video: add PXA3xx accelerator ID
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
PATCH 1/1 video: add PXA3xx accelerator ID
Add ID 99 for PXA3xx frame buffers and use it in pxafb when running on a pxa with acceleration enabled.
--- drivers/video/pxafb.c | 5 +++++ include/linux/fb.h | 1 + 2 files changed, 6 insertions(+), 0 deletions(-)
drivers/video/pxafb.c +5 -0
@@ -2083,6 +2083,11 @@ static int __devinit pxafb_probe(struct platform_device *dev) goto failed; } +#ifdef CONFIG_PXA3XX_GCU +if (cpu_is_pxa3xx ()) + fbi->fb.fix.accel =FB_ACCEL_PXA3XX ; +#endif + fbi->backlight_power = inf->pxafb_backlight_power; fbi->lcd_power = inf->pxafb_lcd_power;
include/linux/fb.h +1 -0
@@ -133,6 +133,7 @@ struct dentry;#define FB_ACCEL_NEOMAGIC_NM2230 96/* NeoMagic NM2230 */ #define FB_ACCEL_NEOMAGIC_NM2360 97/* NeoMagic NM2360 */ #define FB_ACCEL_NEOMAGIC_NM2380 98/* NeoMagic NM2380 */ +#define FB_ACCEL_PXA3XX 99/* PXA3xx */ #define FB_ACCEL_SAVAGE4 0x80/* S3 Savage4 */ #define FB_ACCEL_SAVAGE3D 0x81/* S3 Savage3D */
Signature · 7 lines
1.6.3.1 ------------------------------------------------------------------- List admin: http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm-kernel FAQ: http://www.arm.linux.org.uk/mailinglists/faq.php Etiquette: http://www.arm.linux.org.uk/mailinglists/etiquette.php
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
PATCH 4/4 video: add driver for PXA3xx 2D graphics accelerator
Quoted 12 lines · Daniel Mack
Quoted 11 lines · Eric Miao
Quoted 6 lines · Daniel Mack
Quoted 2 lines
Cool - I'll find a time slot to test this, and maybe someone else wants to have a try either. Thanks.
Will that accelerator ID patch (the one I sent today) go thru your channel or are the fbdev people supposed to pick it?
Better to go through fbdev people - or at least I need an Ack before it goes through my tree.
Quoted 6 lines
Quoted 1 line
Krzysztof Cc'ed.
Krzysztof,
resending this because I don't known whether you got the first copy of that patch.
I'm sending this patch for the the forth time now, this time with the fbdev-devel list copied. Maybe that helps.
This is fairly simple, but it's blocking the pxa3xx-gcu driver to be merged.
Any feedback is appreciated.
Thanks,
Daniel
Quoted 1 line
From
8b200756251e3dcdc91d506e9dbc9846ccaa91d3Mon Sep 17 00:00:00 2001
From: Daniel Mack <[email protected]>
Date: Mon, 20 Jul 2009 11:18:43 +0200
Subject: [PATCH] video: add PXA3xx accelerator ID
Add ID 99 for PXA3xx frame buffers and use it in pxafb when running on a pxa with acceleration enabled.
--- drivers/video/pxafb.c | 5 +++++ include/linux/fb.h | 1 + 2 files changed, 6 insertions(+), 0 deletions(-)
drivers/video/pxafb.c +5 -0
@@ -2083,6 +2083,11 @@ static int __devinit pxafb_probe(struct platform_device *dev) goto failed; } +#ifdef CONFIG_PXA3XX_GCU +if (cpu_is_pxa3xx ()) + fbi->fb.fix.accel =FB_ACCEL_PXA3XX ; +#endif + fbi->backlight_power = inf->pxafb_backlight_power; fbi->lcd_power = inf->pxafb_lcd_power;
include/linux/fb.h +1 -0
@@ -133,6 +133,7 @@ struct dentry;#define FB_ACCEL_NEOMAGIC_NM2230 96/* NeoMagic NM2230 */ #define FB_ACCEL_NEOMAGIC_NM2360 97/* NeoMagic NM2360 */ #define FB_ACCEL_NEOMAGIC_NM2380 98/* NeoMagic NM2380 */ +#define FB_ACCEL_PXA3XX 99/* PXA3xx */ #define FB_ACCEL_SAVAGE4 0x80/* S3 Savage4 */ #define FB_ACCEL_SAVAGE3D 0x81/* S3 Savage3D */
Signature · 1 line
1.6.3.1
PATCH 4/4 video: add driver for PXA3xx 2D graphics accelerator
In reply to a message not in the archive(now used the new AKML address)
Hi,
Was anyone able to give that driver a try?
Haojian, you wanted to have a look IIRC?
I would appreciate seeing this in .35 if possible.
Daniel
Quoted 728 lines · Daniel Mack
This adds a driver for the the 2D graphics accelerator found on PXA3xx processors. Only resource mapping, interrupt handling and a simple ioctl handler is done by the kernel part, the rest of the logic is implemented in DirectFB userspace.
Graphic applications benefit for line drawing, blend, and rectangle and triangle filling operations.
Measurements on a PXA303 using the
df_dokbenchmarking tool follow, where the value in square brackets show the CPU usage during that test.Without accelerator:
Fill Rectangle 3.007 secs ( 175.922 MPixel/sec) [100.3%] Fill Rectangles [10] 3.178 secs ( 182.696 MPixel/sec) [100.0%] Fill Triangles 3.023 secs ( 99.232 MPixel/sec) [100.3%] Blit 3.082 secs ( 39.770 MPixel/sec) [100.0%] Blit 180 3.115 secs ( 28.994 MPixel/sec) [100.3%] Blit with format conversion 3.078 secs ( 18.863 MPixel/sec) [100.0%]With accelerator:
Fill Rectangle 4.377 secs (* 88.433 MPixel/sec) [ 5.9%] Fill Rectangles [10] 5.176 secs (* 87.245 MPixel/sec) [ 0.5%] Fill Triangles 3.025 secs ( 87.437 MPixel/sec) [ 46.6%] Blit 12.177 secs (* 22.250 MPixel/sec) [ 1.3%] Blit 180 8.421 secs (* 32.175 MPixel/sec) [ 4.9%] Blit with format conversion 8.216 secs (* 40.045 MPixel/sec) [ 2.5%]Signed-off-by Daniel Mack <[email protected]>Cc Denis Oliver Kropp <[email protected]>--- drivers/video/Kconfig | 10 + drivers/video/Makefile | 1 + drivers/video/pxa3xx-gcu.c | 591 ++++++++++++++++++++++++++++++++++++++++++++ drivers/video/pxa3xx-gcu.h | 49 ++++ 4 files changed, 651 insertions(+), 0 deletions(-) create mode 100644 drivers/video/pxa3xx-gcu.c create mode 100644 drivers/video/pxa3xx-gcu.hdrivers/video/Kconfig +10
-0@@ -1817,6 +1817,16 @@ config FB_PXA_PARAMETERS <file:Documentation/fb/pxafb.txt> describes the available parameters. +config PXA3XX_GCU + tristate "PXA3xx 2D graphics accelerator driver" + depends on FB_PXA + help + Kernelspace driver for the 2D graphics controller unit (GCU) + found on PXA3xx processors. There is a counterpart driver in the + DirectFB suite, see http://www.directfb.org/ + + If you compile this as a module, it will be called pxa3xx_gcu. + config FB_MBX tristate "2700G LCD framebuffer support" depends on FB && ARCH_PXAdrivers/video/Makefile +1
-0@@ -98,6 +98,7 @@ obj-$(CONFIG_FB_CIRRUS) += cirrusfb.o obj-$(CONFIG_FB_ASILIANT) += asiliantfb.o obj-$(CONFIG_FB_PXA) += pxafb.o obj-$(CONFIG_FB_PXA168) += pxa168fb.o +obj-$(CONFIG_PXA3XX_GCU) += pxa3xx-gcu.o obj-$(CONFIG_FB_W100) += w100fb.o obj-$(CONFIG_FB_TMIO) += tmiofb.o obj-$(CONFIG_FB_AU1100) += au1100fb.odrivers/video/pxa3xx-gcu.c new file +591
-0@@ -0,0 +1,591 @@ +/* +* pxa3xx-gc.c - Linux kernel module for PXA3xx graphics controllers +* +* This driver needs a DirectFB counterpart in user space, communication +* is handled via mmap()ed memory areas and an ioctl. +* +* Copyright (c) 2009 Daniel Mack <[email protected]> +* Copyright (c) 2009 Janine Kropp <[email protected]> +* Copyright (c) 2009 Denis Oliver Kropp <[email protected]> +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. +*/ + +/* +* WARNING: This controller is attached to System Bus 2 of the PXA which +* needs its arbiter to be enabled explictly (CKENB & 1<<9). +* There is currently no way to do this from Linux, so you need to teach +* your bootloader for now. +*/ + +#include <linux/module.h> +#include <linux/version.h> + +#include <linux/platform_device.h> +#include <linux/dma-mapping.h> +#include <linux/miscdevice.h> +#include <linux/interrupt.h> +#include <linux/spinlock.h> +#include <linux/uaccess.h> +#include <linux/ioctl.h> +#include <linux/delay.h> +#include <linux/sched.h> +#include <linux/clk.h> +#include <linux/fs.h> +#include <linux/io.h> + +#include "pxa3xx-gcu.h" + +#define DRV_NAME "pxa3xx-gcu" +#define MISCDEV_MINOR 197 + +#define REG_GCCR 0x00 +#define GCCR_SYNC_CLR (1 << 9) +#define GCCR_BP_RST (1 << 8) +#define GCCR_ABORT (1 << 6) +#define GCCR_STOP (1 << 4) + +#define REG_GCISCR 0x04 +#define REG_GCIECR 0x08 +#define REG_GCRBBR 0x20 +#define REG_GCRBLR 0x24 +#define REG_GCRBHR 0x28 +#define REG_GCRBTR 0x2C +#define REG_GCRBEXHR 0x30 + +#define IE_EOB (1 << 0) +#define IE_EEOB (1 << 5) +#define IE_ALL 0xff + +#define SHARED_SIZE PAGE_ALIGN(sizeof(struct pxa3xx_gcu_shared)) + +/* #define PXA3XX_GCU_DEBUG */ +/* #define PXA3XX_GCU_DEBUG_TIMER */ + +#ifdef PXA3XX_GCU_DEBUG +#define QDUMP (msg) \ + do { \ + QPRINT(priv, KERN_DEBUG, msg); \ + } while (0) +#else +#define QDUMP (msg) do {} while (0) +#endif + +#define QERROR (msg) \ + do { \ + QPRINT(priv, KERN_ERR, msg); \ + } while (0) + +struct pxa3xx_gcu_priv { +void __iomem *mmio_base; +struct clk *clk; +struct pxa3xx_gcu_shared *shared; +dma_addr_t shared_phys; +struct resource *resource_mem; +struct miscdevice misc_dev; +struct file_operations misc_fops; +wait_queue_head_t wait_idle; +wait_queue_head_t wait_next; +spinlock_t spinlock; +struct timeval base_time; +}; + +static inline unsigned long +gc_readl (struct pxa3xx_gcu_priv *priv,unsigned int off) +{ +return __raw_readl (priv->mmio_base + off); +} + +static inline void +gc_writel (struct pxa3xx_gcu_priv *priv,unsigned int off,unsigned long val) +{ +__raw_writel (val, priv->mmio_base + off); +} + +#define QPRINT (priv, level, msg) \ + do { \ + struct timeval tv; \ + struct pxa3xx_gcu_shared *shared = priv->shared; \ + u32 base = gc_readl(priv, REG_GCRBBR); \ + \ + do_gettimeofday(&tv); \ + \ + printk(level "%ld.%03ld.%03ld - %-17s: %-21s (%s, " \ + "hw %5d-%5d, next %5d-%5d, %svalid, STATUS " \ + "0x%02lx, B 0x%08lx [%ld], E %5ld, H %5ld, " \ + "T %5ld)\n", \ + tv.tv_sec - priv->base_time.tv_sec, \ + tv.tv_usec / 1000, tv.tv_usec % 1000, \ + __func__, msg, \ + shared->hw_running ? "running" : " idle", \ + shared->hw_start, \ + shared->hw_end, \ + shared->next_start, \ + shared->next_end, \ + shared->next_valid ? " " : "in", \ + gc_readl(priv, REG_GCISCR), \ + gc_readl(priv, REG_GCRBBR), \ + gc_readl(priv, REG_GCRBLR), \ + (gc_readl(priv, REG_GCRBEXHR) - base) / 4, \ + (gc_readl(priv, REG_GCRBHR) - base) / 4, \ + (gc_readl(priv, REG_GCRBTR) - base) / 4); \ + } while (0) + +static void pxa3xx_gcu_reset (struct pxa3xx_gcu_priv *priv) +{ +QDUMP ("RESET" ); + +/* disable interrupts */ +gc_writel (priv,REG_GCIECR ,0 ); + +/* reset hardware */ +gc_writel (priv,REG_GCCR ,GCCR_ABORT ); +gc_writel (priv,REG_GCCR ,0 ); + +memset (priv->shared,0 ,SHARED_SIZE ); + priv->shared->buffer_phys = priv->shared_phys; + priv->shared->magic =PXA3XX_GCU_SHARED_MAGIC ; + +do_gettimeofday (&priv->base_time); + +/* set up the ring buffer pointers */ +gc_writel (priv,REG_GCRBLR ,0 ); +gc_writel (priv,REG_GCRBBR , priv->shared_phys); +gc_writel (priv,REG_GCRBTR , priv->shared_phys); + +/* enable all IRQs except EOB */ +gc_writel (priv,REG_GCIECR ,IE_ALL & ~IE_EOB ); +} + +static void pxa3xx_gcu_start (struct pxa3xx_gcu_priv *priv) +{ +struct pxa3xx_gcu_shared *sh = priv->shared; + +QDUMP ("Start" ); + +gc_writel (priv,REG_GCRBLR ,0 ); + +/* ring base address */ +gc_writel (priv,REG_GCRBBR , sh->buffer_phys + sh->hw_start *4 ); + +/* ring tail address */ +gc_writel (priv,REG_GCRBTR , sh->buffer_phys + sh->hw_end *4 ); + +/* ring length */ +gc_writel (priv,REG_GCRBLR , + ((sh->hw_end - sh->hw_start +63 ) & ~63 ) *4 ); +} + +static irqreturn_t pxa3xx_gcu_handle_irq (int irq,void *ctx) +{ +struct pxa3xx_gcu_priv *priv = ctx; +struct pxa3xx_gcu_shared *shared = priv->shared; +u32 status =gc_readl (priv,REG_GCISCR ) &IE_ALL ; + +if (!status) +return IRQ_NONE ; + +QDUMP ("-Interrupt" ); + +spin_lock (&priv->spinlock); + + shared->num_interrupts++; + +if (status &IE_EEOB ) { +QDUMP (" [EEOB]" ); + +/* next_valid means user space is not in the process of +* extending the buffer, so we can safely access the +* contents. */ +if (shared->next_valid && + shared->next_start != shared->next_end) { + shared->hw_start = shared->next_start; + shared->hw_end = shared->next_end; + shared->next_start = (shared->hw_end +63 ) & ~63 ; + shared->next_end = shared->next_start; + shared->num_words += shared->hw_end - shared->hw_start; + shared->num_starts++; + +pxa3xx_gcu_start (priv); + +wake_up_all (&priv->wait_next); + }else { +/* There is no more data prepared by the userspace. +* Set hw_running = 0 and wait for the next userspace +* kick-off */ + shared->num_idle++; + shared->hw_running =0 ; + +QDUMP (" '-> Idle." ); + +/* set ring buffer length to zero */ +gc_writel (priv,REG_GCRBLR ,0 ); + +wake_up_all (&priv->wait_next); +wake_up_all (&priv->wait_idle); + } + + shared->num_done++; + }else +QERROR (" [???]" ); + +/* Clear the interrupt */ +gc_writel (priv,REG_GCISCR , status); +spin_unlock (&priv->spinlock); + +return IRQ_HANDLED ; +} + +static int +pxa3xx_gcu_wait_idle (struct pxa3xx_gcu_priv *priv) +{ +int ret =0 ; + +QDUMP ("Waiting for idle..." ); + +while (priv->shared->hw_running) { +int num = priv->shared->num_interrupts; +u32 rbexhr =gc_readl (priv,REG_GCRBEXHR ); + + ret =wait_event_interruptible_timeout (priv->wait_idle, + !priv->shared->hw_running,HZ /4 ); + +if (ret <0 ) +break ; + +if (ret >0 ) +continue ; + +if (gc_readl (priv,REG_GCRBEXHR ) == rbexhr && + priv->shared->num_interrupts == num) { +QERROR ("TIMEOUT" ); + ret = -ETIMEDOUT ; +break ; + } + } + +QDUMP ("done" ); + +return ret; +} + +static int +pxa3xx_gcu_wait_next (struct pxa3xx_gcu_priv *priv) +{ +int ret =0 ; +int num = priv->shared->num_interrupts; + +QDUMP ("Waiting for next..." ); + +while (priv->shared->hw_running && + num == priv->shared->num_interrupts) { +u32 rbexhr =gc_readl (priv,REG_GCRBEXHR ); + + ret =wait_event_interruptible_timeout (priv->wait_next, + (!priv->shared->hw_running || + num != priv->shared->num_interrupts),HZ /4 ); + +if (ret <0 ) +break ; + +if (ret >0 ) +continue ; + +if (gc_readl (priv,REG_GCRBEXHR ) == rbexhr) { +QERROR ("TIMEOUT" ); + ret = -ETIMEDOUT ; +break ; + } + } + +QDUMP ("done" ); + +return ret; +} + +/* Misc device layer */ + +static int +pxa3xx_gcu_misc_ioctl (struct inode *inode,struct file *filp, +unsigned int cmd,unsigned long arg) +{ +unsigned long flags; +struct pxa3xx_gcu_priv *priv = + container_of(filp->f_op, struct pxa3xx_gcu_priv, misc_fops); + +switch (cmd) { +case PXA3XX_GCU_IOCTL_RESET : +spin_lock_irqsave (&priv->spinlock, flags); +pxa3xx_gcu_reset (priv); +spin_unlock_irqrestore (&priv->spinlock, flags); +return 0 ; + +case PXA3XX_GCU_IOCTL_START : +spin_lock_irqsave (&priv->spinlock, flags); +pxa3xx_gcu_start (priv); +spin_unlock_irqrestore (&priv->spinlock, flags); +return 0 ; + +case PXA3XX_GCU_IOCTL_WAIT_IDLE : +/* Does not need to be atomic. There's a lock in user space, +* but anyhow, this is just for statistics. */ + priv->shared->num_wait_idle++; +return pxa3xx_gcu_wait_idle (priv); + +case PXA3XX_GCU_IOCTL_WAIT_NEXT : +/* Does not need to be atomic. There's a lock in user space, +* but anyhow, this is just for statistics. */ + priv->shared->num_wait_next++; +return pxa3xx_gcu_wait_next (priv); + } + +return -ENOSYS ; +} + +static int +pxa3xx_gcu_misc_mmap (struct file *filp,struct vm_area_struct *vma) +{ +unsigned int size = vma->vm_end - vma->vm_start; +struct pxa3xx_gcu_priv *priv = + container_of(filp->f_op, struct pxa3xx_gcu_priv, misc_fops); + +switch (vma->vm_pgoff) { +case 0 : +/* hand out the shared data area */ +if (size !=SHARED_SIZE ) +return -EINVAL ; + +return dma_mmap_coherent (NULL , vma, + priv->shared, priv->shared_phys, size); + +case SHARED_SIZE >>PAGE_SHIFT : +/* hand out the MMIO base for direct register access +* from userspace */ +if (size !=resource_size (priv->resource_mem)) +return -EINVAL ; + + vma->vm_flags |=VM_IO ; + vma->vm_page_prot =pgprot_noncached (vma->vm_page_prot); + +return io_remap_pfn_range (vma, vma->vm_start, + priv->resource_mem->start >>PAGE_SHIFT , + size, vma->vm_page_prot); + } + +return -EINVAL ; +} + + +#ifdef PXA3XX_GCU_DEBUG_TIMER +static struct timer_list pxa3xx_gcu_debug_timer; + +static void pxa3xx_gcu_debug_timedout (unsigned long ptr) +{ +struct pxa3xx_gcu_priv *priv = (struct pxa3xx_gcu_priv *) ptr; + +QERROR ("Timer DUMP" ); + +/* init the timer structure */ +init_timer (&pxa3xx_gcu_debug_timer); + pxa3xx_gcu_debug_timer.function = pxa3xx_gcu_debug_timedout; + pxa3xx_gcu_debug_timer.data = ptr; + pxa3xx_gcu_debug_timer.expires = jiffies +5 *HZ ;/* one second */ + +add_timer (&pxa3xx_gcu_debug_timer); +} + +static void pxa3xx_gcu_init_debug_timer (void ) +{ +pxa3xx_gcu_debug_timedout ((unsigned long ) &pxa3xx_gcu_debug_timer); +} +#else +static inline void pxa3xx_gcu_init_debug_timer (void ) {} +#endif + +static int __devinit pxa3xx_gcu_probe (struct platform_device *dev) +{ +int ret, irq; +struct resource *r; +struct pxa3xx_gcu_priv *priv; + + priv =kzalloc (sizeof (struct pxa3xx_gcu_priv ),GFP_KERNEL ); +if (!priv) +return -ENOMEM ; + +init_waitqueue_head (&priv->wait_idle); +init_waitqueue_head (&priv->wait_next); +spin_lock_init (&priv->spinlock); + +/* we allocate the misc device structure as part of our own allocation, +* so we can get a pointer to our priv structure later on with +* container_of(). This isn't really necessary as we have a fixed minor +* number anyway, but this is to avoid statics. */ + + priv->misc_fops.owner =THIS_MODULE ; + priv->misc_fops.ioctl = pxa3xx_gcu_misc_ioctl; + priv->misc_fops.mmap = pxa3xx_gcu_misc_mmap; + + priv->misc_dev.minor =MISCDEV_MINOR , + priv->misc_dev.name =DRV_NAME , + priv->misc_dev.fops = &priv->misc_fops, + +/* register misc device */ + ret =misc_register (&priv->misc_dev); +if (ret <0 ) { +dev_err (&dev->dev,"misc_register() for minor %d failed\n" , +MISCDEV_MINOR ); + goto err_free_priv; + } + +/* handle IO resources */ + r =platform_get_resource (dev,IORESOURCE_MEM ,0 ); +if (r ==NULL ) { +dev_err (&dev->dev,"no I/O memory resource defined\n" ); + ret = -ENODEV ; + goto err_misc_deregister; + } + +if (!request_mem_region (r->start,resource_size (r), dev->name)) { +dev_err (&dev->dev,"failed to request I/O memory\n" ); + ret = -EBUSY ; + goto err_misc_deregister; + } + + priv->mmio_base =ioremap_nocache (r->start,resource_size (r)); +if (!priv->mmio_base) { +dev_err (&dev->dev,"failed to map I/O memory\n" ); + ret = -EBUSY ; + goto err_free_mem_region; + } + +/* allocate dma memory */ + priv->shared =dma_alloc_coherent (&dev->dev,SHARED_SIZE , + &priv->shared_phys,GFP_KERNEL ); + +if (!priv->shared) { +dev_err (&dev->dev,"failed to allocate DMA memory\n" ); + ret = -ENOMEM ; + goto err_free_io; + } + +/* enable the clock */ + priv->clk =clk_get (&dev->dev,NULL ); +if (IS_ERR (priv->clk)) { +dev_err (&dev->dev,"failed to get clock\n" ); + ret = -ENODEV ; + goto err_free_dma; + } + + ret =clk_enable (priv->clk); +if (ret <0 ) { +dev_err (&dev->dev,"failed to enable clock\n" ); + goto err_put_clk; + } + +/* request the IRQ */ + irq =platform_get_irq (dev,0 ); +if (irq <0 ) { +dev_err (&dev->dev,"no IRQ defined\n" ); + ret = -ENODEV ; + goto err_put_clk; + } + + ret =request_irq (irq, pxa3xx_gcu_handle_irq, +IRQF_DISABLED ,DRV_NAME , priv); +if (ret) { +dev_err (&dev->dev,"request_irq failed\n" ); + ret = -EBUSY ; + goto err_put_clk; + } + +platform_set_drvdata (dev, priv); + priv->resource_mem = r; +pxa3xx_gcu_reset (priv); +pxa3xx_gcu_init_debug_timer (); + +dev_info (&dev->dev,"registered @0x%p, DMA 0x%p (%d bytes), IRQ %d\n" , + (void *) r->start, (void *) priv->shared_phys, +SHARED_SIZE , irq); +return 0 ; + +err_put_clk: +clk_put (priv->clk); + +err_free_dma: +dma_free_coherent (&dev->dev,SHARED_SIZE , + priv->shared, priv->shared_phys); + +err_free_io: +iounmap (priv->mmio_base); +clk_disable (priv->clk); + +err_free_mem_region: +release_mem_region (r->start,resource_size (r)); + +err_misc_deregister: +misc_deregister (&priv->misc_dev); + +err_free_priv: +platform_set_drvdata (dev,NULL ); +kfree (priv); +return ret; +} + +static int __devexit pxa3xx_gcu_remove(struct platform_device *dev) +{ +struct pxa3xx_gcu_priv *priv =platform_get_drvdata (dev); +struct resource *r = priv->resource_mem; + +misc_deregister (&priv->misc_dev); +dma_free_coherent (&dev->dev,SHARED_SIZE , + priv->shared, priv->shared_phys); +iounmap (priv->mmio_base); +release_mem_region (r->start,resource_size (r)); +platform_set_drvdata (dev,NULL ); +clk_disable (priv->clk); +kfree (priv); + +return 0 ; +} + +static struct platform_driver pxa3xx_gcu_driver = { + .probe = pxa3xx_gcu_probe, + .remove =__devexit_p (pxa3xx_gcu_remove), + .driver = { + .owner =THIS_MODULE , + .name =DRV_NAME , + }, +}; + +static int __init pxa3xx_gcu_init (void ) +{ +return platform_driver_register (&pxa3xx_gcu_driver); +} + +static void __exit pxa3xx_gcu_exit(void) +{ +platform_driver_unregister (&pxa3xx_gcu_driver); +} + +module_init (pxa3xx_gcu_init); +module_exit (pxa3xx_gcu_exit); + +MODULE_DESCRIPTION ("PXA3xx graphics controller unit driver" ); +MODULE_LICENSE ("GPL" ); +MODULE_ALIAS_MISCDEV (MISCDEV_MINOR ); +MODULE_AUTHOR ("Janine Kropp <[email protected]>, " +"Denis Oliver Kropp <[email protected]>, " +"Daniel Mack <[email protected]>" ); +drivers/video/pxa3xx-gcu.h new file +49
-0@@ -0,0 +1,49 @@ +#ifndef __PXA3XX_GCU_H__ +#define __PXA3XX_GCU_H__ + +#include <linux/types.h> + +/* Number of 32bit words in display list (ring buffer). */ +#define PXA3XX_GCU_BUFFER_WORDS ((256 * 1024 - 256) / 4) + +/* To be increased when breaking the ABI */ +#define PXA3XX_GCU_SHARED_MAGIC 0x30000001 + +struct pxa3xx_gcu_shared { +u32 buffer[PXA3XX_GCU_BUFFER_WORDS ]; + +bool hw_running; +int hw_start; +int hw_end; + +int next_start; +int next_end; +int next_valid; + +unsigned long buffer_phys; + +unsigned int num_words; +unsigned int num_starts; +unsigned int num_done; +unsigned int num_interrupts; +unsigned int num_wait_idle; +unsigned int num_wait_next; +unsigned int num_idle; + +u32 magic; +}; + +struct pxa3xx_gcu_register { +u32 address;/* in */ +u32 value;/* in/out */ +}; + +/* Initialization and synchronization. +* Hardware is started from user space via MMIO or via ioctl. */ +#define PXA3XX_GCU_IOCTL_RESET _IO('G', 0) +#define PXA3XX_GCU_IOCTL_START _IO('G', 1) +#define PXA3XX_GCU_IOCTL_WAIT_IDLE _IO('G', 2) +#define PXA3XX_GCU_IOCTL_WAIT_NEXT _IO('G', 3) + +#endif /* __PXA3XX_GCU_H__ */ +Signature · 1 line
1.6.3.1
PATCH 4/4 video: add driver for PXA3xx 2D graphics accelerator
Hi,
this patch keeps on lurking around in tree since many months, and I would still like to push that to mainline. Is anyone interested to give some feedback on this? Below is a rebased version that applies to and compiles with Linus' current git HEAD.
Thanks,
Daniel
Quoted 5 lines · Daniel Mack
Was anyone able to give that driver a try?
Haojian, you wanted to have a look IIRC?I would appreciate seeing this in .35 if possible.