Thread (2 messages) 2 messages, 1 author, 2021-01-07

RE: [PATCH 2/3] hv_netvsc: Wait for completion on request NVSP_MSG4_TYPE_SWITCH_DATA_PATH

From: Long Li <longli@microsoft.com>
Date: 2021-01-07 00:00:58
Also in: lkml, netdev

Possibly related (same subject, not in this thread)

Subject: Re: [PATCH 2/3] hv_netvsc: Wait for completion on request
NVSP_MSG4_TYPE_SWITCH_DATA_PATH

Hi Long,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on linus/master] [also build test WARNING on
v5.11-rc2 next-20210104] [If your patch is applied to the wrong git tree, kindly
drop us a note.
And when submitting patch, we suggest to use '--base' as documented in
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgit-
scm.com%2Fdocs%2Fgit-format-
patch&amp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3d454eb468
b85fb08d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6
37455042608743102%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMD
AiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=
90AgH9HlZumRZ4UNC4uD2WIRpZ6ZEnvIdOKOfzYcXpI%3D&amp;reserved=0]

url:
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
ub.com%2F0day-ci%2Flinux%2Fcommits%2FLong-Li%2Fhv_netvsc-Check-VF-
datapath-when-sending-traffic-to-VF%2F20210106-
092237&amp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3d454eb46
8b85fb08d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C
637455042608753098%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwM
DAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata
=vtVJ8pXIOxIYeKdaqT9pD1%2BEuOM3wz4yqsHh8uWsGP4%3D&amp;reserv
ed=0
base:
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgit.k
ernel.org%2Fpub%2Fscm%2Flinux%2Fkernel%2Fgit%2Ftorvalds%2Flinux.git&
amp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3d454eb468b85fb0
8d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C6374550
42608753098%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQ
IjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=FXMG
CFODFoq3KLklxr17iVHiq%2FWmJ3c0fM7vIZRfNmc%3D&amp;reserved=0
e71ba9452f0b5b2e8dc8aa5445198cd9214a6a62
config: i386-allyesconfig (attached as .config)
compiler: gcc-9 (Debian 9.3.0-15) 9.3.0
reproduce (this is a W=1 build):
        #
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
ub.com%2F0day-
ci%2Flinux%2Fcommit%2F8c92b5574da1b0c2aee3eab7da2c4dad8d92572c&a
mp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3d454eb468b85fb08
d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C63745504
2608753098%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIj
oiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=MMXkQ
KENGpyfW0NJs2khBSKTuBExFSZaWHgWyyIj6UU%3D&amp;reserved=0
        git remote add linux-review
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
ub.com%2F0day-
ci%2Flinux&amp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3d454e
b468b85fb08d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0
%7C637455042608753098%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjA
wMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;s
data=uge6PX2NyAe%2BjRtvgOhR5xzN2ltBctZXeZwn0hoYco0%3D&amp;reser
ved=0
        git fetch --no-tags linux-review Long-Li/hv_netvsc-Check-VF-datapath-
when-sending-traffic-to-VF/20210106-092237
        git checkout 8c92b5574da1b0c2aee3eab7da2c4dad8d92572c
        # save the attached .config to linux build tree
        make W=1 ARCH=i386

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <redacted>

All warnings (new ones prefixed by >>):

   drivers/net/hyperv/netvsc.c: In function 'netvsc_send_completion':
quoted
quoted
drivers/net/hyperv/netvsc.c:778:14: warning: cast to pointer from
integer of different size [-Wint-to-pointer-cast]
     778 |   pkt_rqst = (struct nvsp_message *)cmd_rqst;
         |              ^
I think this warning can be safely ignored.

When sending packets over vmbus, the address is passed as u64 and stored internally as u64 in vmbus_next_request_id(). Passing a 32 bit address will not lose any data. Later the address is retrieved from vmbus_request_addr() as a u64. Again, it will not lose data when casting to a 32 bit address.

This method of storing and retrieving addresses are used throughout other hyper-v drivers. If we want to not to trigger this warning, I suggest making a patch to convert all those usages in all hyper-v drivers.

Thanks,
Long

vim +778 drivers/net/hyperv/netvsc.c

   757
   758	static void netvsc_send_completion(struct net_device *ndev,
   759					   struct netvsc_device *net_device,
   760					   struct vmbus_channel
*incoming_channel,
   761					   const struct vmpacket_descriptor
*desc,
   762					   int budget)
   763	{
   764		const struct nvsp_message *nvsp_packet =
hv_pkt_data(desc);
   765		u32 msglen = hv_pkt_datalen(desc);
   766		struct nvsp_message *pkt_rqst;
   767		u64 cmd_rqst;
   768
   769		/* First check if this is a VMBUS completion without data
payload */
   770		if (!msglen) {
   771			cmd_rqst =
vmbus_request_addr(&incoming_channel->requestor,
   772						      (u64)desc->trans_id);
   773			if (cmd_rqst == VMBUS_RQST_ERROR) {
   774				netdev_err(ndev, "Invalid transaction id\n");
   775				return;
   776			}
   777
 > 778			pkt_rqst = (struct nvsp_message *)cmd_rqst;
   779			switch (pkt_rqst->hdr.msg_type) {
   780			case NVSP_MSG4_TYPE_SWITCH_DATA_PATH:
   781				complete(&net_device->channel_init_wait);
   782				break;
   783
   784			default:
   785				netdev_err(ndev, "Unexpected VMBUS
completion!!\n");
   786			}
   787			return;
   788		}
   789
   790		/* Ensure packet is big enough to read header fields */
   791		if (msglen < sizeof(struct nvsp_message_header)) {
   792			netdev_err(ndev, "nvsp_message length too
small: %u\n", msglen);
   793			return;
   794		}
   795
   796		switch (nvsp_packet->hdr.msg_type) {
   797		case NVSP_MSG_TYPE_INIT_COMPLETE:
   798			if (msglen < sizeof(struct nvsp_message_header) +
   799					sizeof(struct
nvsp_message_init_complete)) {
   800				netdev_err(ndev, "nvsp_msg length too
small: %u\n",
   801					   msglen);
   802				return;
   803			}
   804			fallthrough;
   805
   806		case NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE:
   807			if (msglen < sizeof(struct nvsp_message_header) +
   808					sizeof(struct
nvsp_1_message_send_receive_buffer_complete)) {
   809				netdev_err(ndev, "nvsp_msg1 length too
small: %u\n",
   810					   msglen);
   811				return;
   812			}
   813			fallthrough;
   814
   815		case NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE:
   816			if (msglen < sizeof(struct nvsp_message_header) +
   817					sizeof(struct
nvsp_1_message_send_send_buffer_complete)) {
   818				netdev_err(ndev, "nvsp_msg1 length too
small: %u\n",
   819					   msglen);
   820				return;
   821			}
   822			fallthrough;
   823
   824		case NVSP_MSG5_TYPE_SUBCHANNEL:
   825			if (msglen < sizeof(struct nvsp_message_header) +
   826					sizeof(struct
nvsp_5_subchannel_complete)) {
   827				netdev_err(ndev, "nvsp_msg5 length too
small: %u\n",
   828					   msglen);
   829				return;
   830			}
   831			/* Copy the response back */
   832			memcpy(&net_device->channel_init_pkt,
nvsp_packet,
   833			       sizeof(struct nvsp_message));
   834			complete(&net_device->channel_init_wait);
   835			break;
   836
   837		case NVSP_MSG1_TYPE_SEND_RNDIS_PKT_COMPLETE:
   838			netvsc_send_tx_complete(ndev, net_device,
incoming_channel,
   839						desc, budget);
   840			break;
   841
   842		default:
   843			netdev_err(ndev,
   844				   "Unknown send completion type %d
received!!\n",
   845				   nvsp_packet->hdr.msg_type);
   846		}
   847	}
   848

---
0-DAY CI Kernel Test Service, Intel Corporation
https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.
01.org%2Fhyperkitty%2Flist%2Fkbuild-
all%40lists.01.org&amp;data=04%7C01%7Clongli%40microsoft.com%7C695cf3
d454eb468b85fb08d8b1fb3ddd%7C72f988bf86f141af91ab2d7cd011db47%7C1
%7C0%7C637455042608753098%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC
4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&
amp;sdata=AKWfmJrn1C%2BwaqX6wlu95HcPys9K0ju%2FlC%2Bu3O20jAg%3
D&amp;reserved=0
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help