From mboxrd@z Thu Jan 1 00:00:00 1970 From: Maxime Coquelin Subject: Re: [PATCH v3 05/19] vhost: fix error handling when mem table gets updated Date: Thu, 4 Oct 2018 17:06:29 +0200 Message-ID: <9eade352-c26c-f8db-8a6e-b6b81a54c374@redhat.com> References: <20181004081403.8039-1-maxime.coquelin@redhat.com> <20181004081403.8039-6-maxime.coquelin@redhat.com> <20181004145646eucas1p11bcd0adb05bc38ff89e19c60b5af0734~ab6vFE4Uy0940809408eucas1p15@eucas1p1.samsung.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Cc: dgilbert@redhat.com, stable@dpdk.org To: Ilya Maximets , dev@dpdk.org, tiwei.bie@intel.com, zhihong.wang@intel.com, jfreimann@redhat.com, nicknickolaev@gmail.com, bruce.richardson@intel.com, alejandro.lucero@netronome.com Return-path: In-Reply-To: <20181004145646eucas1p11bcd0adb05bc38ff89e19c60b5af0734~ab6vFE4Uy0940809408eucas1p15@eucas1p1.samsung.com> Content-Language: en-US List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On 10/04/2018 04:59 PM, Ilya Maximets wrote: > On 04.10.2018 11:13, Maxime Coquelin wrote: >> When the memory table gets updated, the rings addresses need >> to be translated again. If it fails, we need to exit cleanly >> by unmapping memory regions. >> >> Fixes: d5022533c20a ("vhost: retranslate vring addr when memory table changes") >> Cc: stable@dpdk.org >> >> Signed-off-by: Maxime Coquelin >> --- > > Acked-by: Ilya Maximets > > Minor comments inline. > >> lib/librte_vhost/vhost_user.c | 3 ++- >> 1 file changed, 2 insertions(+), 1 deletion(-) >> >> diff --git a/lib/librte_vhost/vhost_user.c b/lib/librte_vhost/vhost_user.c >> index 8ffe5aa66..b6eae8dc5 100644 >> --- a/lib/librte_vhost/vhost_user.c >> +++ b/lib/librte_vhost/vhost_user.c >> @@ -964,7 +964,8 @@ vhost_user_set_mem_table(struct virtio_net **pdev, struct VhostUserMsg *msg) >> >> dev = translate_ring_addresses(dev, i); >> if (!dev) >> - return VH_RESULT_ERR; >> + goto err_mmap; >> + > > 1. No need to have two empty lines. (You could fix this while applying) Right. I will either fix it while applying, or it will be in next revision, if any. > 2. In current code, error on message handling will cause disconnect and > memory regions will be freed anyway. So, the change is not very > important for master (maybe just for consistency with surrounding > code) but it could be important for stable versions. I may not disconnect directly, as if reply-ack feature is negotiated[0], the slave would reply with a NACK and function handler will return 0. I guess that in that case the master (QEMU) will disconnect anyway, but this is just an assumption from slave point of view. [0]: REPLY_ACK protocol feature is only advertised when IOMMU support is requested because of a bug in an old upstream QEMU version. Thanks! Maxime >> >> *pdev = dev; >> } >>