From: Tony Lu <zlu@ezchip.com>
To: 'Liming Sun' <lsun@ezchip.com>, <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits
Date: Fri, 8 Jan 2016 15:49:28 +0800 [thread overview]
Message-ID: <004b01d149e9$27c3dc50$774b94f0$@com> (raw)
In-Reply-To: <1450813790-11874-3-git-send-email-lsun@ezchip.com>
>-----Original Message-----
>From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Liming Sun
>Sent: Wednesday, December 23, 2015 3:50 AM
>To: dev@dpdk.org
>Subject: [dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue
when
>testpmd quits
>
>1. Fixed the compiling issue of the ethtool example on tilegx
> platform.
>2. Fixed the hung/crash issue when quitting testpmd under high
> traffic rate. The buffer error bit needs to be checked before
> processing the idesc and releasing the buffer. Code logic is
> also simplified.
>
>Signed-off-by: Liming Sun <lsun@ezchip.com>
>---
> drivers/net/mpipe/mpipe_tilegx.c | 46
>+++++++-------------
> .../common/include/arch/tile/rte_rwlock.h | 1 +
> 2 files changed, 17 insertions(+), 30 deletions(-)
>
>diff --git a/drivers/net/mpipe/mpipe_tilegx.c
>b/drivers/net/mpipe/mpipe_tilegx.c
>index 8d006fa..4cb54c3 100644
>--- a/drivers/net/mpipe/mpipe_tilegx.c
>+++ b/drivers/net/mpipe/mpipe_tilegx.c
>@@ -134,7 +134,6 @@ struct mpipe_dev_priv {
> struct rte_mempool *rx_mpool; /* mpool used by the rx queues. */
> unsigned rx_offset; /* Receive head room. */
> unsigned rx_size_code; /* mPIPE rx buffer size code. */
>- unsigned rx_buffers; /* receive buffers on stack. */
> int is_xaui:1, /* Is this an xgbe or gbe? */
> initialized:1, /* Initialized port? */
> running:1; /* Running port? */
>@@ -529,7 +528,6 @@ mpipe_recv_fill_stack(struct mpipe_dev_priv *priv, int
>count)
> mpipe_recv_push(priv, mbuf);
> }
>
>- priv->rx_buffers += count;
> PMD_DEBUG_RX("%s: Filled %d/%d buffers\n", mpipe_name(priv), i,
>count); }
>
>@@ -539,10 +537,9 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
> const int offset = priv->rx_offset & ~RTE_MEMPOOL_ALIGN_MASK;
> uint8_t in_port = priv->port_id;
> struct rte_mbuf *mbuf;
>- unsigned count;
> void *va;
>
>- for (count = 0; count < priv->rx_buffers; count++) {
>+ while (1) {
> va = gxio_mpipe_pop_buffer(priv->context, priv->stack);
> if (!va)
> break;
>@@ -561,10 +558,6 @@ mpipe_recv_flush_stack(struct mpipe_dev_priv *priv)
>
> __rte_mbuf_raw_free(mbuf);
> }
>-
>- PMD_DEBUG_RX("%s: Returned %d/%d buffers\n",
>- mpipe_name(priv), count, priv->rx_buffers);
>- priv->rx_buffers -= count;
> }
>
> static void
>@@ -1246,31 +1239,23 @@ mpipe_recv_flush(struct mpipe_dev_priv *priv)
> gxio_mpipe_iqueue_t *iqueue;
> gxio_mpipe_idesc_t idesc;
> struct rte_mbuf *mbuf;
>- int retries = 0;
> unsigned queue;
>
>- do {
>- mpipe_recv_flush_stack(priv);
>-
>- /* Flush packets sitting in recv queues. */
>- for (queue = 0; queue < priv->nb_rx_queues; queue++) {
>- rx_queue = mpipe_rx_queue(priv, queue);
>- iqueue = &rx_queue->iqueue;
>- while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >=
0) {
>- mbuf = mpipe_recv_mbuf(priv, &idesc,
in_port);
>- rte_pktmbuf_free(mbuf);
>- priv->rx_buffers--;
>- }
>- rte_free(rx_queue->rx_ring_mem);
>- }
>- } while (retries++ < 10 && priv->rx_buffers);
>+ /* Release packets on the buffer stack. */
>+ mpipe_recv_flush_stack(priv);
>
>- if (priv->rx_buffers) {
>- RTE_LOG(ERR, PMD, "%s: Leaked %d receive buffers.\n",
>- mpipe_name(priv), priv->rx_buffers);
>- } else {
>- PMD_DEBUG_RX("%s: Returned all receive buffers.\n",
>- mpipe_name(priv));
>+ /* Flush packets sitting in recv queues. */
>+ for (queue = 0; queue < priv->nb_rx_queues; queue++) {
>+ rx_queue = mpipe_rx_queue(priv, queue);
>+ iqueue = &rx_queue->iqueue;
>+ while (gxio_mpipe_iqueue_try_get(iqueue, &idesc) >= 0) {
>+ /* Skip idesc with the 'buffer error' bit set. */
>+ if (idesc.be)
>+ continue;
>+ mbuf = mpipe_recv_mbuf(priv, &idesc, in_port);
>+ rte_pktmbuf_free(mbuf);
>+ }
>+ rte_free(rx_queue->rx_ring_mem);
> }
> }
>
>@@ -1339,6 +1324,7 @@ mpipe_do_xmit(struct mpipe_tx_queue *tx_queue,
>struct rte_mbuf **tx_pkts,
> .xfer_size = rte_pktmbuf_data_len(mbuf),
> .bound = next ? 0 : 1,
> .stack_idx = mpipe_mbuf_stack_index(priv,
mbuf),
>+ .size = priv->rx_size_code,
> } };
> if (mpipe_local.mbuf_push_debt[port_id] > 0) {
> mpipe_local.mbuf_push_debt[port_id]--;
>diff --git a/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
>b/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
>index 8f67a19..6d609e8 100644
>--- a/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
>+++ b/lib/librte_eal/common/include/arch/tile/rte_rwlock.h
>@@ -38,6 +38,7 @@ extern "C" {
> #endif
>
> #include "generic/rte_rwlock.h"
>+#include <rte_spinlock.h>
>
> static inline void
> rte_rwlock_read_lock_tm(rte_rwlock_t *rwl)
>--
>1.7.1
Acked-by: Zhigang Lu <zlu@ezchip.com>
next prev parent reply other threads:[~2016-01-08 7:49 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-12-22 19:49 [dpdk-dev] [PATCH 0/2] Some enhancement of the tilegx mpipe driver Liming Sun
2015-12-22 19:49 ` [dpdk-dev] [PATCH 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Liming Sun
2016-01-08 7:41 ` Tony Lu
2015-12-22 19:49 ` [dpdk-dev] [PATCH 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits Liming Sun
2016-01-08 7:49 ` Tony Lu [this message]
2016-01-08 14:39 ` [dpdk-dev] [PATCH v2 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Liming Sun
2016-01-08 14:39 ` [dpdk-dev] [PATCH v2 2/2] driver/net/mpipe: fix the crash/hung issue when testpmd quits Liming Sun
2016-02-09 15:59 ` Bruce Richardson
2016-02-10 14:50 ` Liming Sun
2016-02-10 5:15 ` [dpdk-dev] [PATCH v3] " Liming Sun
2016-02-26 17:35 ` Bruce Richardson
2016-02-09 15:55 ` [dpdk-dev] [PATCH v2 1/2] driver/net/mpipe: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Bruce Richardson
2016-02-09 18:40 ` Liming Sun
2016-02-10 14:52 ` Liming Sun
2016-02-10 3:46 ` [dpdk-dev] [PATCH v3 1/2] eal/tile: " Liming Sun
2016-02-10 3:47 ` [dpdk-dev] [PATCH v3 2/2] driver/net/mpipe: some code optimization Liming Sun
2016-02-10 4:04 ` [dpdk-dev] [PATCH v4 1/2] eal/tile: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Liming Sun
2016-02-10 4:04 ` [dpdk-dev] [PATCH v4 2/2] driver/net/mpipe: some code optimization Liming Sun
2016-03-08 19:59 ` [dpdk-dev] [PATCH v4 1/2] eal/tile: add rte_vect.h and enable CONFIG_RTE_LIBRTE_LPM Thomas Monjalon
2016-03-17 14:04 ` Thomas Monjalon
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='004b01d149e9$27c3dc50$774b94f0$@com' \
--to=zlu@ezchip.com \
--cc=dev@dpdk.org \
--cc=lsun@ezchip.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).