4G BARs, our only choice is really to extend the
> vfio region support for a separate file descriptor per region.  The only
> devices I'm aware of with 4G BARs are Nvidia Tesla.  This is possible,
> but I would expect such devices would be extremely rare on 32bit hosts.
> 

Our Chelsio T5 cards can also have 4G bar size. So, it seems this won't work
on 32-bit with current state of kernel vfio driver.

Nevertheless, updating your patch with below diff works fine on 64-bit
for vfio testing on Chelsio T5 cards and it compiles for 32-bit targets
as well.


diff --git a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
index 426953a..6127f5f 100644
--- a/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
+++ b/lib/librte_eal/linuxapp/eal/eal_pci_vfio.c
@@ -728,7 +728,7 @@ pci_vfio_map_resource(struct rte_pci_device *dev)
                struct vfio_region_info reg = { .argsz = sizeof(reg) };
                void *bar_addr;
                struct memreg {
-                       uint32_t offset, size;
+                       unsigned long offset, size;
                } memreg[2] = {};

                reg.index = i;
@@ -771,7 +771,7 @@ pci_vfio_map_resource(struct rte_pci_device *dev)
                                RTE_LOG(DEBUG, EAL,
                                        "Trying to map BAR %d that contains the MSI-X "
                                        "table. Trying offsets: "
-                                       "%04x:%04x, %04x:%04x\n", i,
+                                       "0x%04lx:0x%04lx, 0x%04lx:0x%04lx\n", i,
                                        memreg[0].offset, memreg[0].size,
                                        memreg[1].offset, memreg[1].size);
                        }


Thanks,
Rahul