[PATCH] vmxnet3: fix LRO feature check

Subsystems: networking drivers, the rest, vmware vmxnet3 ethernet driver

STALE3131d

9 messages, 4 authors, 2018-03-20 · open the first message on its own page

[PATCH] vmxnet3: fix LRO feature check

From: Igor Pylypiv <hidden>
Date: 2018-03-17 07:59:48

rxcsum and lro fields were deleted in commit a0d2730c9571 ("net: vmxnet3:
convert to hw_features"). With upgrading to newer version those fields were
resurrected and new code started using uninitialized lro field.
Removing rxcsum and lro fields.

Fixes: 45dac1d6ea04 ("vmxnet3: Changes for vmxnet3 adapter version 2 (fwd)")
Signed-off-by: Igor Pylypiv <redacted>
---
 drivers/net/vmxnet3/vmxnet3_drv.c | 4 +++-
 drivers/net/vmxnet3/vmxnet3_int.h | 3 ---
 2 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/net/vmxnet3/vmxnet3_drv.c b/drivers/net/vmxnet3/vmxnet3_drv.c
index 8b39c160743d..60776c86aaaf 100644
--- a/drivers/net/vmxnet3/vmxnet3_drv.c
+++ b/drivers/net/vmxnet3/vmxnet3_drv.c

@@ -1470,8 +1470,10 @@ vmxnet3_rq_rx_complete(struct vmxnet3_rx_queue *rq,
 			vmxnet3_rx_csum(adapter, skb,
 					(union Vmxnet3_GenericDesc *)rcd);
 			skb->protocol = eth_type_trans(skb, adapter->netdev);
-			if (!rcd->tcp || !adapter->lro)
+			if (!rcd->tcp ||
+			    !adapter->netdev->features & NETIF_F_LRO) {
 				goto not_lro;
+			}
 
 			if (segCnt != 0 && mss != 0) {
 				skb_shinfo(skb)->gso_type = rcd->v4 ?

diff --git a/drivers/net/vmxnet3/vmxnet3_int.h b/drivers/net/vmxnet3/vmxnet3_int.h
index 5ba222920e80..3de4cecda35a 100644
--- a/drivers/net/vmxnet3/vmxnet3_int.h
+++ b/drivers/net/vmxnet3/vmxnet3_int.h

@@ -342,9 +342,6 @@ struct vmxnet3_adapter {
 	u8			__iomem *hw_addr1; /* for BAR 1 */
 	u8                              version;
 
-	bool				rxcsum;
-	bool				lro;
-
 #ifdef VMXNET3_RSS
 	struct UPT1_RSSConf		*rss_conf;
 	bool				rss;

-- 
2.16.2

Re: [PATCH] vmxnet3: fix LRO feature check

From: David Miller <davem@davemloft.net>
Date: 2018-03-18 00:20:33

From: Igor Pylypiv <redacted>
Date: Sat, 17 Mar 2018 00:58:52 -0700

rxcsum and lro fields were deleted in commit a0d2730c9571 ("net: vmxnet3:
convert to hw_features"). With upgrading to newer version those fields were
resurrected and new code started using uninitialized lro field.
Removing rxcsum and lro fields.

Fixes: 45dac1d6ea04 ("vmxnet3: Changes for vmxnet3 adapter version 2 (fwd)")
Signed-off-by: Igor Pylypiv <redacted>

Why are you posting this again?

I applied the copy of this patch which was part of a two part series
posted earlier.

Re: [PATCH] vmxnet3: fix LRO feature check

From: Igor Pylypiv <hidden>
Date: 2018-03-18 01:08:59

The 03/17/2018 20:20, David Miller wrote:

From: Igor Pylypiv <redacted>
Date: Sat, 17 Mar 2018 00:58:52 -0700

quoted

rxcsum and lro fields were deleted in commit a0d2730c9571 ("net: vmxnet3:
convert to hw_features"). With upgrading to newer version those fields were
resurrected and new code started using uninitialized lro field.
Removing rxcsum and lro fields.

Fixes: 45dac1d6ea04 ("vmxnet3: Changes for vmxnet3 adapter version 2 (fwd)")
Signed-off-by: Igor Pylypiv <redacted>

Why are you posting this again?

I applied the copy of this patch which was part of a two part series
posted earlier.

No way!!! 
I know it is hard to believe but I found this bug myself.
That's really odd. I took linux-next-20180316 and patch wasn't there.
Now I see the patch in your tree.

Anyway, let me send another one to delete rxcsum at least...

Re: [PATCH] vmxnet3: fix LRO feature check

From: kbuild test robot <hidden>
Date: 2018-03-18 06:38:02

Hi Igor,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v4.16-rc4]
[also build test WARNING on next-20180316]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Igor-Pylypiv/vmxnet3-fix-LRO-feature-check/20180318-140725
config: i386-randconfig-x003-201811 (attached as .config)
compiler: gcc-7 (Debian 7.3.0-1) 7.3.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

All warnings (new ones prefixed by >>):

   drivers/net/vmxnet3/vmxnet3_drv.c: In function 'vmxnet3_rq_rx_complete':

quoted

drivers/net/vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]

           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~~~~~~~~

vim +1474 drivers/net/vmxnet3/vmxnet3_drv.c

  1255	
  1256	static int
  1257	vmxnet3_rq_rx_complete(struct vmxnet3_rx_queue *rq,
  1258			       struct vmxnet3_adapter *adapter, int quota)
  1259	{
  1260		static const u32 rxprod_reg[2] = {
  1261			VMXNET3_REG_RXPROD, VMXNET3_REG_RXPROD2
  1262		};
  1263		u32 num_pkts = 0;
  1264		bool skip_page_frags = false;
  1265		struct Vmxnet3_RxCompDesc *rcd;
  1266		struct vmxnet3_rx_ctx *ctx = &rq->rx_ctx;
  1267		u16 segCnt = 0, mss = 0;
  1268	#ifdef __BIG_ENDIAN_BITFIELD
  1269		struct Vmxnet3_RxDesc rxCmdDesc;
  1270		struct Vmxnet3_RxCompDesc rxComp;
  1271	#endif
  1272		vmxnet3_getRxComp(rcd, &rq->comp_ring.base[rq->comp_ring.next2proc].rcd,
  1273				  &rxComp);
  1274		while (rcd->gen == rq->comp_ring.gen) {
  1275			struct vmxnet3_rx_buf_info *rbi;
  1276			struct sk_buff *skb, *new_skb = NULL;
  1277			struct page *new_page = NULL;
  1278			dma_addr_t new_dma_addr;
  1279			int num_to_alloc;
  1280			struct Vmxnet3_RxDesc *rxd;
  1281			u32 idx, ring_idx;
  1282			struct vmxnet3_cmd_ring	*ring = NULL;
  1283			if (num_pkts >= quota) {
  1284				/* we may stop even before we see the EOP desc of
  1285				 * the current pkt
  1286				 */
  1287				break;
  1288			}
  1289			BUG_ON(rcd->rqID != rq->qid && rcd->rqID != rq->qid2 &&
  1290			       rcd->rqID != rq->dataRingQid);
  1291			idx = rcd->rxdIdx;
  1292			ring_idx = VMXNET3_GET_RING_IDX(adapter, rcd->rqID);
  1293			ring = rq->rx_ring + ring_idx;
  1294			vmxnet3_getRxDesc(rxd, &rq->rx_ring[ring_idx].base[idx].rxd,
  1295					  &rxCmdDesc);
  1296			rbi = rq->buf_info[ring_idx] + idx;
  1297	
  1298			BUG_ON(rxd->addr != rbi->dma_addr ||
  1299			       rxd->len != rbi->len);
  1300	
  1301			if (unlikely(rcd->eop && rcd->err)) {
  1302				vmxnet3_rx_error(rq, rcd, ctx, adapter);
  1303				goto rcd_done;
  1304			}
  1305	
  1306			if (rcd->sop) { /* first buf of the pkt */
  1307				bool rxDataRingUsed;
  1308				u16 len;
  1309	
  1310				BUG_ON(rxd->btype != VMXNET3_RXD_BTYPE_HEAD ||
  1311				       (rcd->rqID != rq->qid &&
  1312					rcd->rqID != rq->dataRingQid));
  1313	
  1314				BUG_ON(rbi->buf_type != VMXNET3_RX_BUF_SKB);
  1315				BUG_ON(ctx->skb != NULL || rbi->skb == NULL);
  1316	
  1317				if (unlikely(rcd->len == 0)) {
  1318					/* Pretend the rx buffer is skipped. */
  1319					BUG_ON(!(rcd->sop && rcd->eop));
  1320					netdev_dbg(adapter->netdev,
  1321						"rxRing[%u][%u] 0 length\n",
  1322						ring_idx, idx);
  1323					goto rcd_done;
  1324				}
  1325	
  1326				skip_page_frags = false;
  1327				ctx->skb = rbi->skb;
  1328	
  1329				rxDataRingUsed =
  1330					VMXNET3_RX_DATA_RING(adapter, rcd->rqID);
  1331				len = rxDataRingUsed ? rcd->len : rbi->len;
  1332				new_skb = netdev_alloc_skb_ip_align(adapter->netdev,
  1333								    len);
  1334				if (new_skb == NULL) {
  1335					/* Skb allocation failed, do not handover this
  1336					 * skb to stack. Reuse it. Drop the existing pkt
  1337					 */
  1338					rq->stats.rx_buf_alloc_failure++;
  1339					ctx->skb = NULL;
  1340					rq->stats.drop_total++;
  1341					skip_page_frags = true;
  1342					goto rcd_done;
  1343				}
  1344	
  1345				if (rxDataRingUsed) {
  1346					size_t sz;
  1347	
  1348					BUG_ON(rcd->len > rq->data_ring.desc_size);
  1349	
  1350					ctx->skb = new_skb;
  1351					sz = rcd->rxdIdx * rq->data_ring.desc_size;
  1352					memcpy(new_skb->data,
  1353					       &rq->data_ring.base[sz], rcd->len);
  1354				} else {
  1355					ctx->skb = rbi->skb;
  1356	
  1357					new_dma_addr =
  1358						dma_map_single(&adapter->pdev->dev,
  1359							       new_skb->data, rbi->len,
  1360							       PCI_DMA_FROMDEVICE);
  1361					if (dma_mapping_error(&adapter->pdev->dev,
  1362							      new_dma_addr)) {
  1363						dev_kfree_skb(new_skb);
  1364						/* Skb allocation failed, do not
  1365						 * handover this skb to stack. Reuse
  1366						 * it. Drop the existing pkt.
  1367						 */
  1368						rq->stats.rx_buf_alloc_failure++;
  1369						ctx->skb = NULL;
  1370						rq->stats.drop_total++;
  1371						skip_page_frags = true;
  1372						goto rcd_done;
  1373					}
  1374	
  1375					dma_unmap_single(&adapter->pdev->dev,
  1376							 rbi->dma_addr,
  1377							 rbi->len,
  1378							 PCI_DMA_FROMDEVICE);
  1379	
  1380					/* Immediate refill */
  1381					rbi->skb = new_skb;
  1382					rbi->dma_addr = new_dma_addr;
  1383					rxd->addr = cpu_to_le64(rbi->dma_addr);
  1384					rxd->len = rbi->len;
  1385				}
  1386	
  1387	#ifdef VMXNET3_RSS
  1388				if (rcd->rssType != VMXNET3_RCD_RSS_TYPE_NONE &&
  1389				    (adapter->netdev->features & NETIF_F_RXHASH))
  1390					skb_set_hash(ctx->skb,
  1391						     le32_to_cpu(rcd->rssHash),
  1392						     PKT_HASH_TYPE_L3);
  1393	#endif
  1394				skb_put(ctx->skb, rcd->len);
  1395	
  1396				if (VMXNET3_VERSION_GE_2(adapter) &&
  1397				    rcd->type == VMXNET3_CDTYPE_RXCOMP_LRO) {
  1398					struct Vmxnet3_RxCompDescExt *rcdlro;
  1399					rcdlro = (struct Vmxnet3_RxCompDescExt *)rcd;
  1400	
  1401					segCnt = rcdlro->segCnt;
  1402					WARN_ON_ONCE(segCnt == 0);
  1403					mss = rcdlro->mss;
  1404					if (unlikely(segCnt <= 1))
  1405						segCnt = 0;
  1406				} else {
  1407					segCnt = 0;
  1408				}
  1409			} else {
  1410				BUG_ON(ctx->skb == NULL && !skip_page_frags);
  1411	
  1412				/* non SOP buffer must be type 1 in most cases */
  1413				BUG_ON(rbi->buf_type != VMXNET3_RX_BUF_PAGE);
  1414				BUG_ON(rxd->btype != VMXNET3_RXD_BTYPE_BODY);
  1415	
  1416				/* If an sop buffer was dropped, skip all
  1417				 * following non-sop fragments. They will be reused.
  1418				 */
  1419				if (skip_page_frags)
  1420					goto rcd_done;
  1421	
  1422				if (rcd->len) {
  1423					new_page = alloc_page(GFP_ATOMIC);
  1424					/* Replacement page frag could not be allocated.
  1425					 * Reuse this page. Drop the pkt and free the
  1426					 * skb which contained this page as a frag. Skip
  1427					 * processing all the following non-sop frags.
  1428					 */
  1429					if (unlikely(!new_page)) {
  1430						rq->stats.rx_buf_alloc_failure++;
  1431						dev_kfree_skb(ctx->skb);
  1432						ctx->skb = NULL;
  1433						skip_page_frags = true;
  1434						goto rcd_done;
  1435					}
  1436					new_dma_addr = dma_map_page(&adapter->pdev->dev,
  1437								    new_page,
  1438								    0, PAGE_SIZE,
  1439								    PCI_DMA_FROMDEVICE);
  1440					if (dma_mapping_error(&adapter->pdev->dev,
  1441							      new_dma_addr)) {
  1442						put_page(new_page);
  1443						rq->stats.rx_buf_alloc_failure++;
  1444						dev_kfree_skb(ctx->skb);
  1445						ctx->skb = NULL;
  1446						skip_page_frags = true;
  1447						goto rcd_done;
  1448					}
  1449	
  1450					dma_unmap_page(&adapter->pdev->dev,
  1451						       rbi->dma_addr, rbi->len,
  1452						       PCI_DMA_FROMDEVICE);
  1453	
  1454					vmxnet3_append_frag(ctx->skb, rcd, rbi);
  1455	
  1456					/* Immediate refill */
  1457					rbi->page = new_page;
  1458					rbi->dma_addr = new_dma_addr;
  1459					rxd->addr = cpu_to_le64(rbi->dma_addr);
  1460					rxd->len = rbi->len;
  1461				}
  1462			}
  1463	
  1464	
  1465			skb = ctx->skb;
  1466			if (rcd->eop) {
  1467				u32 mtu = adapter->netdev->mtu;
  1468				skb->len += skb->data_len;
  1469	
  1470				vmxnet3_rx_csum(adapter, skb,
  1471						(union Vmxnet3_GenericDesc *)rcd);
  1472				skb->protocol = eth_type_trans(skb, adapter->netdev);
  1473				if (!rcd->tcp ||

1474				    !adapter->netdev->features & NETIF_F_LRO) {

  1475					goto not_lro;
  1476				}
  1477	
  1478				if (segCnt != 0 && mss != 0) {
  1479					skb_shinfo(skb)->gso_type = rcd->v4 ?
  1480						SKB_GSO_TCPV4 : SKB_GSO_TCPV6;
  1481					skb_shinfo(skb)->gso_size = mss;
  1482					skb_shinfo(skb)->gso_segs = segCnt;
  1483				} else if (segCnt != 0 || skb->len > mtu) {
  1484					u32 hlen;
  1485	
  1486					hlen = vmxnet3_get_hdr_len(adapter, skb,
  1487						(union Vmxnet3_GenericDesc *)rcd);
  1488					if (hlen == 0)
  1489						goto not_lro;
  1490	
  1491					skb_shinfo(skb)->gso_type =
  1492						rcd->v4 ? SKB_GSO_TCPV4 : SKB_GSO_TCPV6;
  1493					if (segCnt != 0) {
  1494						skb_shinfo(skb)->gso_segs = segCnt;
  1495						skb_shinfo(skb)->gso_size =
  1496							DIV_ROUND_UP(skb->len -
  1497								hlen, segCnt);
  1498					} else {
  1499						skb_shinfo(skb)->gso_size = mtu - hlen;
  1500					}
  1501				}
  1502	not_lro:
  1503				if (unlikely(rcd->ts))
  1504					__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), rcd->tci);
  1505	
  1506				if (adapter->netdev->features & NETIF_F_LRO)
  1507					netif_receive_skb(skb);
  1508				else
  1509					napi_gro_receive(&rq->napi, skb);
  1510	
  1511				ctx->skb = NULL;
  1512				num_pkts++;
  1513			}
  1514	
  1515	rcd_done:
  1516			/* device may have skipped some rx descs */
  1517			ring->next2comp = idx;
  1518			num_to_alloc = vmxnet3_cmd_ring_desc_avail(ring);
  1519			ring = rq->rx_ring + ring_idx;
  1520			while (num_to_alloc) {
  1521				vmxnet3_getRxDesc(rxd, &ring->base[ring->next2fill].rxd,
  1522						  &rxCmdDesc);
  1523				BUG_ON(!rxd->addr);
  1524	
  1525				/* Recv desc is ready to be used by the device */
  1526				rxd->gen = ring->gen;
  1527				vmxnet3_cmd_ring_adv_next2fill(ring);
  1528				num_to_alloc--;
  1529			}
  1530	
  1531			/* if needed, update the register */
  1532			if (unlikely(rq->shared->updateRxProd)) {
  1533				VMXNET3_WRITE_BAR0_REG(adapter,
  1534						       rxprod_reg[ring_idx] + rq->qid * 8,
  1535						       ring->next2fill);
  1536			}
  1537	
  1538			vmxnet3_comp_ring_adv_next2proc(&rq->comp_ring);
  1539			vmxnet3_getRxComp(rcd,
  1540					  &rq->comp_ring.base[rq->comp_ring.next2proc].rcd, &rxComp);
  1541		}
  1542	
  1543		return num_pkts;
  1544	}
  1545	

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

Re: [PATCH] vmxnet3: fix LRO feature check

From: kbuild test robot <hidden>
Date: 2018-03-18 11:24:14

Hi Igor,

Thank you for the patch! Perhaps something to improve:

[auto build test WARNING on v4.16-rc4]
[also build test WARNING on next-20180316]
[if your patch is applied to the wrong git tree, please drop us a note to help improve the system]

url:    https://github.com/0day-ci/linux/commits/Igor-Pylypiv/vmxnet3-fix-LRO-feature-check/20180318-140725
config: x86_64-randconfig-s3-03181820 (attached as .config)
compiler: gcc-7 (Debian 7.3.0-1) 7.3.0
reproduce:
        # save the attached .config to linux build tree
        make ARCH=x86_64 

All warnings (new ones prefixed by >>):

   In file included from include/linux/kernel.h:10:0,
                    from include/linux/list.h:9,
                    from include/linux/module.h:9,
                    from drivers/net//vmxnet3/vmxnet3_drv.c:27:
   drivers/net//vmxnet3/vmxnet3_drv.c: In function 'vmxnet3_rq_rx_complete':
   drivers/net//vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]
           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~
   include/linux/compiler.h:58:30: note: in definition of macro '__trace_if'
     if (__builtin_constant_p(!!(cond)) ? !!(cond) :   \
                                 ^~~~

quoted

drivers/net//vmxnet3/vmxnet3_drv.c:1473:4: note: in expansion of macro 'if'

       if (!rcd->tcp ||
       ^~
   drivers/net//vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]
           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~
   include/linux/compiler.h:58:42: note: in definition of macro '__trace_if'
     if (__builtin_constant_p(!!(cond)) ? !!(cond) :   \
                                             ^~~~

quoted

drivers/net//vmxnet3/vmxnet3_drv.c:1473:4: note: in expansion of macro 'if'

       if (!rcd->tcp ||
       ^~
   drivers/net//vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]
           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~
   include/linux/compiler.h:69:16: note: in definition of macro '__trace_if'
      ______r = !!(cond);     \
                   ^~~~

quoted

drivers/net//vmxnet3/vmxnet3_drv.c:1473:4: note: in expansion of macro 'if'

       if (!rcd->tcp ||
       ^~

vim +/if +1473 drivers/net//vmxnet3/vmxnet3_drv.c

  1255	
  1256	static int
  1257	vmxnet3_rq_rx_complete(struct vmxnet3_rx_queue *rq,
  1258			       struct vmxnet3_adapter *adapter, int quota)
  1259	{
  1260		static const u32 rxprod_reg[2] = {
  1261			VMXNET3_REG_RXPROD, VMXNET3_REG_RXPROD2
  1262		};
  1263		u32 num_pkts = 0;
  1264		bool skip_page_frags = false;
  1265		struct Vmxnet3_RxCompDesc *rcd;
  1266		struct vmxnet3_rx_ctx *ctx = &rq->rx_ctx;
  1267		u16 segCnt = 0, mss = 0;
  1268	#ifdef __BIG_ENDIAN_BITFIELD
  1269		struct Vmxnet3_RxDesc rxCmdDesc;
  1270		struct Vmxnet3_RxCompDesc rxComp;
  1271	#endif
  1272		vmxnet3_getRxComp(rcd, &rq->comp_ring.base[rq->comp_ring.next2proc].rcd,
  1273				  &rxComp);
  1274		while (rcd->gen == rq->comp_ring.gen) {
  1275			struct vmxnet3_rx_buf_info *rbi;
  1276			struct sk_buff *skb, *new_skb = NULL;
  1277			struct page *new_page = NULL;
  1278			dma_addr_t new_dma_addr;
  1279			int num_to_alloc;
  1280			struct Vmxnet3_RxDesc *rxd;
  1281			u32 idx, ring_idx;
  1282			struct vmxnet3_cmd_ring	*ring = NULL;
  1283			if (num_pkts >= quota) {
  1284				/* we may stop even before we see the EOP desc of
  1285				 * the current pkt
  1286				 */
  1287				break;
  1288			}
  1289			BUG_ON(rcd->rqID != rq->qid && rcd->rqID != rq->qid2 &&
  1290			       rcd->rqID != rq->dataRingQid);
  1291			idx = rcd->rxdIdx;
  1292			ring_idx = VMXNET3_GET_RING_IDX(adapter, rcd->rqID);
  1293			ring = rq->rx_ring + ring_idx;
  1294			vmxnet3_getRxDesc(rxd, &rq->rx_ring[ring_idx].base[idx].rxd,
  1295					  &rxCmdDesc);
  1296			rbi = rq->buf_info[ring_idx] + idx;
  1297	
  1298			BUG_ON(rxd->addr != rbi->dma_addr ||
  1299			       rxd->len != rbi->len);
  1300	
  1301			if (unlikely(rcd->eop && rcd->err)) {
  1302				vmxnet3_rx_error(rq, rcd, ctx, adapter);
  1303				goto rcd_done;
  1304			}
  1305	
  1306			if (rcd->sop) { /* first buf of the pkt */
  1307				bool rxDataRingUsed;
  1308				u16 len;
  1309	
  1310				BUG_ON(rxd->btype != VMXNET3_RXD_BTYPE_HEAD ||
  1311				       (rcd->rqID != rq->qid &&
  1312					rcd->rqID != rq->dataRingQid));
  1313	
  1314				BUG_ON(rbi->buf_type != VMXNET3_RX_BUF_SKB);
  1315				BUG_ON(ctx->skb != NULL || rbi->skb == NULL);
  1316	
  1317				if (unlikely(rcd->len == 0)) {
  1318					/* Pretend the rx buffer is skipped. */
  1319					BUG_ON(!(rcd->sop && rcd->eop));
  1320					netdev_dbg(adapter->netdev,
  1321						"rxRing[%u][%u] 0 length\n",
  1322						ring_idx, idx);
  1323					goto rcd_done;
  1324				}
  1325	
  1326				skip_page_frags = false;
  1327				ctx->skb = rbi->skb;
  1328	
  1329				rxDataRingUsed =
  1330					VMXNET3_RX_DATA_RING(adapter, rcd->rqID);
  1331				len = rxDataRingUsed ? rcd->len : rbi->len;
  1332				new_skb = netdev_alloc_skb_ip_align(adapter->netdev,
  1333								    len);
  1334				if (new_skb == NULL) {
  1335					/* Skb allocation failed, do not handover this
  1336					 * skb to stack. Reuse it. Drop the existing pkt
  1337					 */
  1338					rq->stats.rx_buf_alloc_failure++;
  1339					ctx->skb = NULL;
  1340					rq->stats.drop_total++;
  1341					skip_page_frags = true;
  1342					goto rcd_done;
  1343				}
  1344	
  1345				if (rxDataRingUsed) {
  1346					size_t sz;
  1347	
  1348					BUG_ON(rcd->len > rq->data_ring.desc_size);
  1349	
  1350					ctx->skb = new_skb;
  1351					sz = rcd->rxdIdx * rq->data_ring.desc_size;
  1352					memcpy(new_skb->data,
  1353					       &rq->data_ring.base[sz], rcd->len);
  1354				} else {
  1355					ctx->skb = rbi->skb;
  1356	
  1357					new_dma_addr =
  1358						dma_map_single(&adapter->pdev->dev,
  1359							       new_skb->data, rbi->len,
  1360							       PCI_DMA_FROMDEVICE);
  1361					if (dma_mapping_error(&adapter->pdev->dev,
  1362							      new_dma_addr)) {
  1363						dev_kfree_skb(new_skb);
  1364						/* Skb allocation failed, do not
  1365						 * handover this skb to stack. Reuse
  1366						 * it. Drop the existing pkt.
  1367						 */
  1368						rq->stats.rx_buf_alloc_failure++;
  1369						ctx->skb = NULL;
  1370						rq->stats.drop_total++;
  1371						skip_page_frags = true;
  1372						goto rcd_done;
  1373					}
  1374	
  1375					dma_unmap_single(&adapter->pdev->dev,
  1376							 rbi->dma_addr,
  1377							 rbi->len,
  1378							 PCI_DMA_FROMDEVICE);
  1379	
  1380					/* Immediate refill */
  1381					rbi->skb = new_skb;
  1382					rbi->dma_addr = new_dma_addr;
  1383					rxd->addr = cpu_to_le64(rbi->dma_addr);
  1384					rxd->len = rbi->len;
  1385				}
  1386	
  1387	#ifdef VMXNET3_RSS
  1388				if (rcd->rssType != VMXNET3_RCD_RSS_TYPE_NONE &&
  1389				    (adapter->netdev->features & NETIF_F_RXHASH))
  1390					skb_set_hash(ctx->skb,
  1391						     le32_to_cpu(rcd->rssHash),
  1392						     PKT_HASH_TYPE_L3);
  1393	#endif
  1394				skb_put(ctx->skb, rcd->len);
  1395	
  1396				if (VMXNET3_VERSION_GE_2(adapter) &&
  1397				    rcd->type == VMXNET3_CDTYPE_RXCOMP_LRO) {
  1398					struct Vmxnet3_RxCompDescExt *rcdlro;
  1399					rcdlro = (struct Vmxnet3_RxCompDescExt *)rcd;
  1400	
  1401					segCnt = rcdlro->segCnt;
  1402					WARN_ON_ONCE(segCnt == 0);
  1403					mss = rcdlro->mss;
  1404					if (unlikely(segCnt <= 1))
  1405						segCnt = 0;
  1406				} else {
  1407					segCnt = 0;
  1408				}
  1409			} else {
  1410				BUG_ON(ctx->skb == NULL && !skip_page_frags);
  1411	
  1412				/* non SOP buffer must be type 1 in most cases */
  1413				BUG_ON(rbi->buf_type != VMXNET3_RX_BUF_PAGE);
  1414				BUG_ON(rxd->btype != VMXNET3_RXD_BTYPE_BODY);
  1415	
  1416				/* If an sop buffer was dropped, skip all
  1417				 * following non-sop fragments. They will be reused.
  1418				 */
  1419				if (skip_page_frags)
  1420					goto rcd_done;
  1421	
  1422				if (rcd->len) {
  1423					new_page = alloc_page(GFP_ATOMIC);
  1424					/* Replacement page frag could not be allocated.
  1425					 * Reuse this page. Drop the pkt and free the
  1426					 * skb which contained this page as a frag. Skip
  1427					 * processing all the following non-sop frags.
  1428					 */
  1429					if (unlikely(!new_page)) {
  1430						rq->stats.rx_buf_alloc_failure++;
  1431						dev_kfree_skb(ctx->skb);
  1432						ctx->skb = NULL;
  1433						skip_page_frags = true;
  1434						goto rcd_done;
  1435					}
  1436					new_dma_addr = dma_map_page(&adapter->pdev->dev,
  1437								    new_page,
  1438								    0, PAGE_SIZE,
  1439								    PCI_DMA_FROMDEVICE);
  1440					if (dma_mapping_error(&adapter->pdev->dev,
  1441							      new_dma_addr)) {
  1442						put_page(new_page);
  1443						rq->stats.rx_buf_alloc_failure++;
  1444						dev_kfree_skb(ctx->skb);
  1445						ctx->skb = NULL;
  1446						skip_page_frags = true;
  1447						goto rcd_done;
  1448					}
  1449	
  1450					dma_unmap_page(&adapter->pdev->dev,
  1451						       rbi->dma_addr, rbi->len,
  1452						       PCI_DMA_FROMDEVICE);
  1453	
  1454					vmxnet3_append_frag(ctx->skb, rcd, rbi);
  1455	
  1456					/* Immediate refill */
  1457					rbi->page = new_page;
  1458					rbi->dma_addr = new_dma_addr;
  1459					rxd->addr = cpu_to_le64(rbi->dma_addr);
  1460					rxd->len = rbi->len;
  1461				}
  1462			}
  1463	
  1464	
  1465			skb = ctx->skb;
  1466			if (rcd->eop) {
  1467				u32 mtu = adapter->netdev->mtu;
  1468				skb->len += skb->data_len;
  1469	
  1470				vmxnet3_rx_csum(adapter, skb,
  1471						(union Vmxnet3_GenericDesc *)rcd);
  1472				skb->protocol = eth_type_trans(skb, adapter->netdev);

1473				if (!rcd->tcp ||

  1474				    !adapter->netdev->features & NETIF_F_LRO) {
  1475					goto not_lro;
  1476				}
  1477	
  1478				if (segCnt != 0 && mss != 0) {
  1479					skb_shinfo(skb)->gso_type = rcd->v4 ?
  1480						SKB_GSO_TCPV4 : SKB_GSO_TCPV6;
  1481					skb_shinfo(skb)->gso_size = mss;
  1482					skb_shinfo(skb)->gso_segs = segCnt;
  1483				} else if (segCnt != 0 || skb->len > mtu) {
  1484					u32 hlen;
  1485	
  1486					hlen = vmxnet3_get_hdr_len(adapter, skb,
  1487						(union Vmxnet3_GenericDesc *)rcd);
  1488					if (hlen == 0)
  1489						goto not_lro;
  1490	
  1491					skb_shinfo(skb)->gso_type =
  1492						rcd->v4 ? SKB_GSO_TCPV4 : SKB_GSO_TCPV6;
  1493					if (segCnt != 0) {
  1494						skb_shinfo(skb)->gso_segs = segCnt;
  1495						skb_shinfo(skb)->gso_size =
  1496							DIV_ROUND_UP(skb->len -
  1497								hlen, segCnt);
  1498					} else {
  1499						skb_shinfo(skb)->gso_size = mtu - hlen;
  1500					}
  1501				}
  1502	not_lro:
  1503				if (unlikely(rcd->ts))
  1504					__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), rcd->tci);
  1505	
  1506				if (adapter->netdev->features & NETIF_F_LRO)
  1507					netif_receive_skb(skb);
  1508				else
  1509					napi_gro_receive(&rq->napi, skb);
  1510	
  1511				ctx->skb = NULL;
  1512				num_pkts++;
  1513			}
  1514	
  1515	rcd_done:
  1516			/* device may have skipped some rx descs */
  1517			ring->next2comp = idx;
  1518			num_to_alloc = vmxnet3_cmd_ring_desc_avail(ring);
  1519			ring = rq->rx_ring + ring_idx;
  1520			while (num_to_alloc) {
  1521				vmxnet3_getRxDesc(rxd, &ring->base[ring->next2fill].rxd,
  1522						  &rxCmdDesc);
  1523				BUG_ON(!rxd->addr);
  1524	
  1525				/* Recv desc is ready to be used by the device */
  1526				rxd->gen = ring->gen;
  1527				vmxnet3_cmd_ring_adv_next2fill(ring);
  1528				num_to_alloc--;
  1529			}
  1530	
  1531			/* if needed, update the register */
  1532			if (unlikely(rq->shared->updateRxProd)) {
  1533				VMXNET3_WRITE_BAR0_REG(adapter,
  1534						       rxprod_reg[ring_idx] + rq->qid * 8,
  1535						       ring->next2fill);
  1536			}
  1537	
  1538			vmxnet3_comp_ring_adv_next2proc(&rq->comp_ring);
  1539			vmxnet3_getRxComp(rcd,
  1540					  &rq->comp_ring.base[rq->comp_ring.next2proc].rcd, &rxComp);
  1541		}
  1542	
  1543		return num_pkts;
  1544	}
  1545	

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

Re: [PATCH] vmxnet3: fix LRO feature check

From: David Miller <davem@davemloft.net>
Date: 2018-03-20 14:57:24

From: kbuild test robot <redacted>
Date: Sun, 18 Mar 2018 14:37:35 +0800

All warnings (new ones prefixed by >>):

   drivers/net/vmxnet3/vmxnet3_drv.c: In function 'vmxnet3_rq_rx_complete':

quoted

drivers/net/vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]

           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~~~~~~~~

Igor, I will fix this up for you.  But it is clear that this patch wasn't tested
very well.

Because !adapter->netdev->features evaluates wholly before the "& NETIF_F_LRO",
the flags aren't being tested properly at all.

Re: [PATCH] vmxnet3: fix LRO feature check

From: Igor Pylypiv <hidden>
Date: 2018-03-20 18:24:52

The 03/20/2018 10:57, David Miller wrote:

From: kbuild test robot <redacted>
Date: Sun, 18 Mar 2018 14:37:35 +0800

quoted

All warnings (new ones prefixed by >>):

   drivers/net/vmxnet3/vmxnet3_drv.c: In function 'vmxnet3_rq_rx_complete':

quoted

drivers/net/vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]

           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~~~~~~~~

Igor, I will fix this up for you.  But it is clear that this patch wasn't tested
very well.

Because !adapter->netdev->features evaluates wholly before the "& NETIF_F_LRO",
the flags aren't being tested properly at all.

My bad.

I have even been looking at C operator precedence table:

----------------------------
| Operator | Associativity |
|--------------------------|
| ++ --    | right-to-left |
| + -      |               |
| ! ~      |               |
| (type)   |               |
| *        |               |
| &        |               |
| sizeof   |               |
----------------------------

According to this table '&' will be evaluated first, because it is on the right side.

But yeah, that was "Address of", not "Bitwise AND".

Re: [PATCH] vmxnet3: fix LRO feature check

From: Shrikrishna Khare <hidden>
Date: 2018-03-20 18:57:50


On Tue, 20 Mar 2018, David Miller wrote:

From: kbuild test robot <redacted>
Date: Sun, 18 Mar 2018 14:37:35 +0800

quoted

All warnings (new ones prefixed by >>):

   drivers/net/vmxnet3/vmxnet3_drv.c: In function 'vmxnet3_rq_rx_complete':

quoted

drivers/net/vmxnet3/vmxnet3_drv.c:1474:8: warning: suggest parentheses around operand of '!' or change '&' to '&&' or '!' to '~' [-Wparentheses]

           !adapter->netdev->features & NETIF_F_LRO) {
           ^~~~~~~~~~~~~~~~~~~~~~~~~~

Igor, I will fix this up for you.  But it is clear that this patch wasn't tested
very well.

Because !adapter->netdev->features evaluates wholly before the "& NETIF_F_LRO",
the flags aren't being tested properly at all.

Ronak's patch that fixes this issue has the right code, and is already 
accepted in net tree (Commit: 034f405793897a3c8f642935f5494b86c340cde7). 
We no longer need Igor's patch.

Thanks,
Shri

Re: [PATCH] vmxnet3: fix LRO feature check

From: David Miller <davem@davemloft.net>
Date: 2018-03-20 18:59:05

From: Shrikrishna Khare <redacted>
Date: Tue, 20 Mar 2018 11:42:47 -0700

Ronak's patch that fixes this issue has the right code, and is already 
accepted in net tree (Commit: 034f405793897a3c8f642935f5494b86c340cde7). 
We no longer need Igor's patch.

That explains why things in my actual tree are fine.

Thanks.

`h`	back out one level
`j`	next message in thread
`k`	previous message in thread
`l`	drill in
`Esc`	close help / fold thread tree
`?`	toggle this help