From: Aniket Randive <aniket.randive@oss.qualcomm.com>
To: mukesh.savaliya@oss.qualcomm.com,
viken.dadhaniya@oss.qualcomm.com, andi.shyti@kernel.org,
sumit.semwal@linaro.org, christian.koenig@amd.com
Cc: linux-i2c@vger.kernel.org, linux-arm-msm@vger.kernel.org,
linux-kernel@vger.kernel.org, linux-media@vger.kernel.org,
dri-devel@lists.freedesktop.org, linaro-mm-sig@lists.linaro.org,
naresh.maramaina@oss.qualcomm.com,
aniket.randive@oss.qualcomm.com
Subject: [PATCH V4] i2c: qcom-geni: Avoid extra TX DMA TRE for single read message in GPI mode
Date: Fri, 10 Apr 2026 15:49:49 +0530 [thread overview]
Message-ID: <20260410101949.2315058-1-aniket.randive@oss.qualcomm.com> (raw)
In GPI mode, the I2C GENI driver programs an extra TX DMA transfer
descriptor (TRE) on the TX channel when handling a single read message.
This results in an unintended write phase being issued on the I2C bus,
even though a read transaction does not require any TX data.
For a single-byte read, the correct hardware sequence consists of the
CONFIG and GO commands followed by a single RX DMA TRE. Programming an
additional TX DMA TRE is redundant, causes unnecessary DMA buffer
mapping on the TX channel, and may lead to incorrect bus behavior.
Update the transfer logic to avoid programming a TX DMA TRE for single
read messages in GPI mode.
Co-developed-by: Maramaina Naresh <naresh.maramaina@oss.qualcomm.com>
Signed-off-by: Maramaina Naresh <naresh.maramaina@oss.qualcomm.com>
Signed-off-by: Aniket Randive <aniket.randive@oss.qualcomm.com>
---
Changes in v4:
- Added some more description in comment and changed the label name.
Changes in v3:
- Added comment in the driver for better readability and changed the
position of 'skip_dma' label to allow dma engine configuration.
Changes in v2:
- Updated the commit message.
drivers/i2c/busses/i2c-qcom-geni.c | 24 +++++++++++++++++++-----
1 file changed, 19 insertions(+), 5 deletions(-)
diff --git a/drivers/i2c/busses/i2c-qcom-geni.c b/drivers/i2c/busses/i2c-qcom-geni.c
index a4acb78fafb6..a482a4c60744 100644
--- a/drivers/i2c/busses/i2c-qcom-geni.c
+++ b/drivers/i2c/busses/i2c-qcom-geni.c
@@ -625,8 +625,8 @@ static int geni_i2c_gpi(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[],
{
struct gpi_i2c_config *peripheral;
unsigned int flags;
- void *dma_buf;
- dma_addr_t addr;
+ void *dma_buf = NULL;
+ dma_addr_t addr = 0;
enum dma_data_direction map_dirn;
enum dma_transfer_direction dma_dirn;
struct dma_async_tx_descriptor *desc;
@@ -639,6 +639,16 @@ static int geni_i2c_gpi(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[],
gi2c_gpi_xfer = &gi2c->i2c_multi_desc_config;
msg_idx = gi2c_gpi_xfer->msg_idx_cnt;
+ /*
+ * Skip TX DMA mapping for a read message (I2C_M_RD) to avoid
+ * programming an extra TX DMA TRE that would cause an unintended
+ * write cycle on the I2C bus before the actual read operation.
+ */
+ if (op == I2C_WRITE && msgs[msg_idx].flags & I2C_M_RD) {
+ peripheral->multi_msg = true;
+ goto skip_tx_dma_map;
+ }
+
dma_buf = i2c_get_dma_safe_msg_buf(&msgs[msg_idx], 1);
if (!dma_buf) {
ret = -ENOMEM;
@@ -658,6 +668,7 @@ static int geni_i2c_gpi(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[],
goto out;
}
+skip_tx_dma_map:
if (gi2c->is_tx_multi_desc_xfer) {
flags = DMA_CTRL_ACK;
@@ -740,9 +751,12 @@ static int geni_i2c_gpi(struct geni_i2c_dev *gi2c, struct i2c_msg msgs[],
return 0;
err_config:
- dma_unmap_single(gi2c->se.dev->parent, addr,
- msgs[msg_idx].len, map_dirn);
- i2c_put_dma_safe_msg_buf(dma_buf, &msgs[msg_idx], false);
+ /* Avoid DMA unmap as the write operation skipped DMA mapping */
+ if (dma_buf) {
+ dma_unmap_single(gi2c->se.dev->parent, addr,
+ msgs[msg_idx].len, map_dirn);
+ i2c_put_dma_safe_msg_buf(dma_buf, &msgs[msg_idx], false);
+ }
out:
gi2c->err = ret;
--
2.34.1
next reply other threads:[~2026-04-10 10:20 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-10 10:19 Aniket Randive [this message]
2026-04-10 11:22 ` [PATCH V4] i2c: qcom-geni: Avoid extra TX DMA TRE for single read message in GPI mode Mukesh Kumar Savaliya
2026-04-10 21:01 ` Andi Shyti
2026-04-11 23:50 ` Claude review: " Claude Code Review Bot
2026-04-11 23:50 ` Claude Code Review Bot
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=20260410101949.2315058-1-aniket.randive@oss.qualcomm.com \
--to=aniket.randive@oss.qualcomm.com \
--cc=andi.shyti@kernel.org \
--cc=christian.koenig@amd.com \
--cc=dri-devel@lists.freedesktop.org \
--cc=linaro-mm-sig@lists.linaro.org \
--cc=linux-arm-msm@vger.kernel.org \
--cc=linux-i2c@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-media@vger.kernel.org \
--cc=mukesh.savaliya@oss.qualcomm.com \
--cc=naresh.maramaina@oss.qualcomm.com \
--cc=sumit.semwal@linaro.org \
--cc=viken.dadhaniya@oss.qualcomm.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