Hi,
>-----Original Message-----
>From: Gupta, Ajay Kumar
>Sent: Thursday, May 13, 2010 3:41 PM
>To: Kalliguddi, Hema; Gadiyar, Anand; me@xxxxxxxxxxxxxxx
>Cc: linux-usb@xxxxxxxxxxxxxxx; linux-omap@xxxxxxxxxxxxxxx
>Subject: RE: [PATCH 2/5] musb: use system DMA to fix Inventra
>DMA issue on RTL-1.4
>
>Hi,
>> >> > > Another approach to use PIO mode in opposite direction would
>> >> > > increase
>> >> the
>> >> > > cpu loading and thus using system DMA is preferred workaround.
>> >> > >
>> >> > > Signed-off-by: Anand Gadiyar <gadiyar@xxxxxx>
>> >> > > Signed-off-by: Ajay Kumar Gupta <ajay.gupta@xxxxxx>
>> >> >
>> >> > I think falling back to pio is better than this patch.
>> >At the cost of cpu, which certainly is not preferred.
>> >
>> >>> It will most likely be only one transfer.
>> >
>> >How about host mode with multiple devices connected and doing
>> >transfers?
>> >Falling back to PIO would kill the cpu.
>> >
>> >>> Another approach is to allocate dma channels on a transfer basis.
>> >
>> >Can you elaborate this?
>>
>> It might be good idea to allocate the dma channels on
>tarnsfer basis as
>> Felipe
>> Suggested. The musb driver allocates dma channels for the
>first 8 enabled
>> endpoints and higher endpoints works in PIO mode.
>> For system dma if there are more nummber of Rx endpoints
>enabled but not
>> used for data
>> Transfer you might end up having many sdma channles
>allocated biut not
>> used which will
>> Impact in the system of not utilizing the sdma channels effectively.
>
>Felipe, Is this what you meant? If so then I have a patch
>(copied below)
>to fix this and I can post this one along with others.
>
>-------------- cut here -----------------
>Currently DMA channels are allocated and they remain allocated
>even if there is no active data transfer. Added channel_release()
>whenever there is no pending request.
>
>Signed-off-by: Ajay Kumar Gupta <ajay.gupta@xxxxxx>
>---
> drivers/usb/musb/musb_gadget.c | 27 +++++++++++++++++++++------
> drivers/usb/musb/musb_host.c | 14 ++++++++++++--
> 2 files changed, 33 insertions(+), 8 deletions(-)
>
>diff --git a/drivers/usb/musb/musb_gadget.c
>b/drivers/usb/musb/musb_gadget.c
>index fd842af..477a009 100644
>--- a/drivers/usb/musb/musb_gadget.c
>+++ b/drivers/usb/musb/musb_gadget.c
>@@ -297,9 +297,13 @@ static void txstate(struct musb *musb,
>struct musb_request *req)
> csr);
>
> #ifndef CONFIG_MUSB_PIO_ONLY
>- if (is_dma_capable() && musb_ep->dma) {
>+
>+ if (is_dma_capable() && musb->dma_controller) {
> struct dma_controller *c = musb->dma_controller;
>
>+ if (!musb_ep->dma)
>+ musb_ep->dma = c->channel_alloc(c,
>musb_ep->hw_ep, 1);
If the channel_alloc returns failure then there should be a way to fallback to PIO mode.
I am posting a patch for dynamic dma channel allocation for mentor dma for gadget driver.
~Hema
>+
> use_dma = (request->dma != DMA_ADDR_INVALID);
>
> /* MUSB_TXCSR_P_ISO is still set correctly */
>@@ -433,6 +437,7 @@ void musb_g_tx(struct musb *musb, u8 epnum)
> u8 __iomem *mbase = musb->mregs;
> struct musb_ep *musb_ep =
>&musb->endpoints[epnum].ep_in;
> void __iomem *epio = musb->endpoints[epnum].regs;
>+ struct dma_controller *c = musb->dma_controller;
> struct dma_channel *dma;
>
> musb_ep_select(mbase, epnum);
>@@ -535,6 +540,10 @@ void musb_g_tx(struct musb *musb, u8 epnum)
> if (!request) {
> DBG(4, "%s idle now\n",
> musb_ep->end_point.name);
>+ if (musb_ep->dma) {
>+
>c->channel_release(musb_ep->dma);
>+ musb_ep->dma = NULL;
>+ }
> return;
> }
> }
>@@ -585,6 +594,7 @@ static void rxstate(struct musb *musb,
>struct musb_request *req)
> struct usb_request *request = &req->request;
> struct musb_ep *musb_ep =
>&musb->endpoints[epnum].ep_out;
> void __iomem *epio = musb->endpoints[epnum].regs;
>+ struct dma_controller *c = musb->dma_controller;
> unsigned fifo_count = 0;
> u16 len = musb_ep->packet_sz;
> u16 csr = musb_readw(epio, MUSB_RXCSR);
>@@ -601,8 +611,10 @@ static void rxstate(struct musb *musb,
>struct musb_request *req)
> return;
> }
>
>+ if (is_dma_capable() && musb->dma_controller && !musb_ep->dma)
>+ musb_ep->dma = c->channel_alloc(c, musb_ep->hw_ep, 0);
>+
> if ((is_cppi_enabled() || is_cppi41_enabled()) &&
>musb_ep->dma) {
>- struct dma_controller *c = musb->dma_controller;
> struct dma_channel *channel = musb_ep->dma;
>
> /* NOTE: CPPI won't actually stop advancing the DMA
>@@ -633,11 +645,9 @@ static void rxstate(struct musb *musb,
>struct musb_request *req)
> if (request->actual < request->length) {
> #ifdef CONFIG_USB_INVENTRA_DMA
> if (is_dma_capable() && musb_ep->dma) {
>- struct dma_controller *c;
> struct dma_channel *channel;
> int use_dma = 0;
>
>- c = musb->dma_controller;
> channel = musb_ep->dma;
>
> /* We use DMA Req mode 0 in rx_csr, and DMA controller
>operates in
>@@ -719,7 +729,6 @@ static void rxstate(struct musb *musb,
>struct musb_request *req)
>
> #ifdef CONFIG_USB_TUSB_OMAP_DMA
> if (tusb_dma_omap() && musb_ep->dma) {
>- struct dma_controller *c =
>musb->dma_controller;
> struct dma_channel *channel =
>musb_ep->dma;
> u32 dma_addr = request->dma +
>request->actual;
> int ret;
>@@ -764,6 +773,7 @@ void musb_g_rx(struct musb *musb, u8 epnum)
> void __iomem *mbase = musb->mregs;
> struct musb_ep *musb_ep =
>&musb->endpoints[epnum].ep_out;
> void __iomem *epio = musb->endpoints[epnum].regs;
>+ struct dma_controller *c = musb->dma_controller;
> struct dma_channel *dma;
>
> musb_ep_select(mbase, epnum);
>@@ -838,8 +848,13 @@ void musb_g_rx(struct musb *musb, u8 epnum)
> musb_g_giveback(musb_ep, request, 0);
>
> request = next_request(musb_ep);
>- if (!request)
>+ if (!request) {
>+ if (musb_ep->dma) {
>+ c->channel_release(musb_ep->dma);
>+ musb_ep->dma = NULL;
>+ }
> return;
>+ }
> }
>
> /* analyze request if the ep is hot */
>diff --git a/drivers/usb/musb/musb_host.c
>b/drivers/usb/musb/musb_host.c
>index 89c8c35..001a1d6 100644
>--- a/drivers/usb/musb/musb_host.c
>+++ b/drivers/usb/musb/musb_host.c
>@@ -460,11 +460,21 @@ static void musb_advance_schedule(struct
>musb *musb, struct urb *urb,
> */
> if (list_empty(&qh->hep->urb_list)) {
> struct list_head *head;
>+ struct dma_controller *dma = musb->dma_controller;
>
>- if (is_in)
>+ if (is_in) {
> ep->rx_reinit = 1;
>- else
>+ if (ep->rx_channel) {
>+ dma->channel_release(ep->rx_channel);
>+ ep->rx_channel = NULL;
>+ }
>+ } else {
> ep->tx_reinit = 1;
>+ if (ep->tx_channel) {
>+ dma->channel_release(ep->tx_channel);
>+ ep->tx_channel = NULL;
>+ }
>+ }
>
> /* Clobber old pointers to this qh */
> musb_ep_set_qh(ep, is_in, NULL);
>--
>1.6.2.4
>
>
>-----------------------------------------
>-Ajay
>>
>--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
