From: Maxime Coquelin <maxime.coquelin@redhat.com>
To: Nicolas Chautru <nicolas.chautru@intel.com>, dev@dpdk.org
Cc: hernan.vargas@intel.com, stable@dpdk.org
Subject: Re: [PATCH v1 5/9] baseband/acc: prevent to dequeue more than requested
Date: Fri, 10 Feb 2023 10:39:04 +0100 [thread overview]
Message-ID: <a96102c6-9765-9395-a462-389c52ef1ffb@redhat.com> (raw)
In-Reply-To: <20230209221929.265059-6-nicolas.chautru@intel.com>
On 2/9/23 23:19, Nicolas Chautru wrote:
> Add support for corner-case when more operations are
> requested than expected, in the case of encoder muxing
> operations.
>
> Fixes: e640f6cdfa84 ("baseband/acc200: add LDPC processing")
> Cc: stable@dpdk.org
>
> Signed-off-by: Nicolas Chautru <nicolas.chautru@intel.com>
> ---
> drivers/baseband/acc/rte_vrb_pmd.c | 27 +++++++++++++++------------
> 1 file changed, 15 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/baseband/acc/rte_vrb_pmd.c b/drivers/baseband/acc/rte_vrb_pmd.c
> index 8540e3d31c..b251ad25c6 100644
> --- a/drivers/baseband/acc/rte_vrb_pmd.c
> +++ b/drivers/baseband/acc/rte_vrb_pmd.c
> @@ -2641,7 +2641,8 @@ vrb_enqueue_ldpc_dec(struct rte_bbdev_queue_data *q_data,
> /* Dequeue one encode operations from device in CB mode. */
> static inline int
> vrb_dequeue_enc_one_op_cb(struct acc_queue *q, struct rte_bbdev_enc_op **ref_op,
> - uint16_t *dequeued_ops, uint32_t *aq_dequeued, uint16_t *dequeued_descs)
> + uint16_t *dequeued_ops, uint32_t *aq_dequeued, uint16_t *dequeued_descs,
> + uint16_t max_requested_ops)
> {
> union acc_dma_desc *desc, atom_desc;
> union acc_dma_rsp_desc rsp;
> @@ -2654,6 +2655,9 @@ vrb_dequeue_enc_one_op_cb(struct acc_queue *q, struct rte_bbdev_enc_op **ref_op,
> desc = q->ring_addr + desc_idx;
> atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, __ATOMIC_RELAXED);
>
> + if (*dequeued_ops + desc->req.numCBs > max_requested_ops)
> + return -1;
> +
> /* Check fdone bit. */
> if (!(atom_desc.rsp.val & ACC_FDONE))
> return -1;
> @@ -2695,7 +2699,7 @@ vrb_dequeue_enc_one_op_cb(struct acc_queue *q, struct rte_bbdev_enc_op **ref_op,
> static inline int
> vrb_dequeue_enc_one_op_tb(struct acc_queue *q, struct rte_bbdev_enc_op **ref_op,
> uint16_t *dequeued_ops, uint32_t *aq_dequeued,
> - uint16_t *dequeued_descs)
> + uint16_t *dequeued_descs, uint16_t max_requested_ops)
> {
> union acc_dma_desc *desc, *last_desc, atom_desc;
> union acc_dma_rsp_desc rsp;
> @@ -2706,6 +2710,9 @@ vrb_dequeue_enc_one_op_tb(struct acc_queue *q, struct rte_bbdev_enc_op **ref_op,
> desc = acc_desc_tail(q, *dequeued_descs);
> atom_desc.atom_hdr = __atomic_load_n((uint64_t *)desc, __ATOMIC_RELAXED);
>
> + if (*dequeued_ops + 1 > max_requested_ops)
> + return -1;
> +
> /* Check fdone bit. */
> if (!(atom_desc.rsp.val & ACC_FDONE))
> return -1;
> @@ -2966,25 +2973,23 @@ vrb_dequeue_enc(struct rte_bbdev_queue_data *q_data,
>
> cbm = op->turbo_enc.code_block_mode;
>
> - for (i = 0; i < num; i++) {
> + for (i = 0; i < avail; i++) {
> if (cbm == RTE_BBDEV_TRANSPORT_BLOCK)
> ret = vrb_dequeue_enc_one_op_tb(q, &ops[dequeued_ops],
> &dequeued_ops, &aq_dequeued,
> - &dequeued_descs);
> + &dequeued_descs, num);
> else
> ret = vrb_dequeue_enc_one_op_cb(q, &ops[dequeued_ops],
> &dequeued_ops, &aq_dequeued,
> - &dequeued_descs);
> + &dequeued_descs, num);
> if (ret < 0)
> break;
> - if (dequeued_ops >= num)
> - break;
> }
>
> q->aq_dequeued += aq_dequeued;
> q->sw_ring_tail += dequeued_descs;
>
> - /* Update enqueue stats */
> + /* Update enqueue stats. */
> q_data->queue_stats.dequeued_count += dequeued_ops;
>
> return dequeued_ops;
> @@ -3010,15 +3015,13 @@ vrb_dequeue_ldpc_enc(struct rte_bbdev_queue_data *q_data,
> if (cbm == RTE_BBDEV_TRANSPORT_BLOCK)
> ret = vrb_dequeue_enc_one_op_tb(q, &ops[dequeued_ops],
> &dequeued_ops, &aq_dequeued,
> - &dequeued_descs);
> + &dequeued_descs, num);
> else
> ret = vrb_dequeue_enc_one_op_cb(q, &ops[dequeued_ops],
> &dequeued_ops, &aq_dequeued,
> - &dequeued_descs);
> + &dequeued_descs, num);
> if (ret < 0)
> break;
> - if (dequeued_ops >= num)
> - break;
> }
>
> q->aq_dequeued += aq_dequeued;
Reviewed-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Thanks,
Maxime
next prev parent reply other threads:[~2023-02-10 9:39 UTC|newest]
Thread overview: 31+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20230209221929.265059-1-nicolas.chautru@intel.com>
2023-02-09 22:19 ` [PATCH v1 1/9] baseband/acc: protection for TB negative scenario Nicolas Chautru
2023-02-10 8:17 ` Maxime Coquelin
2023-02-10 17:28 ` Chautru, Nicolas
2023-02-10 17:58 ` [PATCH v2 0/9] baseband/acc: VRB PMD fixes Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 1/9] baseband/acc: protection for TB negative scenario Nicolas Chautru
2023-02-23 8:56 ` Maxime Coquelin
2023-02-23 17:25 ` Chautru, Nicolas
2023-02-10 17:58 ` [PATCH v2 2/9] baseband/acc: remove interrupt support on VRB1 Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 3/9] baseband/acc: add explicit mbuf append for soft output Nicolas Chautru
2023-02-22 11:20 ` Maxime Coquelin
2023-02-10 17:58 ` [PATCH v2 4/9] baseband/acc: prevent to dequeue more than requested Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 5/9] baseband/acc: fix iteration counter in TB mode Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 6/9] baseband/acc: fix potential arithmetic overflow Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 7/9] baseband/acc: add support for 4GUL CRC drop in VRB PMD Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 8/9] baseband/acc: add support for 4GUL with SO and TB Nicolas Chautru
2023-02-10 17:58 ` [PATCH v2 9/9] baseband/acc: remove printf from PMD function Nicolas Chautru
2023-02-23 9:28 ` [PATCH v2 0/9] baseband/acc: VRB PMD fixes Maxime Coquelin
2023-02-09 22:19 ` [PATCH v1 2/9] baseband/acc: add support for 4GUL with SO and TB Nicolas Chautru
2023-02-10 8:30 ` Maxime Coquelin
2023-02-10 17:29 ` Chautru, Nicolas
2023-02-09 22:19 ` [PATCH v1 3/9] baseband/acc: remove interrupt support on VRB1 Nicolas Chautru
2023-02-10 8:31 ` Maxime Coquelin
2023-02-09 22:19 ` [PATCH v1 4/9] baseband/acc: add explicit mbuf apend for soft output Nicolas Chautru
2023-02-10 8:40 ` Maxime Coquelin
2023-02-10 17:35 ` Chautru, Nicolas
2023-02-09 22:19 ` [PATCH v1 5/9] baseband/acc: prevent to dequeue more than requested Nicolas Chautru
2023-02-10 9:39 ` Maxime Coquelin [this message]
2023-02-09 22:19 ` [PATCH v1 6/9] baseband/acc: fix iteration counter in TB mode Nicolas Chautru
2023-02-10 9:40 ` Maxime Coquelin
2023-02-09 22:19 ` [PATCH v1 7/9] baseband/acc: fix potential arithmetic overflow Fix potential issue of overflow causing coverity warning Nicolas Chautru
2023-02-10 9:41 ` Maxime Coquelin
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=a96102c6-9765-9395-a462-389c52ef1ffb@redhat.com \
--to=maxime.coquelin@redhat.com \
--cc=dev@dpdk.org \
--cc=hernan.vargas@intel.com \
--cc=nicolas.chautru@intel.com \
--cc=stable@dpdk.org \
/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).