[PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Hi,
this patch converts driver/video/cyber200fb.c to framebuffer_alloc:

======== drivers/video/cyber2000fb.c 1.33 ========
D 1.33 03/09/13 23:21:10+02:00 kronos@kronoz.cjb.net 35 34 108/99/1650
P drivers/video/cyber2000fb.c
C switch to framebuffer_alloc
------------------------------------------------

===== drivers/video/cyber2000fb.c 1.32 vs 1.33 =====
--- 1.32/drivers/video/cyber2000fb.c	Fri Aug 22 08:27:08 2003
+++ 1.33/drivers/video/cyber2000fb.c	Sat Sep 13 23:21:10 2003
@@ -62,7 +62,7 @@
 #include "cyber2000fb.h"
 
 struct cfb_info {
-	struct fb_info		fb;
+	struct fb_info		*fb;
 	struct display_switch	*dispsw;
 	struct display		*display;
 	struct pci_dev		*dev;
@@ -97,6 +97,8 @@
 MODULE_PARM(default_font, "s");
 MODULE_PARM_DESC(default_font, "Default font name");
 
+static void release_cfb_info(struct fb_info *);
+
 /*
  * Our access methods.
  */
@@ -148,10 +150,10 @@
 static void
 cyber2000fb_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 	unsigned long dst, col;
 
-	if (!(cfb->fb.var.accel_flags & FB_ACCELF_TEXT)) {
+	if (!(info->var.accel_flags & FB_ACCELF_TEXT)) {
 		cfb_fillrect(info, rect);
 		return;
 	}
@@ -161,12 +163,12 @@
 	cyber2000fb_writew(rect->height - 1, CO_REG_PIXHEIGHT, cfb);
 
 	col = rect->color;
-	if (cfb->fb.var.bits_per_pixel > 8)
-		col = ((u32 *)cfb->fb.pseudo_palette)[col];
+	if (info->var.bits_per_pixel > 8)
+		col = ((u32 *)info->pseudo_palette)[col];
 	cyber2000fb_writel(col, CO_REG_FGCOLOUR, cfb);
 
-	dst = rect->dx + rect->dy * cfb->fb.var.xres_virtual;
-	if (cfb->fb.var.bits_per_pixel == 24) {
+	dst = rect->dx + rect->dy * info->var.xres_virtual;
+	if (info->var.bits_per_pixel == 24) {
 		cyber2000fb_writeb(dst, CO_REG_X_PHASE, cfb);
 		dst *= 3;
 	}
@@ -180,11 +182,11 @@
 static void
 cyber2000fb_copyarea(struct fb_info *info, const struct fb_copyarea *region)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 	unsigned int cmd = CO_CMD_L_PATTERN_FGCOL;
 	unsigned long src, dst;
 
-	if (!(cfb->fb.var.accel_flags & FB_ACCELF_TEXT)) {
+	if (!(info->var.accel_flags & FB_ACCELF_TEXT)) {
 		cfb_copyarea(info, region);
 		return;
 	}
@@ -193,8 +195,8 @@
 	cyber2000fb_writew(region->width - 1, CO_REG_PIXWIDTH, cfb);
 	cyber2000fb_writew(region->height - 1, CO_REG_PIXHEIGHT, cfb);
 
-	src = region->sx + region->sy * cfb->fb.var.xres_virtual;
-	dst = region->dx + region->dy * cfb->fb.var.xres_virtual;
+	src = region->sx + region->sy * info->var.xres_virtual;
+	dst = region->dx + region->dy * info->var.xres_virtual;
 
 	if (region->sx < region->dx) {
 		src += region->width - 1;
@@ -203,12 +205,12 @@
 	}
 
 	if (region->sy < region->dy) {
-		src += (region->height - 1) * cfb->fb.var.xres_virtual;
-		dst += (region->height - 1) * cfb->fb.var.xres_virtual;
+		src += (region->height - 1) * info->var.xres_virtual;
+		dst += (region->height - 1) * info->var.xres_virtual;
 		cmd |= CO_CMD_L_INC_UP;
 	}
 
-	if (cfb->fb.var.bits_per_pixel == 24) {
+	if (info->var.bits_per_pixel == 24) {
 		cyber2000fb_writeb(dst, CO_REG_X_PHASE, cfb);
 		src *= 3;
 		dst *= 3;
@@ -234,10 +236,10 @@
 
 static int cyber2000fb_sync(struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 	int count = 100000;
 
-	if (!(cfb->fb.var.accel_flags & FB_ACCELF_TEXT))
+	if (!(info->var.accel_flags & FB_ACCELF_TEXT))
 		return 0;
 
 	while (cyber2000fb_readb(CO_REG_CONTROL, cfb) & CO_CTRL_BUSY) {
@@ -269,12 +271,12 @@
 cyber2000fb_setcolreg(u_int regno, u_int red, u_int green, u_int blue,
 		      u_int transp, struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
-	struct fb_var_screeninfo *var = &cfb->fb.var;
+	struct cfb_info *cfb = info->par;
+	struct fb_var_screeninfo *var = &info->var;
 	u32 pseudo_val;
 	int ret = 1;
 
-	switch (cfb->fb.fix.visual) {
+	switch (info->fix.visual) {
 	default:
 		return 1;
 
@@ -400,7 +402,7 @@
 	 * Now set our pseudo palette for the CFB16/24/32 drivers.
 	 */
 	if (regno < 16)
-		((u32 *)cfb->fb.pseudo_palette)[regno] = pseudo_val;
+		((u32 *)info->pseudo_palette)[regno] = pseudo_val;
 
 	return ret;
 }
@@ -744,7 +746,7 @@
 static int
 cyber2000fb_check_var(struct fb_var_screeninfo *var, struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 	struct par_info hw;
 	unsigned int mem;
 	int err;
@@ -831,8 +833,8 @@
 	}
 
 	mem = var->xres_virtual * var->yres_virtual * (var->bits_per_pixel / 8);
-	if (mem > cfb->fb.fix.smem_len)
-		var->yres_virtual = cfb->fb.fix.smem_len * 8 /
+	if (mem > info->fix.smem_len)
+		var->yres_virtual = info->fix.smem_len * 8 /
 			(var->bits_per_pixel * var->xres_virtual);
 
 	if (var->yres > var->yres_virtual)
@@ -853,8 +855,8 @@
 
 static int cyber2000fb_set_par(struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
-	struct fb_var_screeninfo *var = &cfb->fb.var;
+	struct cfb_info *cfb = info->par;
+	struct fb_var_screeninfo *var = &info->var;
 	struct par_info hw;
 	unsigned int mem;
 
@@ -924,7 +926,7 @@
 		hw.fetch <<= 1;
 	hw.fetch += 1;
 
-	cfb->fb.fix.line_length	= var->xres_virtual * var->bits_per_pixel / 8;
+	info->fix.line_length	= var->xres_virtual * var->bits_per_pixel / 8;
 
 	/*
 	 * Same here - if the size of the video mode exceeds the
@@ -933,8 +935,8 @@
 	 * In theory, since NetWinders contain just one VGA card,
 	 * we should never end up hitting this problem.
 	 */
-	mem = cfb->fb.fix.line_length * var->yres_virtual;
-	BUG_ON(mem > cfb->fb.fix.smem_len);
+	mem = info->fix.line_length * var->yres_virtual;
+	BUG_ON(mem > info->fix.smem_len);
 
 	/*
 	 * 8bpp displays are always pseudo colour.  16bpp and above
@@ -943,11 +945,11 @@
 	 * palettes, true colour does not.)
 	 */
 	if (var->bits_per_pixel == 8)
-		cfb->fb.fix.visual = FB_VISUAL_PSEUDOCOLOR;
+		info->fix.visual = FB_VISUAL_PSEUDOCOLOR;
 	else if (hw.ramdac & RAMDAC_BYPASS)
-		cfb->fb.fix.visual = FB_VISUAL_TRUECOLOR;
+		info->fix.visual = FB_VISUAL_TRUECOLOR;
 	else
-		cfb->fb.fix.visual = FB_VISUAL_DIRECTCOLOR;
+		info->fix.visual = FB_VISUAL_DIRECTCOLOR;
 
 	cyber2000fb_set_timing(cfb, &hw);
 	cyber2000fb_update_start(cfb, var);
@@ -962,18 +964,18 @@
 static int
 cyber2000fb_pan_display(struct fb_var_screeninfo *var, struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 
 	if (cyber2000fb_update_start(cfb, var))
 		return -EINVAL;
 
-	cfb->fb.var.xoffset = var->xoffset;
-	cfb->fb.var.yoffset = var->yoffset;
+	info->var.xoffset = var->xoffset;
+	info->var.yoffset = var->yoffset;
 
 	if (var->vmode & FB_VMODE_YWRAP) {
-		cfb->fb.var.vmode |= FB_VMODE_YWRAP;
+		info->var.vmode |= FB_VMODE_YWRAP;
 	} else {
-		cfb->fb.var.vmode &= ~FB_VMODE_YWRAP;
+		info->var.vmode &= ~FB_VMODE_YWRAP;
 	}
 
 	return 0;
@@ -998,7 +1000,7 @@
  */
 static int cyber2000fb_blank(int blank, struct fb_info *info)
 {
-	struct cfb_info *cfb = (struct cfb_info *)info;
+	struct cfb_info *cfb = info->par;
 	unsigned int sync = 0;
 	int i;
 
@@ -1111,7 +1113,7 @@
 
 void cyber2000fb_get_fb_var(struct cfb_info *cfb, struct fb_var_screeninfo *var)
 {
-	memcpy(var, &cfb->fb.var, sizeof(struct fb_var_screeninfo));
+	memcpy(var, &cfb->fb->var, sizeof(struct fb_var_screeninfo));
 }
 
 /*
@@ -1122,13 +1124,13 @@
 	if (int_cfb_info != NULL) {
 		info->dev	      = int_cfb_info->dev;
 		info->regs	      = int_cfb_info->regs;
-		info->fb	      = int_cfb_info->fb.screen_base;
-		info->fb_size	      = int_cfb_info->fb.fix.smem_len;
+		info->fb	      = int_cfb_info->fb->screen_base;
+		info->fb_size	      = int_cfb_info->fb->fix.smem_len;
 		info->enable_extregs  = cyber2000fb_enable_extregs;
 		info->disable_extregs = cyber2000fb_disable_extregs;
 		info->info            = int_cfb_info;
 
-		strlcpy(info->dev_name, int_cfb_info->fb.fix.id, sizeof(info->dev_name));
+		strlcpy(info->dev_name, int_cfb_info->fb->fix.id, sizeof(info->dev_name));
 	}
 
 	return int_cfb_info != NULL;
@@ -1220,17 +1222,19 @@
 }
 
 static struct cfb_info * __devinit
-cyberpro_alloc_fb_info(unsigned int id, char *name)
+cyberpro_alloc_fb_info(unsigned int id, char *name, struct device *dev)
 {
 	struct cfb_info *cfb;
+	struct fb_info *fb_info;
 
-	cfb = kmalloc(sizeof(struct cfb_info) +
-		       sizeof(u32) * 16, GFP_KERNEL);
+	fb_info = framebuffer_alloc(sizeof(struct cfb_info) + 32 * 16, dev);
 
-	if (!cfb)
+	if (!fb_info)
 		return NULL;
 
-	memset(cfb, 0, sizeof(struct cfb_info));
+	fb_info->release = release_cfb_info;
+	cfb = fb_info->par;
+	cfb->fb = fb_info;
 
 	cfb->id			= id;
 
@@ -1248,43 +1252,43 @@
 	else
 		cfb->divisors[3] = 6;
 
-	strcpy(cfb->fb.fix.id, name);
+	strcpy(fb_info->fix.id, name);
 
-	cfb->fb.fix.type	= FB_TYPE_PACKED_PIXELS;
-	cfb->fb.fix.type_aux	= 0;
-	cfb->fb.fix.xpanstep	= 0;
-	cfb->fb.fix.ypanstep	= 1;
-	cfb->fb.fix.ywrapstep	= 0;
+	fb_info->fix.type	= FB_TYPE_PACKED_PIXELS;
+	fb_info->fix.type_aux	= 0;
+	fb_info->fix.xpanstep	= 0;
+	fb_info->fix.ypanstep	= 1;
+	fb_info->fix.ywrapstep	= 0;
 
 	switch (id) {
 	case ID_IGA_1682:
-		cfb->fb.fix.accel = 0;
+		fb_info->fix.accel = 0;
 		break;
 
 	case ID_CYBERPRO_2000:
-		cfb->fb.fix.accel = FB_ACCEL_IGS_CYBER2000;
+		fb_info->fix.accel = FB_ACCEL_IGS_CYBER2000;
 		break;
 
 	case ID_CYBERPRO_2010:
-		cfb->fb.fix.accel = FB_ACCEL_IGS_CYBER2010;
+		fb_info->fix.accel = FB_ACCEL_IGS_CYBER2010;
 		break;
 
 	case ID_CYBERPRO_5000:
-		cfb->fb.fix.accel = FB_ACCEL_IGS_CYBER5000;
+		fb_info->fix.accel = FB_ACCEL_IGS_CYBER5000;
 		break;
 	}
 
-	cfb->fb.var.nonstd	= 0;
-	cfb->fb.var.activate	= FB_ACTIVATE_NOW;
-	cfb->fb.var.height	= -1;
-	cfb->fb.var.width	= -1;
-	cfb->fb.var.accel_flags	= FB_ACCELF_TEXT;
-
-	cfb->fb.fbops		= &cyber2000fb_ops;
-	cfb->fb.flags		= FBINFO_FLAG_DEFAULT;
-	cfb->fb.pseudo_palette	= (void *)(cfb + 1);
+	fb_info->var.nonstd	= 0;
+	fb_info->var.activate	= FB_ACTIVATE_NOW;
+	fb_info->var.height	= -1;
+	fb_info->var.width	= -1;
+	fb_info->var.accel_flags	= FB_ACCELF_TEXT;
+
+	fb_info->fbops		= &cyber2000fb_ops;
+	fb_info->flags		= FBINFO_FLAG_DEFAULT;
+	fb_info->pseudo_palette	= (void *)(cfb + 1);
 
-	fb_alloc_cmap(&cfb->fb.cmap, NR_PALETTE, 0);
+	fb_alloc_cmap(&fb_info->cmap, NR_PALETTE, 0);
 
 	return cfb;
 }
@@ -1296,9 +1300,9 @@
 		/*
 		 * Free the colourmap
 		 */
-		fb_alloc_cmap(&cfb->fb.cmap, 0, 0);
+		fb_alloc_cmap(&cfb->fb->cmap, 0, 0);
 
-		kfree(cfb);
+		kfree(cfb->fb);
 	}
 }
 
@@ -1363,23 +1367,22 @@
 	default:		smem_size = 0x00100000; break;
 	}
 
-	cfb->fb.fix.smem_len   = smem_size;
-	cfb->fb.fix.mmio_len   = MMIO_SIZE;
-	cfb->fb.screen_base    = cfb->region;
-	cfb->fb.dev            = &cfb->dev->dev;
+	cfb->fb->fix.smem_len   = smem_size;
+	cfb->fb->fix.mmio_len   = MMIO_SIZE;
+	cfb->fb->screen_base    = cfb->region;
 
 	err = -EINVAL;
-	if (!fb_find_mode(&cfb->fb.var, &cfb->fb, NULL, NULL, 0,
+	if (!fb_find_mode(&cfb->fb->var, cfb->fb, NULL, NULL, 0,
 	    		  &cyber2000fb_default_mode, 8)) {
-		printk("%s: no valid mode found\n", cfb->fb.fix.id);
+		printk("%s: no valid mode found\n", cfb->fb->fix.id);
 		goto failed;
 	}
 
-	cfb->fb.var.yres_virtual = cfb->fb.fix.smem_len * 8 /
-			(cfb->fb.var.bits_per_pixel * cfb->fb.var.xres_virtual);
+	cfb->fb->var.yres_virtual = cfb->fb->fix.smem_len * 8 /
+			(cfb->fb->var.bits_per_pixel * cfb->fb->var.xres_virtual);
 
-	if (cfb->fb.var.yres_virtual < cfb->fb.var.yres)
-		cfb->fb.var.yres_virtual = cfb->fb.var.yres;
+	if (cfb->fb->var.yres_virtual < cfb->fb->var.yres)
+		cfb->fb->var.yres_virtual = cfb->fb->var.yres;
 
 //	fb_set_var(&cfb->fb.var, -1, &cfb->fb);
 
@@ -1389,18 +1392,18 @@
 	 * the precision and fit the results into 32-bit registers.
 	 *  (1953125000 * 512 = 1e12)
 	 */
-	h_sync = 1953125000 / cfb->fb.var.pixclock;
-	h_sync = h_sync * 512 / (cfb->fb.var.xres + cfb->fb.var.left_margin +
-		 cfb->fb.var.right_margin + cfb->fb.var.hsync_len);
-	v_sync = h_sync / (cfb->fb.var.yres + cfb->fb.var.upper_margin +
-		 cfb->fb.var.lower_margin + cfb->fb.var.vsync_len);
+	h_sync = 1953125000 / cfb->fb->var.pixclock;
+	h_sync = h_sync * 512 / (cfb->fb->var.xres + cfb->fb->var.left_margin +
+		 cfb->fb->var.right_margin + cfb->fb->var.hsync_len);
+	v_sync = h_sync / (cfb->fb->var.yres + cfb->fb->var.upper_margin +
+		 cfb->fb->var.lower_margin + cfb->fb->var.vsync_len);
 
 	printk(KERN_INFO "%s: %dKiB VRAM, using %dx%d, %d.%03dkHz, %dHz\n",
-		cfb->fb.fix.id, cfb->fb.fix.smem_len >> 10,
-		cfb->fb.var.xres, cfb->fb.var.yres,
+		cfb->fb->fix.id, cfb->fb->fix.smem_len >> 10,
+		cfb->fb->var.xres, cfb->fb->var.yres,
 		h_sync / 1000, h_sync % 1000, v_sync);
 
-	err = register_framebuffer(&cfb->fb);
+	err = register_framebuffer(cfb->fb);
 
 failed:
 	return err;
@@ -1420,7 +1423,7 @@
 	 * Restore the old video mode and the palette.
 	 * We also need to tell fbcon to redraw the console.
 	 */
-	cyber2000fb_set_par(&cfb->fb);
+	cyber2000fb_set_par(cfb->fb);
 }
 
 #ifdef CONFIG_ARCH_SHARK
@@ -1445,8 +1448,8 @@
 		goto failed_ioremap;
 
 	cfb->regs = cfb->region + MMIO_OFFSET;
-	cfb->fb.fix.mmio_start = FB_START + MMIO_OFFSET;
-	cfb->fb.fix.smem_start = FB_START;
+	cfb->fb->fix.mmio_start = FB_START + MMIO_OFFSET;
+	cfb->fb->fix.smem_start = FB_START;
 
 	/*
 	 * Bring up the hardware.  This is expected to enable access
@@ -1541,7 +1544,7 @@
 	 */
 	val = cyber2000_grphr(EXT_BUS_CTL, cfb);
 	if (!(val & EXT_BUS_CTL_PCIBURST_WRITE)) {
-		printk(KERN_INFO "%s: enabling PCI bursts\n", cfb->fb.fix.id);
+		printk(KERN_INFO "%s: enabling PCI bursts\n", cfb->fb->fix.id);
 
 		val |= EXT_BUS_CTL_PCIBURST_WRITE;
 
@@ -1572,7 +1575,7 @@
 		return err;
 
 	err = -ENOMEM;
-	cfb = cyberpro_alloc_fb_info(id->driver_data, name);
+	cfb = cyberpro_alloc_fb_info(id->driver_data, name, &dev->dev);
 	if (!cfb)
 		goto failed_release;
 
@@ -1583,8 +1586,8 @@
 		goto failed_ioremap;
 
 	cfb->regs = cfb->region + MMIO_OFFSET;
-	cfb->fb.fix.mmio_start = pci_resource_start(dev, 0) + MMIO_OFFSET;
-	cfb->fb.fix.smem_start = pci_resource_start(dev, 0);
+	cfb->fb->fix.mmio_start = pci_resource_start(dev, 0) + MMIO_OFFSET;
+	cfb->fb->fix.smem_start = pci_resource_start(dev, 0);
 
 	/*
 	 * Bring up the hardware.  This is expected to enable access
@@ -1635,6 +1638,16 @@
 	return err;
 }
 
+static void release_cfb_info(struct fb_info *info) {
+	struct cfb_info *cfb = info->par;
+
+	iounmap(cfb->region);
+	fb_alloc_cmap(&info->cmap, 0, 0);
+
+	if (cfb->dev)
+		pci_release_regions(cfb->dev);
+}
+
 static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
 {
 	struct cfb_info *cfb = pci_get_drvdata(dev);
@@ -1645,12 +1658,10 @@
 		 * we will be leaving hooks that could cause
 		 * oopsen laying around.
 		 */
-		if (unregister_framebuffer(&cfb->fb))
+		if (unregister_framebuffer(cfb->fb))
 			printk(KERN_WARNING "%s: danger Will Robinson, "
 				"danger danger!  Oopsen imminent!\n",
-				cfb->fb.fix.id);
-		iounmap(cfb->region);
-		cyberpro_free_fb_info(cfb);
+				cfb->fb->fix.id);
 
 		/*
 		 * Ensure that the driver data is no longer
@@ -1659,8 +1670,6 @@
 		pci_set_drvdata(dev, NULL);
 		if (cfb == int_cfb_info)
 			int_cfb_info = NULL;
-
-		pci_release_regions(dev);
 	}
 }
 


Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
"La Via prosegue senza fine."
	Bilbo Baggins - Il Signore degli Anelli - J.R.R. Tolkien


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Mon, Sep 15, 2003 at 09:43:29PM +0200, Kronos wrote:
> Hi,
> this patch converts driver/video/cyber200fb.c to framebuffer_alloc:
> 
> ======== drivers/video/cyber2000fb.c 1.33 ========
> D 1.33 03/09/13 23:21:10+02:00 kronos@kronoz.cjb.net 35 34 108/99/1650
> P drivers/video/cyber2000fb.c
> C switch to framebuffer_alloc
> ------------------------------------------------
> 
> ===== drivers/video/cyber2000fb.c 1.32 vs 1.33 =====
> --- 1.32/drivers/video/cyber2000fb.c	Fri Aug 22 08:27:08 2003
> +++ 1.33/drivers/video/cyber2000fb.c	Sat Sep 13 23:21:10 2003
> @@ -62,7 +62,7 @@
>  #include "cyber2000fb.h"
>  
>  struct cfb_info {
> -	struct fb_info		fb;
> +	struct fb_info		*fb;

Oh god, do we have to add yet another level of indirection all over
the framebuffer code?

> @@ -1635,6 +1638,16 @@
>  	return err;
>  }
>  
> +static void release_cfb_info(struct fb_info *info) {
> +	struct cfb_info *cfb = info->par;
> +
> +	iounmap(cfb->region);
> +	fb_alloc_cmap(&info->cmap, 0, 0);
> +
> +	if (cfb->dev)
> +		pci_release_regions(cfb->dev);
> +}
> +
>  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
>  {
>  	struct cfb_info *cfb = pci_get_drvdata(dev);

Who says "cfb->dev" remains valid after the PCI device has been removed.
This looks like a perfect use-after-free bug waiting to happen.

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Mon, Sep 15, 2003 at 10:07:42PM +0100, Russell King ha scritto: 
> >  struct cfb_info {
> > -	struct fb_info		fb;
> > +	struct fb_info		*fb;
> 
> Oh god, do we have to add yet another level of indirection all over
> the framebuffer code?

Ok, I've been to vague...

Now there is  a class_dev embedded in fb_info which  registered with the
driver model. We need a dynamically allocated struct fb_info.

> 
> > @@ -1635,6 +1638,16 @@
> >  	return err;
> >  }
> >  
> > +static void release_cfb_info(struct fb_info *info) {
> > +	struct cfb_info *cfb = info->par;
> > +
> > +	iounmap(cfb->region);
> > +	fb_alloc_cmap(&info->cmap, 0, 0);
> > +
> > +	if (cfb->dev)
> > +		pci_release_regions(cfb->dev);
> > +}
> > +
> >  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
> >  {
> >  	struct cfb_info *cfb = pci_get_drvdata(dev);
> 
> Who says "cfb->dev" remains valid after the PCI device has been removed.
> This looks like a perfect use-after-free bug waiting to happen.

cfb->dev is  refcounted, it  won't go  away until we  are done  with the
cleanup. Maybe I misread  driver core code...

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
Windows NT: Designed for the Internet. The Internet: Designed for Unix.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Mon, Sep 15, 2003 at 11:28:09PM +0200, Kronos wrote:
> > > @@ -1635,6 +1638,16 @@
> > >  	return err;
> > >  }
> > >  
> > > +static void release_cfb_info(struct fb_info *info) {
> > > +	struct cfb_info *cfb = info->par;
> > > +
> > > +	iounmap(cfb->region);
> > > +	fb_alloc_cmap(&info->cmap, 0, 0);
> > > +
> > > +	if (cfb->dev)
> > > +		pci_release_regions(cfb->dev);
> > > +}
> > > +
> > >  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
> > >  {
> > >  	struct cfb_info *cfb = pci_get_drvdata(dev);
> > 
> > Who says "cfb->dev" remains valid after the PCI device has been removed.
> > This looks like a perfect use-after-free bug waiting to happen.
> 
> cfb->dev is  refcounted, it  won't go  away until we  are done  with the
> cleanup. Maybe I misread  driver core code...

pci_request_regions / pci_release_regions does not perform any reference
counting on the pci device.

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Mon, Sep 15, 2003 at 10:33:31PM +0100, Russell King ha scritto: 
> > > > +static void release_cfb_info(struct fb_info *info) {
> > > > +	struct cfb_info *cfb = info->par;
> > > > +
> > > > +	iounmap(cfb->region);
> > > > +	fb_alloc_cmap(&info->cmap, 0, 0);
> > > > +
> > > > +	if (cfb->dev)
> > > > +		pci_release_regions(cfb->dev);
> > > > +}
> > > > +
> > > >  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
> > > >  {
> > > >  	struct cfb_info *cfb = pci_get_drvdata(dev);
> > > 
> > > Who says "cfb->dev" remains valid after the PCI device has been removed.
> > > This looks like a perfect use-after-free bug waiting to happen.
> > 
> > cfb->dev is  refcounted, it  won't go  away until we  are done  with the
> > cleanup. Maybe I misread  driver core code...
> 
> pci_request_regions / pci_release_regions does not perform any reference
> counting on the pci device.

I was referring to class_dev stuff, class_device_add in
drivers/base/class.c:262

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
"L'abilita` politica e` l'abilita` di prevedere quello che
 accadra` domani, la prossima settimana, il prossimo mese e
 l'anno prossimo. E di essere cosi` abili, piu` tardi,
 da spiegare  perche' non e` accaduto."


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Mon, Sep 15, 2003 at 09:43:29PM +0200, Kronos wrote:
> @@ -1635,6 +1638,16 @@
>  	return err;
>  }
>  
> +static void release_cfb_info(struct fb_info *info) {
> +	struct cfb_info *cfb = info->par;
> +
> +	iounmap(cfb->region);
> +	fb_alloc_cmap(&info->cmap, 0, 0);
> +
> +	if (cfb->dev)
> +		pci_release_regions(cfb->dev);
> +}
> +
>  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
>  {
>  	struct cfb_info *cfb = pci_get_drvdata(dev);
> @@ -1645,12 +1658,10 @@
>  		 * we will be leaving hooks that could cause
>  		 * oopsen laying around.
>  		 */
> -		if (unregister_framebuffer(&cfb->fb))
> +		if (unregister_framebuffer(cfb->fb))
>  			printk(KERN_WARNING "%s: danger Will Robinson, "
>  				"danger danger!  Oopsen imminent!\n",
> -				cfb->fb.fix.id);
> -		iounmap(cfb->region);
> -		cyberpro_free_fb_info(cfb);
> +				cfb->fb->fix.id);
>  
>  		/*
>  		 * Ensure that the driver data is no longer
> @@ -1659,8 +1670,6 @@
>  		pci_set_drvdata(dev, NULL);
>  		if (cfb == int_cfb_info)
>  			int_cfb_info = NULL;
> -
> -		pci_release_regions(dev);
>  	}
>  }

There is another reason why the above is fundamentally flawed - who says
that "release_cfb_info" will still be in module space by the time you
need to call it?

Eg, you unload your framebuffer driver module immediately after the device
has gone away, but someone is keeping the sysfs files associated with
the fb_info open.  They close them, you call the release function - oops.

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Mon, Sep 15, 2003 at 10:40:42PM +0100, Russell King ha scritto: 
> > +static void release_cfb_info(struct fb_info *info) {
> > +	struct cfb_info *cfb = info->par;
> > +
> > +	iounmap(cfb->region);
> > +	fb_alloc_cmap(&info->cmap, 0, 0);
> > +
> > +	if (cfb->dev)
> > +		pci_release_regions(cfb->dev);
> > +}
> > +
> >  static void __devexit cyberpro_pci_remove(struct pci_dev *dev)
> >  {
> >  	struct cfb_info *cfb = pci_get_drvdata(dev);
> > @@ -1645,12 +1658,10 @@
> >  		 * we will be leaving hooks that could cause
> >  		 * oopsen laying around.
> >  		 */
> > -		if (unregister_framebuffer(&cfb->fb))
> > +		if (unregister_framebuffer(cfb->fb))
> >  			printk(KERN_WARNING "%s: danger Will Robinson, "
> >  				"danger danger!  Oopsen imminent!\n",
> > -				cfb->fb.fix.id);
> > -		iounmap(cfb->region);
> > -		cyberpro_free_fb_info(cfb);
> > +				cfb->fb->fix.id);
> >  
> >  		/*
> >  		 * Ensure that the driver data is no longer
> > @@ -1659,8 +1670,6 @@
> >  		pci_set_drvdata(dev, NULL);
> >  		if (cfb == int_cfb_info)
> >  			int_cfb_info = NULL;
> > -
> > -		pci_release_regions(dev);
> >  	}
> >  }
> 
> There is another reason why the above is fundamentally flawed - who says
> that "release_cfb_info" will still be in module space by the time you
> need to call it?
> 
> Eg, you unload your framebuffer driver module immediately after the device
> has gone away, but someone is keeping the sysfs files associated with
> the fb_info open.

I'm quite sure that this can't happen. If someone is keeping a sysfs
file open module use count won't be zero. Right?

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
Alcuni pensano che io sia una persona orribile, ma non vero. Ho il
cuore di un ragazzino - in un vaso sulla scrivania.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Tue, Sep 16, 2003 at 12:17:42AM +0200, Kronos wrote:
> Il Mon, Sep 15, 2003 at 10:40:42PM +0100, Russell King ha scritto: 
> > There is another reason why the above is fundamentally flawed - who says
> > that "release_cfb_info" will still be in module space by the time you
> > need to call it?
> > 
> > Eg, you unload your framebuffer driver module immediately after the device
> > has gone away, but someone is keeping the sysfs files associated with
> > the fb_info open.
> 
> I'm quite sure that this can't happen. If someone is keeping a sysfs
> file open module use count won't be zero. Right?

Where are you handling the module use count of the framebuffer driver?

I'm also confused why you want to keep the device IO regions around
until all sysfs files have been closed.  I hope you're not thinking
of touching the hardware after you've returned from the drivers
->remove method?

Is there somewhere I can view the core changes?

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Mon, Sep 15, 2003 at 11:58:32PM +0100, Russell King ha scritto: 
> On Tue, Sep 16, 2003 at 12:17:42AM +0200, Kronos wrote:
> > Il Mon, Sep 15, 2003 at 10:40:42PM +0100, Russell King ha scritto: 
> > > There is another reason why the above is fundamentally flawed - who says
> > > that "release_cfb_info" will still be in module space by the time you
> > > need to call it?
> > > 
> > > Eg, you unload your framebuffer driver module immediately after the device
> > > has gone away, but someone is keeping the sysfs files associated with
> > > the fb_info open.
> > 
> > I'm quite sure that this can't happen. If someone is keeping a sysfs
> > file open module use count won't be zero. Right?
> 
> Where are you handling the module use count of the framebuffer driver?

Every   attribute   (ie.  sysfs   file)   has   a  .owner   field   (see
include/linux/sysfs.h), the module  use count is handled  by sysfs. If a
sysfs file is open the module can't be unloaded.

> I'm also confused why you want to keep the device IO regions around
> until all sysfs files have been closed.  I hope you're not thinking
> of touching the hardware after you've returned from the drivers
> ->remove method?

I tought  that some  sysfs file  may use those  things. I was  sure that
class_dev hold a  reference to pci_dev.dev, but that's  not the case. So
yes, that's a bad idea. I'll move pci stuff in the unregister function.

> Is there somewhere I can view the core changes?

They are on fbdev for review, I put a copy here:
http://web.tiscali.it/kronoz/linux/fbdev-class_dev-fbmem.c.diff

The patch is agains James Simmons tree.

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
La vispa candela bruciava l'erbetta
creando un'essenza alquanto sospetta;
sembrava l'odore di una sacrestia
ma presto mi accorsi che era maria.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Tue, Sep 16, 2003 at 03:40:09PM +0200, Kronos wrote:
> Il Mon, Sep 15, 2003 at 11:58:32PM +0100, Russell King ha scritto: 
> > On Tue, Sep 16, 2003 at 12:17:42AM +0200, Kronos wrote:
> > > Il Mon, Sep 15, 2003 at 10:40:42PM +0100, Russell King ha scritto: 
> > > > There is another reason why the above is fundamentally flawed - who says
> > > > that "release_cfb_info" will still be in module space by the time you
> > > > need to call it?
> > > > 
> > > > Eg, you unload your framebuffer driver module immediately after the device
> > > > has gone away, but someone is keeping the sysfs files associated with
> > > > the fb_info open.
> > > 
> > > I'm quite sure that this can't happen. If someone is keeping a sysfs
> > > file open module use count won't be zero. Right?
> > 
> > Where are you handling the module use count of the framebuffer driver?
> 
> Every   attribute   (ie.  sysfs   file)   has   a  .owner   field   (see
> include/linux/sysfs.h), the module  use count is handled  by sysfs. If a
> sysfs file is open the module can't be unloaded.

The .owner field is set to the module which defined the sysfs attribute.
This wouldn't be the driver, but would be some other part of the framebuffer
layer.

The first instance I see in the patch at the URL below (fbmem.c):

 static CLASS_DEVICE_ATTR(dev, S_IRUGO, show_dev, NULL);

The owner field would be the module containing fbmem.o, which is separate
from the driver - this doesn't protect the driver code from being unloaded,
and its the driver code which is called via fb_info->release() when this
file is eventually closed.

> > I'm also confused why you want to keep the device IO regions around
> > until all sysfs files have been closed.  I hope you're not thinking
> > of touching the hardware after you've returned from the drivers
> > ->remove method?
> 
> I tought  that some  sysfs file  may use those  things. I was  sure that
> class_dev hold a  reference to pci_dev.dev, but that's  not the case. So
> yes, that's a bad idea. I'll move pci stuff in the unregister function.
> 
> > Is there somewhere I can view the core changes?
> 
> They are on fbdev for review, I put a copy here:
> http://web.tiscali.it/kronoz/linux/fbdev-class_dev-fbmem.c.diff

When I tried to get at the fbdev site, it seemed to be down.  Maybe
some of the kernel documentation needs to be updated?

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Tue, Sep 16, 2003 at 02:44:59PM +0100, Russell King ha scritto: 
> On Tue, Sep 16, 2003 at 03:40:09PM +0200, Kronos wrote:
> > Il Mon, Sep 15, 2003 at 11:58:32PM +0100, Russell King ha scritto: 
> > > On Tue, Sep 16, 2003 at 12:17:42AM +0200, Kronos wrote:
> > > > Il Mon, Sep 15, 2003 at 10:40:42PM +0100, Russell King ha scritto: 
> > > > > There is another reason why the above is fundamentally flawed - who says
> > > > > that "release_cfb_info" will still be in module space by the time you
> > > > > need to call it?
> > > > > 
> > > > > Eg, you unload your framebuffer driver module immediately after the device
> > > > > has gone away, but someone is keeping the sysfs files associated with
> > > > > the fb_info open.
> > > > 
> > > > I'm quite sure that this can't happen. If someone is keeping a sysfs
> > > > file open module use count won't be zero. Right?
> > > 
> > > Where are you handling the module use count of the framebuffer driver?
> > 
> > Every   attribute   (ie.  sysfs   file)   has   a  .owner   field   (see
> > include/linux/sysfs.h), the module  use count is handled  by sysfs. If a
> > sysfs file is open the module can't be unloaded.
> 
> The .owner field is set to the module which defined the sysfs attribute.
> This wouldn't be the driver, but would be some other part of the framebuffer
> layer.

Good point. I  was thinking  at files  created by  the driver  itself, I
forgot about  dev attribute. I  can overwrite  .owner of  dev attribute,
fbmem can't be built as module. Sounds good?

> > > Is there somewhere I can view the core changes?
> > 
> > They are on fbdev for review, I put a copy here:
> > http://web.tiscali.it/kronoz/linux/fbdev-class_dev-fbmem.c.diff
> 
> When I tried to get at the fbdev site, it seemed to be down.  Maybe
> some of the kernel documentation needs to be updated?

Try here: http://linux-fbdev.sf.net
Mailing lists are here: http://linux-fbdev.sourceforge.net/mlist.html

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
"It is more complicated than you think"
                -- The Eighth Networking Truth from RFC 1925


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Tue, Sep 16, 2003 at 04:17:13PM +0200, Kronos wrote:
> Good point. I  was thinking  at files  created by  the driver  itself, I
> forgot about  dev attribute. I  can overwrite  .owner of  dev attribute,
> fbmem can't be built as module. Sounds good?

How about we do the job properly first time around?

- fbmem.c is responsible for managing the lifetime of the fb_info structure,
  and the lifetime of the driver-private data is equal to the lifetime
  of the fb_info structure.

- framebuffer_alloc() creates the fb_info and the driver-private data
  in one object, as per your patch.
- fb_add_class_device creates the class device, as per your patch
  (don't you think it'd be a good idea to return the error to the
  driver if class_device_register() fails?)
- fb_remove_class_device unregisters the class device, as per your
  patch.  This should not cause the fb_info structures to be released.
- framebuffer_free() marks the structure available for freeing.
- at some point later, once all references have done, release_fb_info()
  is called.  This releases the memory space allocated by
  framebuffer_alloc().  This callback should _not_ call the driver
  code.

So far, this fits fairly nicely.  Now on to the driver.

- bus-specific ->probe function
  driver calls framebuffer_alloc(), initialises whatever it needs to.
  Driver then calls register_fb() which in turn eventually calls
  fb_add_class_device().

- bus-specific ->remove function
  driver calls unregister_framebuffer() which eventually calls
  fb_remove_class_device().  driver releases its resources, shuts down
  hardware, and generally cleans up.  The very last step that it does
  is call framebuffer_free().

I believe there was an issue concerning EDID stuff with the above method.
If this requires data from the fb_info or another structure whose lifetime
is directly connected to fb_info, then there isn't a problem.  If it
requires something from the driver module, things get a little more
complicated, and there is a solution to this.  I don't see any instances
of this from looking at your fbdev-class_dev-all patch though.

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Kronos]

Il Tue, Sep 16, 2003 at 03:52:30PM +0100, Russell King ha scritto: 
> - framebuffer_alloc() creates the fb_info and the driver-private data
>   in one object, as per your patch.

> - framebuffer_free() marks the structure available for freeing.


So framebuffer_alloc calls class_device_get and framebuffer_free calls
class_device_put. Ok, I'll rework the patch.

Luca
-- 
Reply-To: kronos@kronoz.cjb.net
Home: http://kronoz.cjb.net
Porc i' mond che cio' sott i piedi!
V. Catozzo


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Tue, Sep 16, 2003 at 05:17:01PM +0200, Kronos wrote:
> Il Tue, Sep 16, 2003 at 03:52:30PM +0100, Russell King ha scritto: 
> > - framebuffer_alloc() creates the fb_info and the driver-private data
> >   in one object, as per your patch.
> 
> > - framebuffer_free() marks the structure available for freeing.
> 
> 
> So framebuffer_alloc calls class_device_get and framebuffer_free calls
> class_device_put. Ok, I'll rework the patch.

You could call class_device_del() instead of class_device_unregister()
from unregister_framebuffer(), and call class_device_put() in
framebuffer_free().

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[James Simmons]

> - fbmem.c is responsible for managing the lifetime of the fb_info structure,
>   and the lifetime of the driver-private data is equal to the lifetime
>   of the fb_info structure.

This is not always the case. For dual headed graphics cards you can have 
two struct fb_info and one driver-private data. If the fbdev driver
releases one framebuffer you still have one framebuffer left that is using 
the driver-private data.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[Russell King]

On Wed, Sep 17, 2003 at 08:37:29PM +0100, James Simmons wrote:
> 
> > - fbmem.c is responsible for managing the lifetime of the fb_info structure,
> >   and the lifetime of the driver-private data is equal to the lifetime
> >   of the fb_info structure.
> 
> This is not always the case. For dual headed graphics cards you can have 
> two struct fb_info and one driver-private data. If the fbdev driver
> releases one framebuffer you still have one framebuffer left that is using 
> the driver-private data.

If this is not true, then how can anyone even be thinking about applying
the sysfs / driver model to the framebuffer drivers.

Unless this sort of data lifetime rules are clearly and unabiguously
defined, there is no way that a sysfs / driver model "port" will ever
come anywhere near being stable.

You can't get around these problems with module usage counting - module
usage refcounting is solely about code lifetimes.  This is all about data
lifetimes - when is the data created, when can it be safely destroyed,
etc.

-- 
Russell King (rmk@arm.linux.org.uk)	http://www.arm.linux.org.uk/personal/
Linux kernel maintainer of:
  2.6 ARM Linux   - http://www.arm.linux.org.uk/
  2.6 PCMCIA      - http://pcmcia.arm.linux.org.uk/
  2.6 Serial core


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[James Simmons]

> > > - fbmem.c is responsible for managing the lifetime of the fb_info structure,
> > >   and the lifetime of the driver-private data is equal to the lifetime
> > >   of the fb_info structure.
> > 
> > This is not always the case. For dual headed graphics cards you can have 
> > two struct fb_info and one driver-private data. If the fbdev driver
> > releases one framebuffer you still have one framebuffer left that is using 
> > the driver-private data.
> 
> If this is not true, then how can anyone even be thinking about applying
> the sysfs / driver model to the framebuffer drivers.
> 
> Unless this sort of data lifetime rules are clearly and unabiguously
> defined, there is no way that a sysfs / driver model "port" will ever
> come anywhere near being stable.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [PATCH] cyber2000fb: New framebuffer_alloc API and class_dev changes

[James Simmons]

> > > - fbmem.c is responsible for managing the lifetime of the fb_info structure,
> > >   and the lifetime of the driver-private data is equal to the lifetime
> > >   of the fb_info structure.
> > 
> > This is not always the case. For dual headed graphics cards you can have 
> > two struct fb_info and one driver-private data. If the fbdev driver
> > releases one framebuffer you still have one framebuffer left that is using 
> > the driver-private data.
> 
> If this is not true, then how can anyone even be thinking about applying
> the sysfs / driver model to the framebuffer drivers.
> 
> Unless this sort of data lifetime rules are clearly and unabiguously
> defined, there is no way that a sysfs / driver model "port" will ever
> come anywhere near being stable.

Sorry about the earlier email sent too early. 

   I'm attempting to look long term. We will have graphics cards with 
built in video as well. Plus 3D stuff. So my attempt was to create a driver 
core that was independent of any api. Then we would have mappers to 
different api's (fbdev/v4l etc). Thus different api sturctures could share 
the same core.  
   As for sysfs. I really don't see the need to export a par to userspace. 
In fact it could be dangerous to mess with. Now fb_info is important to 
export to userspace.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf