From: Amit Prakash Shukla <amitprakashs@marvell.com>
To: Vamsi Attunuru <vattunuru@marvell.com>
Cc: <dev@dpdk.org>, <jerinj@marvell.com>, <stable@dpdk.org>
Subject: [PATCH v5 11/12] dma/cnxk: add completion ring tail wrap check
Date: Wed, 23 Aug 2023 16:45:24 +0530 [thread overview]
Message-ID: <20230823111525.3975662-11-amitprakashs@marvell.com> (raw)
In-Reply-To: <20230823111525.3975662-1-amitprakashs@marvell.com>
From: Vamsi Attunuru <vattunuru@marvell.com>
Adds a check to avoid tail wrap when completion desc ring
is full. Also patch increase max desc size to 2048.
Fixes: b56f1e2dad38 ("dma/cnxk: add channel operations")
Fixes: 3340c3e22783 ("dma/cnxk: add scatter-gather copy")
Fixes: 681851b347ad ("dma/cnxk: support CN10K DMA engine")
Cc: stable@dpdk.org
Signed-off-by: Vamsi Attunuru <vattunuru@marvell.com>
---
v2:
- Fix for bugs observed in v1.
- Squashed few commits.
v3:
- Resolved review suggestions.
- Code improvement.
v4:
- Resolved checkpatch warnings.
v5:
- Updated commit message.
- Split the commits.
drivers/dma/cnxk/cnxk_dmadev.c | 22 ++++++++++++++++++++--
drivers/dma/cnxk/cnxk_dmadev.h | 2 +-
2 files changed, 21 insertions(+), 3 deletions(-)
diff --git a/drivers/dma/cnxk/cnxk_dmadev.c b/drivers/dma/cnxk/cnxk_dmadev.c
index 0b77543f6a..89ff4c18ac 100644
--- a/drivers/dma/cnxk/cnxk_dmadev.c
+++ b/drivers/dma/cnxk/cnxk_dmadev.c
@@ -434,6 +434,11 @@ cnxk_dmadev_copy(void *dev_private, uint16_t vchan, rte_iova_t src, rte_iova_t d
header->cn9k.ptr = (uint64_t)comp_ptr;
STRM_INC(dpi_conf->c_desc, tail);
+ if (unlikely(dpi_conf->c_desc.tail == dpi_conf->c_desc.head)) {
+ STRM_DEC(dpi_conf->c_desc, tail);
+ return -ENOSPC;
+ }
+
header->cn9k.nfst = 1;
header->cn9k.nlst = 1;
@@ -494,6 +499,11 @@ cnxk_dmadev_copy_sg(void *dev_private, uint16_t vchan, const struct rte_dma_sge
header->cn9k.ptr = (uint64_t)comp_ptr;
STRM_INC(dpi_conf->c_desc, tail);
+ if (unlikely(dpi_conf->c_desc.tail == dpi_conf->c_desc.head)) {
+ STRM_DEC(dpi_conf->c_desc, tail);
+ return -ENOSPC;
+ }
+
/*
* For inbound case, src pointers are last pointers.
* For all other cases, src pointers are first pointers.
@@ -561,6 +571,11 @@ cn10k_dmadev_copy(void *dev_private, uint16_t vchan, rte_iova_t src, rte_iova_t
header->cn10k.ptr = (uint64_t)comp_ptr;
STRM_INC(dpi_conf->c_desc, tail);
+ if (unlikely(dpi_conf->c_desc.tail == dpi_conf->c_desc.head)) {
+ STRM_DEC(dpi_conf->c_desc, tail);
+ return -ENOSPC;
+ }
+
header->cn10k.nfst = 1;
header->cn10k.nlst = 1;
@@ -613,6 +628,11 @@ cn10k_dmadev_copy_sg(void *dev_private, uint16_t vchan, const struct rte_dma_sge
header->cn10k.ptr = (uint64_t)comp_ptr;
STRM_INC(dpi_conf->c_desc, tail);
+ if (unlikely(dpi_conf->c_desc.tail == dpi_conf->c_desc.head)) {
+ STRM_DEC(dpi_conf->c_desc, tail);
+ return -ENOSPC;
+ }
+
header->cn10k.nfst = nb_src & DPI_MAX_POINTER;
header->cn10k.nlst = nb_dst & DPI_MAX_POINTER;
fptr = &src[0];
@@ -695,8 +715,6 @@ cnxk_dmadev_completed_status(void *dev_private, uint16_t vchan, const uint16_t n
struct cnxk_dpi_compl_s *comp_ptr;
int cnt;
- RTE_SET_USED(last_idx);
-
for (cnt = 0; cnt < nb_cpls; cnt++) {
comp_ptr = c_desc->compl_ptr[c_desc->head];
status[cnt] = comp_ptr->cdata;
diff --git a/drivers/dma/cnxk/cnxk_dmadev.h b/drivers/dma/cnxk/cnxk_dmadev.h
index f375143b16..9c6c898d23 100644
--- a/drivers/dma/cnxk/cnxk_dmadev.h
+++ b/drivers/dma/cnxk/cnxk_dmadev.h
@@ -9,7 +9,7 @@
#define DPI_MAX_POINTER 15
#define STRM_INC(s, var) ((s).var = ((s).var + 1) & (s).max_cnt)
#define STRM_DEC(s, var) ((s).var = ((s).var - 1) == -1 ? (s).max_cnt : ((s).var - 1))
-#define DPI_MAX_DESC 1024
+#define DPI_MAX_DESC 2048
#define DPI_MIN_DESC 2
#define MAX_VCHANS_PER_QUEUE 4
--
2.25.1
next prev parent reply other threads:[~2023-08-23 11:16 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <20230731121225.1545318-1-amitprakashs@marvell.com>
2023-08-18 9:01 ` [PATCH v3 1/8] common/cnxk: use unique name for DPI memzone Amit Prakash Shukla
2023-08-18 9:01 ` [PATCH v3 2/8] dma/cnxk: changes for dmadev driver Amit Prakash Shukla
2023-08-21 13:27 ` [PATCH v3 1/8] common/cnxk: use unique name for DPI memzone Jerin Jacob
2023-08-21 17:49 ` [PATCH v4 " Amit Prakash Shukla
2023-08-21 17:49 ` [PATCH v4 2/8] dma/cnxk: changes for dmadev driver Amit Prakash Shukla
2023-08-23 11:15 ` [PATCH v5 01/12] common/cnxk: use unique name for DPI memzone Amit Prakash Shukla
2023-08-23 11:15 ` [PATCH v5 03/12] dma/cnxk: set dmadev to ready state Amit Prakash Shukla
2023-08-23 11:15 ` [PATCH v5 04/12] dma/cnxk: flag support for dma device Amit Prakash Shukla
2023-08-23 11:15 ` [PATCH v5 05/12] dma/cnxk: allocate completion ring buffer Amit Prakash Shukla
2023-08-23 11:15 ` [PATCH v5 06/12] dma/cnxk: chunk buffer failure return code Amit Prakash Shukla
2023-08-23 11:15 ` Amit Prakash Shukla [this message]
2023-08-23 15:30 ` [PATCH v5 01/12] common/cnxk: use unique name for DPI memzone Jerin Jacob
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=20230823111525.3975662-11-amitprakashs@marvell.com \
--to=amitprakashs@marvell.com \
--cc=dev@dpdk.org \
--cc=jerinj@marvell.com \
--cc=stable@dpdk.org \
--cc=vattunuru@marvell.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).