Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S933129Ab3CLQVq (ORCPT ); Tue, 12 Mar 2013 12:21:46 -0400 Received: from mail-ia0-f170.google.com ([209.85.210.170]:51336 "EHLO mail-ia0-f170.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932923Ab3CLQVn (ORCPT ); Tue, 12 Mar 2013 12:21:43 -0400 Date: Tue, 12 Mar 2013 12:22:33 -0400 From: Matt Porter To: "Nori, Sekhar" Cc: Tony Lindgren , Grant Likely , Rob Herring , Vinod Koul , Mark Brown , "Cousson, Benoit" , Russell King , Rob Landley , Andrew Morton , Devicetree Discuss , Linux OMAP List , Linux ARM Kernel List , Linux DaVinci Kernel List , Linux Kernel Mailing List , Linux Documentation List , Linux MMC List , Linux SPI Devel List , Arnd Bergmann Subject: Re: [PATCH v9 3/9] ARM: edma: add AM33XX support to the private EDMA API Message-ID: <20130312162233.GY6209@beef> References: <1362586540-10393-1-git-send-email-mporter@ti.com> <1362586540-10393-4-git-send-email-mporter@ti.com> <76caa6ddf7a143b381f30aa8fdfef3cc@DLEE71.ent.ti.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <76caa6ddf7a143b381f30aa8fdfef3cc@DLEE71.ent.ti.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 10270 Lines: 314 On Tue, Mar 12, 2013 at 06:45:46AM +0000, Sekhar Nori wrote: > > > On 3/6/2013 9:45 PM, Matt Porter wrote: > > Adds support for parsing the TI EDMA DT data into the > > required EDMA private API platform data. Enables runtime > > PM support to initialize the EDMA hwmod. Adds AM33XX EDMA > > crossbar event mux support. Enables build on OMAP. > > > > Signed-off-by: Matt Porter > > Acked-by: Sekhar Nori > > --- > > arch/arm/common/edma.c | 300 ++++++++++++++++++++++++++++++++++-- > > arch/arm/mach-omap2/Kconfig | 1 + > > include/linux/platform_data/edma.h | 1 + > > 3 files changed, 292 insertions(+), 10 deletions(-) > > > > diff --git a/arch/arm/common/edma.c b/arch/arm/common/edma.c > > index a1db6cd..e68ac38 100644 > > --- a/arch/arm/common/edma.c > > +++ b/arch/arm/common/edma.c > > @@ -24,6 +24,13 @@ > > #include > > #include > > #include > > +#include > > +#include > > +#include > > +#include > > +#include > > +#include > > +#include > > > > #include > > > > @@ -1369,31 +1376,278 @@ void edma_clear_event(unsigned channel) > > EXPORT_SYMBOL(edma_clear_event); > > > > /*-----------------------------------------------------------------------*/ > > +static int edma_of_read_u32_to_s8_array(const struct device_node *np, > > + const char *propname, s8 *out_values, > > + size_t sz) > > +{ > > + int ret; > > + > > + ret = of_property_read_u8_array(np, propname, out_values, sz); > > + if (ret) > > + return ret; > > + > > + /* Terminate it */ > > + *out_values++ = -1; > > + *out_values++ = -1; > > + > > + return 0; > > +} > > + > > +static int edma_of_read_u32_to_s16_array(const struct device_node *np, > > + const char *propname, s16 *out_values, > > + size_t sz) > > +{ > > + int ret; > > + > > + ret = of_property_read_u16_array(np, propname, out_values, sz); > > + if (ret) > > + return ret; > > + > > + /* Terminate it */ > > + *out_values++ = -1; > > + *out_values++ = -1; > > + > > + return 0; > > +} > > + > > +static int edma_xbar_event_map(struct device *dev, > > + struct device_node *node, > > + struct edma_soc_info *pdata, int len) > > +{ > > It will be nice to separate the xbar feature from DT'fication of the > existing driver. Right now because of the mix the patch has become > pretty big and its becoming tough to review in isolation. Sure, I'll do that on v10. > > + int ret = 0; > > + int i; > > + struct resource res; > > + void *xbar; > > + const s16 (*xbar_chans)[2]; > > + u32 shift, offset, mux; > > + > > + xbar_chans = devm_kzalloc(dev, > > + len/sizeof(s16) + 2*sizeof(s16), > > + GFP_KERNEL); > > + if (!xbar_chans) > > + return -ENOMEM; > > + > > + ret = of_address_to_resource(node, 1, &res); > > + if (ret) > > + return -EIO; > > + > > + xbar = devm_ioremap(dev, res.start, resource_size(&res)); > > + if (!xbar) > > + return -ENOMEM; > > + > > + ret = edma_of_read_u32_to_s16_array(node, > > + "ti,edma-xbar-event-map", > > + (s16 *)xbar_chans, > > + len/sizeof(u32)); > > + if (ret) > > + return -EIO; > > + > > + for (i = 0; xbar_chans[i][0] != -1; i++) { > > + shift = (xbar_chans[i][1] % 4) * 8; > > + offset = xbar_chans[i][1] >> 2; > > + offset <<= 2; > > + mux = readl((void *)((u32)xbar + offset)); > > + mux &= ~(0xff << shift); > > + mux |= xbar_chans[i][0] << shift; > > + writel(mux, (void *)((u32)xbar + offset)); > > + } > > + > > + pdata->xbar_chans = xbar_chans; > > + > > + return 0; > > +} > > + > > +static int edma_of_parse_dt(struct device *dev, > > + struct device_node *node, > > + struct edma_soc_info *pdata) > > +{ > > + int ret = 0; > > + u32 value; > > + struct property *prop; > > + size_t sz; > > + struct edma_rsv_info *rsv_info; > > + const s16 (*rsv_chans)[2], (*rsv_slots)[2]; > > + const s8 (*queue_tc_map)[2], (*queue_priority_map)[2]; > > + > > + memset(pdata, 0, sizeof(struct edma_soc_info)); > > + > > + ret = of_property_read_u32(node, "dma-channels", &value); > > + if (ret < 0) > > + return ret; > > + pdata->n_channel = value; > > + > > + ret = of_property_read_u32(node, "ti,edma-regions", &value); > > + if (ret < 0) > > + return ret; > > + pdata->n_region = value; > > + > > + ret = of_property_read_u32(node, "ti,edma-slots", &value); > > + if (ret < 0) > > + return ret; > > + pdata->n_slot = value; > > + > > + pdata->n_cc = 1; > > + pdata->n_tc = 3; > > Will this mean the DT portion of this driver cannot be used on SoCs > where there are two CCs like DA850? If you are hard-coding this, will it > make sense to set to to EDMA_MAX_CC instead? Okay I see a comment down > below saying DT will support only one CC. Not sure why, but this is > probably related to that. Yeah, this series is aging quickly with all the work done on Davinci DT support recently. The actual parsing implementation was intended to be short-term for am33xx only when written. I'll update and test on DA850 with DT support. > > + > > + rsv_info = > > + devm_kzalloc(dev, sizeof(struct edma_rsv_info), GFP_KERNEL); > > + if (!rsv_info) > > + return -ENOMEM; > > + pdata->rsv = rsv_info; > > + > > + /* Build the reserved channel/slots arrays */ > > + prop = of_find_property(node, "ti,edma-reserved-channels", &sz); > > + if (prop) { > > + rsv_chans = devm_kzalloc(dev, > > + sz/sizeof(s16) + 2*sizeof(s16), > > + GFP_KERNEL); > > + if (!rsv_chans) > > + return -ENOMEM; > > + pdata->rsv->rsv_chans = rsv_chans; > > + > > + ret = edma_of_read_u32_to_s16_array(node, > > + "ti,edma-reserved-channels", > > + (s16 *)rsv_chans, > > + sz/sizeof(u32)); > > Is this binding accepted? I cant find it in v3.9-rc1. IMHO, this > configuration should not be through DT. This is configuration material > for a given application (which channels should Linux running on ARM use > vs which channels should DSP use?) but not hardware description. > Probably configfs is useful here but I myself need to go through the > details. No, there's been no review by the DT maintainers as of yet. I agree it could either fall into the grey area where sometimes these things are permitted or we simply should find a different way. configfs is, indeed, and obvious choice. Most of these parameters are tuning characteristics that in any application I can think of could probably wait until user space is available...and so configfs is ok here. The only downside might be somebody wanting the rootfs device to be able to allocate a channel with something other than the default queue...that can be fixed with some creative initramfs. > On AM335x the usage of this is further limited use since applications > which need DMA from PRU or M3 are limited (at least I don't know of any). I don't know of any either, it's simply conjecture since EDMA is available to those cores. > I know its frustrating to get these comments piece by piece and after v9 > has already been posted. Sorry about that. I think it will be easier to Not a problem. I'm happy to have other eyes on the DT portion. > drop this and some other may-not-really-be-a-hardware-description > bindings like "ti,edma-reserved-slots" for now and just get the basic > support in. The other ones can then be discussed/handled separately. That works for me. It'll be less to be reviewed when Rob/Grant are able to look at it. > > + if (ret < 0) > > + return ret; > > + } > > > > -static int __init edma_probe(struct platform_device *pdev) > > + prop = of_find_property(node, "ti,edma-reserved-slots", &sz); > > + if (prop) { > > + rsv_slots = devm_kzalloc(dev, > > + sz/sizeof(s16) + 2*sizeof(s16), > > + GFP_KERNEL); > > + if (!rsv_slots) > > + return -ENOMEM; > > + pdata->rsv->rsv_slots = rsv_slots; > > + > > + ret = edma_of_read_u32_to_s16_array(node, > > + "ti,edma-reserved-slots", > > + (s16 *)rsv_slots, > > + sz/sizeof(u32)); > > + if (ret < 0) > > + return ret; > > + } > > + > > + prop = of_find_property(node, "ti,edma-queue-tc-map", &sz); > > Again, this is application-driven configuration from EDMA IP > point-of-view. For some of these may be you can leave some sane defaults > which will work on most systems (AM335x included) and then we can look > at how this can be configured when an actual need arises (and at that > time we can look at the best way of doing it). Same thing for a number > of other properties below. Yep, will go to all defaults on these. > > + if (!prop) > > + return -EINVAL; > > + > > + queue_tc_map = devm_kzalloc(dev, > > + sz/sizeof(s8) + 2*sizeof(s8), > > + GFP_KERNEL); > > + if (!queue_tc_map) > > + return -ENOMEM; > > + pdata->queue_tc_mapping = queue_tc_map; > > + > > + ret = edma_of_read_u32_to_s8_array(node, > > + "ti,edma-queue-tc-map", > > + (s8 *)queue_tc_map, > > + sz/sizeof(u32)); > > + if (ret < 0) > > + return ret; > > + > > + prop = of_find_property(node, "ti,edma-queue-priority-map", &sz); > > + if (!prop) > > + return -EINVAL; > > + > > + queue_priority_map = devm_kzalloc(dev, > > + sz/sizeof(s8) + 2*sizeof(s8), > > + GFP_KERNEL); > > + if (!queue_priority_map) > > + return -ENOMEM; > > + pdata->queue_priority_mapping = queue_priority_map; > > + > > + ret = edma_of_read_u32_to_s8_array(node, > > + "ti,edma-queue-tc-map", > > + (s8 *)queue_priority_map, > > + sz/sizeof(u32)); > > + if (ret < 0) > > + return ret; > > + > > + ret = of_property_read_u32(node, "ti,edma-default-queue", &value); > > + if (ret < 0) > > + return ret; > > + pdata->default_queue = value; > > + > > + prop = of_find_property(node, "ti,edma-xbar-event-map", &sz); > > + if (prop) > > + ret = edma_xbar_event_map(dev, node, pdata, sz); > > + > > + return ret; > > +} > > Thanks, > Sekhar -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/