From: Zhi Li <lznuaa@gmail.com>
To: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Cc: Frank Li <Frank.Li@nxp.com>,
gustavo.pimentel@synopsys.com, hongxing.zhu@nxp.com,
Lucas Stach <l.stach@pengutronix.de>,
dl-linux-imx <linux-imx@nxp.com>,
linux-pci@vger.kernel.org, vkoul@kernel.org,
lorenzo.pieralisi@arm.com, robh@kernel.org, kw@linux.com,
Bjorn Helgaas <bhelgaas@google.com>,
Shawn Guo <shawnguo@kernel.org>
Subject: Re: [PATCH v2 3/5] dmaengine: dw-edma: add flags at struct dw_edma_chip
Date: Fri, 4 Mar 2022 11:16:11 -0600 [thread overview]
Message-ID: <CAHrpEqSFj8eTc9adhh21Ju_=P8Rb0xXVTa5QMcY-w-8xn+m5xg@mail.gmail.com> (raw)
In-Reply-To: <20220304161922.GL12451@workstation>
i
On Fri, Mar 4, 2022 at 10:19 AM Manivannan Sadhasivam
<manivannan.sadhasivam@linaro.org> wrote:
>
> On Thu, Mar 03, 2022 at 12:46:33PM -0600, Frank Li wrote:
> > Allow user don't enable remote MSI irq.
> > PCI ep probe dma locally, don't want to raise irq
> > to remote PCI host.
> >
> > Add option allow force 32bit register access even at
> > 64bit system. i.MX8 hardware only allowed 32bit register
> > access.
> >
> > Add option allow EP side probe dma. remote side dma is
> > continue physical memory, local memory is scatter list.
> >
> > Signed-off-by: Frank Li <Frank.Li@nxp.com>
> > ---
> > Change from v1 to v2
> > - none
> >
> > drivers/dma/dw-edma/dw-edma-core.c | 7 ++++++-
> > drivers/dma/dw-edma/dw-edma-v0-core.c | 14 ++++++++++----
> > include/linux/dma/edma.h | 7 +++++++
> > 3 files changed, 23 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/dma/dw-edma/dw-edma-core.c b/drivers/dma/dw-edma/dw-edma-core.c
> > index 0cb66434f9e14..a43bb26c8bf96 100644
> > --- a/drivers/dma/dw-edma/dw-edma-core.c
> > +++ b/drivers/dma/dw-edma/dw-edma-core.c
> > @@ -336,6 +336,7 @@ dw_edma_device_transfer(struct dw_edma_transfer *xfer)
> > struct dw_edma_desc *desc;
> > u32 cnt = 0;
> > int i;
> > + bool b;
> >
> > if (!chan->configured)
> > return NULL;
> > @@ -424,7 +425,11 @@ dw_edma_device_transfer(struct dw_edma_transfer *xfer)
> > chunk->ll_region.sz += burst->sz;
> > desc->alloc_sz += burst->sz;
> >
> > - if (chan->dir == EDMA_DIR_WRITE) {
> > + b = (chan->dir == EDMA_DIR_WRITE);
> > + if (chan->chip->flags & DW_EDMA_CHIP_LOCAL_EP)
> > + b = !b;
> > +
> > + if (b) {
>
> I've added a patch that uses the xfer direction and channel direction to
> find out whether it is a read operation or not. Please take a look:
>
> https://git.linaro.org/landing-teams/working/qualcomm/kernel.git/commit/?h=tracking-qcomlt-sdx55-drivers&id=d6a3f432204614ad8531321949a8a9e2c3e94c3b
>
I think your patch is better. Can I include your patch into this sequence ?
> The patch could also make use of the "DW_EDMA_CHIP_LOCAL" flag if you
> prefer.
>
> > burst->sar = src_addr;
> > if (xfer->type == EDMA_XFER_CYCLIC) {
> > burst->dar = xfer->xfer.cyclic.paddr;
> > diff --git a/drivers/dma/dw-edma/dw-edma-v0-core.c b/drivers/dma/dw-edma/dw-edma-v0-core.c
> > index 6e2f83e31a03a..d5c2415e2c616 100644
> > --- a/drivers/dma/dw-edma/dw-edma-v0-core.c
> > +++ b/drivers/dma/dw-edma/dw-edma-v0-core.c
> > @@ -307,13 +307,18 @@ u32 dw_edma_v0_core_status_abort_int(struct dw_edma_chip *chip, enum dw_edma_dir
> > static void dw_edma_v0_core_write_chunk(struct dw_edma_chunk *chunk)
> > {
> > struct dw_edma_burst *child;
> > + struct dw_edma_chan *chan = chunk->chan;
> > struct dw_edma_v0_lli __iomem *lli;
> > struct dw_edma_v0_llp __iomem *llp;
> > u32 control = 0, i = 0;
> > + u32 rie = 0;
> > int j;
> >
> > lli = chunk->ll_region.vaddr;
> >
> > + if (!(chan->chip->flags & DW_EDMA_CHIP_NO_MSI))
> > + rie = DW_EDMA_V0_RIE;
> > +
> > if (chunk->cb)
> > control = DW_EDMA_V0_CB;
> >
> > @@ -321,7 +326,7 @@ static void dw_edma_v0_core_write_chunk(struct dw_edma_chunk *chunk)
> > list_for_each_entry(child, &chunk->burst->list, list) {
> > j--;
> > if (!j)
> > - control |= (DW_EDMA_V0_LIE | DW_EDMA_V0_RIE);
> > + control |= (DW_EDMA_V0_LIE | rie);
>
> I think the MSI makes sense only for eDMA access from remote. So how
> about reusing the same flag you used above?
Understand, DW_EDMA_CHIP_NO_MSI is more direct. User can know what
exactly control.
DW_EDMA_CHIP_LOCAL is quite general. User don't know what to do from naming.
I am okay for both naming.
If I pick up your patch for dma dir, Only below two flags need,
#define DW_EDMA_CHIP_NO_MSI BIT(0)
#define DW_EDMA_CHIP_REG32BIT BIT(1)
vs
#define DW_EDMA_CHIP_LOCAL BIT(0)
#define DW_EDMA_CHIP_REG32BIT BIT(1)
which one is better?
>
> control |= DW_EDMA_V0_LIE;
>
> /* Raise MSI only if the eDMA is accessed by remote */
> if (!(chan->chip->flags & DW_EDMA_CHIP_LOCAL))
> control |= DW_EDMA_V0_RIE;
>
> >
> > /* Channel control */
> > SET_LL_32(&lli[i].control, control);
> > @@ -420,15 +425,16 @@ void dw_edma_v0_core_start(struct dw_edma_chunk *chunk, bool first)
> > SET_CH_32(chip, chan->dir, chan->id, ch_control1,
> > (DW_EDMA_V0_CCS | DW_EDMA_V0_LLE));
> > /* Linked list */
> > - #ifdef CONFIG_64BIT
> > + if (!(chan->chip->flags & DW_EDMA_CHIP_REG32BIT) &&
> > + IS_ENABLED(CONFIG_64BIT)) {
> > SET_CH_64(chip, chan->dir, chan->id, llp.reg,
> > chunk->ll_region.paddr);
>
> Why you are doing 32bit access here only and not in other places?
>
Only here access DBI register, other place is access dma link list,
which is memory.
> Thanks,
> Mani
>
> > - #else /* CONFIG_64BIT */
> > + } else {
> > SET_CH_32(chip, chan->dir, chan->id, llp.lsb,
> > lower_32_bits(chunk->ll_region.paddr));
> > SET_CH_32(chip, chan->dir, chan->id, llp.msb,
> > upper_32_bits(chunk->ll_region.paddr));
> > - #endif /* CONFIG_64BIT */
> > + }
> > }
> > /* Doorbell */
> > SET_RW_32(chip, chan->dir, doorbell,
> > diff --git a/include/linux/dma/edma.h b/include/linux/dma/edma.h
> > index fcfbc0f47f83d..e74ee15d9832a 100644
> > --- a/include/linux/dma/edma.h
> > +++ b/include/linux/dma/edma.h
> > @@ -33,6 +33,10 @@ enum dw_edma_map_format {
> > EDMA_MF_HDMA_COMPAT = 0x5
> > };
> >
> > +#define DW_EDMA_CHIP_NO_MSI BIT(0)
> > +#define DW_EDMA_CHIP_REG32BIT BIT(1)
> > +#define DW_EDMA_CHIP_LOCAL_EP BIT(2)
>
> As per my understanding the
what's you mean?
>
> > +
> > /**
> > * struct dw_edma_chip - representation of DesignWare eDMA controller hardware
> > * @dev: struct device of the eDMA controller
> > @@ -40,6 +44,8 @@ enum dw_edma_map_format {
> > * @nr_irqs: total dma irq number
> > * reg64bit if support 64bit write to register
> > * @ops DMA channel to IRQ number mapping
> > + * @flags - DW_EDMA_CHIP_NO_MSI can't generate remote MSI irq
> > + * - DW_EDMA_CHIP_REG32BIT only support 32bit register write
> > * @wr_ch_cnt DMA write channel number
> > * @rd_ch_cnt DMA read channel number
> > * @rg_region DMA register region
> > @@ -53,6 +59,7 @@ struct dw_edma_chip {
> > int id;
> > int nr_irqs;
> > const struct dw_edma_core_ops *ops;
> > + u32 flags;
> >
> > void __iomem *reg_base;
> >
> > --
> > 2.24.0.rc1
> >
next prev parent reply other threads:[~2022-03-04 17:16 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-03 18:46 [PATCH v2 1/5] dmaengine: dw-edma: fix dw_edma_probe() can't be call globally Frank Li
2022-03-03 18:46 ` [PATCH v2 2/5] dmaengine: dw-edma-pcie: don't touch internal struct dw_edma Frank Li
2022-03-03 18:46 ` [PATCH v2 3/5] dmaengine: dw-edma: add flags at struct dw_edma_chip Frank Li
2022-03-04 16:19 ` Manivannan Sadhasivam
2022-03-04 17:16 ` Zhi Li [this message]
2022-03-07 13:59 ` Manivannan Sadhasivam
2022-03-04 21:31 ` Bjorn Helgaas
2022-03-03 18:46 ` [PATCH v2 4/5] PCI: imx6: add PCIe embedded DMA support Frank Li
2022-03-04 21:33 ` Bjorn Helgaas
2022-03-05 1:31 ` Zhi Li
2022-03-09 12:01 ` Manivannan Sadhasivam
2022-03-09 18:31 ` Zhi Li
2022-03-09 18:42 ` Manivannan Sadhasivam
2022-03-09 19:14 ` Zhi Li
2022-03-09 20:48 ` Zhi Li
2022-03-10 4:45 ` Manivannan Sadhasivam
2022-03-03 18:46 ` [PATCH v2 5/5] PCI: endpoint: functions/pci-epf-test: Support PCI controller DMA Frank Li
2022-03-10 22:07 ` [PATCH v2 1/5] dmaengine: dw-edma: fix dw_edma_probe() can't be call globally Zhi Li
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='CAHrpEqSFj8eTc9adhh21Ju_=P8Rb0xXVTa5QMcY-w-8xn+m5xg@mail.gmail.com' \
--to=lznuaa@gmail.com \
--cc=Frank.Li@nxp.com \
--cc=bhelgaas@google.com \
--cc=gustavo.pimentel@synopsys.com \
--cc=hongxing.zhu@nxp.com \
--cc=kw@linux.com \
--cc=l.stach@pengutronix.de \
--cc=linux-imx@nxp.com \
--cc=linux-pci@vger.kernel.org \
--cc=lorenzo.pieralisi@arm.com \
--cc=manivannan.sadhasivam@linaro.org \
--cc=robh@kernel.org \
--cc=shawnguo@kernel.org \
--cc=vkoul@kernel.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).