Fwd: [PATCH] ATI Mach64 accelerated imgblit (sligthly improved)

Fwd: [PATCH] ATI Mach64 accelerated imgblit (sligthly improved)

[Alexander Kern]

[-- Attachment #1: Type: text/plain, Size: 66 bytes --]

Sorry, forgot to say it's against 2.6.1-mm1 + last fbdev.diff.gz.

[-- Attachment #2: Alexander Kern <alex.kern@gmx.de>: [PATCH] ATI Mach64 accelerated imgblit (sligthly improved) --]
[-- Type: message/rfc822, Size: 11848 bytes --]

[-- Attachment #2.1.1: Type: text/plain, Size: 186 bytes --]

Hi,

here is a patch. Now it works with SUN12x22 font and 8, 16 and 32 bpp. I have 
the strange issues with 24bpp. Should we simply drop acceleration for this 
resolution?

Regards Alex

[-- Attachment #2.1.2: accel_imgblit.patch --]
[-- Type: text/x-diff, Size: 10799 bytes --]

diff -r -u -X /data/patches/exclude /usr/src/linux-2.6.orig/drivers/video/aty/atyfb_base.c /usr/src/linux-2.6.dev/drivers/video/aty/atyfb_base.c
--- /usr/src/linux-2.6.orig/drivers/video/aty/atyfb_base.c	2004-01-17 00:48:16.000000000 +0100
+++ /usr/src/linux-2.6.dev/drivers/video/aty/atyfb_base.c	2004-01-13 17:38:03.000000000 +0100
@@ -1134,11 +1130,11 @@
 		v_total = v_sync_end + debug.upper_margin;
 
 		hSync = 1000000000 / (pixclock_in_ps * h_total);
-		vRefresh = (hSync * 1000.0) / v_total;
+		vRefresh = (hSync * 1000) / v_total;
         	if (par->crtc.gen_cntl & CRTC_INTERLACE_EN)
-            	vRefresh *= 2.0;
+            	vRefresh *= 2;
         	if (par->crtc.gen_cntl & CRTC_DBL_SCAN_EN)
-            	vRefresh /= 2.0;
+            	vRefresh /= 2;
 
 		printk("atyfb: atyfb_set_par\n");
 		printk(" Set Visible Mode to %ix%i-%i\n", var->xres, var->yres, var->bits_per_pixel);
diff -r -u -X /data/patches/exclude /usr/src/linux-2.6.orig/drivers/video/aty/mach64_accel.c /usr/src/linux-2.6.dev/drivers/video/aty/mach64_accel.c
--- /usr/src/linux-2.6.orig/drivers/video/aty/mach64_accel.c	2004-01-09 07:59:04.000000000 +0100
+++ /usr/src/linux-2.6.dev/drivers/video/aty/mach64_accel.c	2004-01-17 00:23:23.000000000 +0100
@@ -12,23 +12,42 @@
     /*
      *  Generic Mach64 routines
      */
+     
+/* this is for DMA GUI engine! work to be continue */
+typedef struct {
+	u32 frame_buf_offset;
+	u32 system_mem_addr;
+	u32 command;
+	u32 reserved;
+} BM_DESCRIPTOR_ENTRY;
+
+#define LAST_DESCRIPTOR (1 << 31)
+#define SYSTEM_TO_FRAME_BUFFER 0
+
+static u32 rotation24bpp(u32 dx, u32 direction)
+{
+	u32 rotation;
+	if (direction & DST_X_LEFT_TO_RIGHT) {
+		rotation = (dx / 4) % 6;
+	} else {
+		rotation = ((dx + 2) / 4) % 6;
+	}
+
+	return ((rotation << 8) | DST_24_ROTATION_ENABLE);
+}
 
 void aty_reset_engine(const struct atyfb_par *par)
 {
 	/* reset engine */
 	aty_st_le32(GEN_TEST_CNTL,
-		    aty_ld_le32(GEN_TEST_CNTL, par) & ~GUI_ENGINE_ENABLE,
-		    par);
+		aty_ld_le32(GEN_TEST_CNTL, par) & ~GUI_ENGINE_ENABLE, par);
 	/* enable engine */
 	aty_st_le32(GEN_TEST_CNTL,
-		    aty_ld_le32(GEN_TEST_CNTL, par) | GUI_ENGINE_ENABLE,
-		    par);
+		aty_ld_le32(GEN_TEST_CNTL, par) | GUI_ENGINE_ENABLE, par);
 	/* ensure engine is not locked up by clearing any FIFO or */
 	/* HOST errors */
 	aty_st_le32(BUS_CNTL,
-		    aty_ld_le32(BUS_CNTL,
-				par) | BUS_HOST_ERR_ACK | BUS_FIFO_ERR_ACK,
-		    par);
+		aty_ld_le32(BUS_CNTL, par) | BUS_HOST_ERR_ACK | BUS_FIFO_ERR_ACK, par);
 }
 
 static void reset_GTC_3D_engine(const struct atyfb_par *par)
@@ -51,7 +70,7 @@
 	if (info->var.bits_per_pixel == 24) {
 		/* In 24 bpp, the engine is in 8 bpp - this requires that all */
 		/* horizontal coordinates and widths must be adjusted */
-		pitch_value = pitch_value * 3;
+		pitch_value *= 3;
 	}
 
 	/* On GTC (RagePro), we need to reset the 3D engine before */
@@ -146,7 +165,7 @@
 	aty_st_le32(DP_CHAIN_MASK, par->crtc.dp_chain_mask, par);
 
 	wait_for_fifo(5, par);
-	aty_st_le32(SCALE_3D_CNTL, 0, par);
+ 	aty_st_le32(SCALE_3D_CNTL, 0, par);
 	aty_st_le32(Z_CNTL, 0, par);
 	aty_st_le32(CRTC_INT_CNTL, aty_ld_le32(CRTC_INT_CNTL, par) & ~0x20,
 		    par);
@@ -174,8 +193,7 @@
 {
 	struct atyfb_par *par = (struct atyfb_par *) info->par;
 	u32 dy = area->dy, sy = area->sy, direction = DST_LAST_PEL;
-	u32 sx = area->sx, dx = area->dx, width = area->width;	
-	u32 pitch_value;
+	u32 sx = area->sx, dx = area->dx, width = area->width, rotation = 0;
 
 	if (!area->width || !area->height)
 		return;
@@ -186,11 +204,9 @@
 		return;
 	}
 
-	pitch_value = info->var.xres_virtual;
 	if (info->var.bits_per_pixel == 24) {
 		/* In 24 bpp, the engine is in 8 bpp - this requires that all */
 		/* horizontal coordinates and widths must be adjusted */
-		pitch_value *= 3;
 		sx *= 3;
 		dx *= 3;
 		width *= 3;
@@ -208,18 +224,22 @@
 	} else
 		direction |= DST_X_LEFT_TO_RIGHT;
 
+	if (info->var.bits_per_pixel == 24) {
+		rotation = rotation24bpp(dx, direction);
+	}
+
 	wait_for_fifo(4, par);
 	aty_st_le32(DP_SRC, FRGD_SRC_BLIT, par);
 	aty_st_le32(SRC_Y_X, (sx << 16) | sy, par);
 	aty_st_le32(SRC_HEIGHT1_WIDTH1, (width << 16) | area->height, par);
-	aty_st_le32(DST_CNTL, direction, par);
+	aty_st_le32(DST_CNTL, direction | rotation, par);
 	draw_rect(dx, dy, width, area->height, par);
 }
 
 void atyfb_fillrect(struct fb_info *info, const struct fb_fillrect *rect)
 {
 	struct atyfb_par *par = (struct atyfb_par *) info->par;
-	u32 color = rect->color, dx = rect->dx, width = rect->width;
+	u32 color = rect->color, dx = rect->dx, width = rect->width, rotation = 0;
 
 	if (!rect->width || !rect->height)
 		return;
@@ -238,6 +258,7 @@
 		/* horizontal coordinates and widths must be adjusted */
 		dx *= 3;
 		width *= 3;
+		rotation = rotation24bpp(dx, DST_X_LEFT_TO_RIGHT);
 	}
 
 	wait_for_fifo(3, par);
@@ -247,15 +268,160 @@
 		    par);
 	aty_st_le32(DST_CNTL,
 		    DST_LAST_PEL | DST_Y_TOP_TO_BOTTOM |
-		    DST_X_LEFT_TO_RIGHT, par);
+		    DST_X_LEFT_TO_RIGHT | rotation, par);
 	draw_rect(dx, rect->dy, width, rect->height, par);
 }
 
 void atyfb_imageblit(struct fb_info *info, const struct fb_image *image)
 {
 	struct atyfb_par *par = (struct atyfb_par *) info->par;
-    
-	if (par->blitter_may_be_busy)
-		wait_for_idle(par);
-	cfb_imageblit(info, image);
+	u32 dwords, dx = image->dx, dy = image->dy, width = image->width;
+	u32 pix_width_save, pix_width, host_cntl, rotation = 0, src, mix;
+
+	if (!image->width || !image->height)
+		return;
+	if (!par->accel_flags ||
+	    (image->depth != 1 && info->var.bits_per_pixel != image->depth)) {
+		if (par->blitter_may_be_busy)
+			wait_for_idle(par);
+
+		cfb_imageblit(info, image);
+		return;
+	}
+
+	wait_for_idle(par);
+	pix_width = pix_width_save = aty_ld_le32(DP_PIX_WIDTH, par);
+	host_cntl = aty_ld_le32(HOST_CNTL, par) | HOST_BYTE_ALIGN;
+
+	switch (image->depth) {
+	case 1:
+	    pix_width &= ~(BYTE_ORDER_MASK | HOST_MASK);
+	    pix_width |= (BYTE_ORDER_MSB_TO_LSB | HOST_1BPP);
+	    break;
+	case 4:
+	    pix_width &= ~(BYTE_ORDER_MASK | HOST_MASK);
+	    pix_width |= (BYTE_ORDER_MSB_TO_LSB | HOST_4BPP);
+	    break;
+	case 8:
+	    pix_width &= ~HOST_MASK;
+	    pix_width |= HOST_8BPP;
+	    break;
+	case 15:
+	    pix_width &= ~HOST_MASK;
+	    pix_width |= HOST_15BPP;
+	    break;
+	case 16:
+	    pix_width &= ~HOST_MASK;
+	    pix_width |= HOST_16BPP;
+	    break;
+	case 24:
+	    pix_width &= ~HOST_MASK;
+	    pix_width |= HOST_24BPP;
+	    break;
+	case 32:
+	    pix_width &= ~HOST_MASK;
+	    pix_width |= HOST_32BPP;
+	    break;
+	}
+
+	if (info->var.bits_per_pixel == 24) {
+		/* In 24 bpp, the engine is in 8 bpp - this requires that all */
+		/* horizontal coordinates and widths must be adjusted */
+		dx *= 3;
+		width *= 3;
+
+		rotation = rotation24bpp(dx, DST_X_LEFT_TO_RIGHT);
+
+		pix_width &= ~DST_MASK;
+		pix_width |= DST_8BPP;
+
+		/* ab IIC we have DP_HOST_TRIPLE_EN bit */
+		/* FIXME: 24bpp + SUN12x22 corruption
+		if (image->width % 8 == 0) */
+			pix_width |= DP_HOST_TRIPLE_EN;
+	}
+
+	dwords = (((((image->width * image->depth) + 7) / 8) * image->height * image->depth)+ 3) / 4 ;
+
+	if(image->depth == 1) {
+		u32 fg, bg;
+		if (info->fix.visual == FB_VISUAL_TRUECOLOR ||
+		    info->fix.visual == FB_VISUAL_DIRECTCOLOR) {
+			fg = ((u32*)(info->pseudo_palette))[image->fg_color];
+			bg = ((u32*)(info->pseudo_palette))[image->bg_color];
+		} else {
+			fg = image->fg_color;
+			bg = image->bg_color;
+		}
+
+		wait_for_fifo(2, par);
+		aty_st_le32(DP_BKGD_CLR, bg, par);
+		aty_st_le32(DP_FRGD_CLR, fg, par);
+		src = MONO_SRC_HOST | FRGD_SRC_FRGD_CLR | BKGD_SRC_BKGD_CLR;
+		mix = FRGD_MIX_S | BKGD_MIX_S;
+	} else {
+		src = MONO_SRC_ONE | FRGD_SRC_HOST;
+		mix = FRGD_MIX_D_XOR_S | BKGD_MIX_D;
+	}
+
+	wait_for_fifo(6, par);
+	aty_st_le32(DP_WRITE_MASK, 0xFFFFFFFF, par);
+	aty_st_le32(DP_PIX_WIDTH, pix_width, par);
+	aty_st_le32(DP_MIX, mix, par);
+	aty_st_le32(DP_SRC, src, par);
+	aty_st_le32(HOST_CNTL, host_cntl, par);
+	aty_st_le32(DST_CNTL, DST_Y_TOP_TO_BOTTOM | DST_X_LEFT_TO_RIGHT | rotation, par);
+
+	draw_rect(dx, dy, width, image->height, par);
+
+	/* manual triple each pixel, freezes the box */
+	if (0 && info->var.bits_per_pixel == 24 && !(pix_width & DP_HOST_TRIPLE_EN)) {
+		int inbit, outbit, mult24, byte_id_in_dword, width;
+		u8 *pbitmapin, *pbitmapout;
+		u32 hostdword;
+
+		pbitmapin = (u8*)image->data;
+
+		for(width = image->width, inbit = 7, mult24 = 0; dwords; dwords--) {
+			for(hostdword = 0, pbitmapout = (u8*)&hostdword, byte_id_in_dword = 0;
+				byte_id_in_dword < 4;
+				byte_id_in_dword++, pbitmapout++) {
+				for(outbit = 7; outbit >= 0; outbit--) {
+					*pbitmapout |= (((*pbitmapin >> inbit) & 1) << outbit);
+					mult24++;
+					/* next bit */
+					if (mult24 == 3) {
+						mult24 = 0;
+						inbit--;
+						width--;
+					}
+
+					/* next byte */
+					if (inbit < 0 || width == 0) {
+						pbitmapin++;
+						inbit = 7;
+
+						if(width == 0) {
+						    width = image->width;
+						    outbit = 0;
+						}
+					}
+				}
+			}
+			wait_for_fifo(1, par);
+			aty_st_le32(HOST_DATA0, hostdword, par);
+		}
+	} else {
+		u32 *pbitmap;
+		for(pbitmap = (u32*)(image->data); dwords; dwords--, pbitmap++) {
+			wait_for_fifo(1, par);
+			aty_st_le32(HOST_DATA0, *pbitmap, par);
+		}
+	}
+
+	wait_for_idle(par);
+
+	/* restore pix_width */
+	wait_for_fifo(1, par);
+	aty_st_le32(DP_PIX_WIDTH, pix_width_save, par);
 }
diff -r -u -X /data/patches/exclude /usr/src/linux-2.6.orig/include/video/mach64.h /usr/src/linux-2.6.dev/include/video/mach64.h
--- /usr/src/linux-2.6.orig/include/video/mach64.h	2004-01-17 00:48:16.000000000 +0100
+++ /usr/src/linux-2.6.dev/include/video/mach64.h	2004-01-15 00:27:29.000000000 +0100
@@ -983,13 +983,14 @@
 #define DP_CHAIN_32BPP		0x8080
 
 /* DP_PIX_WIDTH register constants */
-#define DST_1BPP		0
-#define DST_4BPP		1
-#define DST_8BPP		2
-#define DST_15BPP		3
-#define DST_16BPP		4
-#define DST_24BPP		5
-#define DST_32BPP		6
+#define DST_1BPP		0x0
+#define DST_4BPP		0x1
+#define DST_8BPP		0x2
+#define DST_15BPP		0x3
+#define DST_16BPP		0x4
+#define DST_24BPP		0x5
+#define DST_32BPP		0x6
+#define DST_MASK		0xF
 #define SRC_1BPP		0x000
 #define SRC_4BPP		0x100
 #define SRC_8BPP		0x200
@@ -997,6 +998,8 @@
 #define SRC_16BPP		0x400
 #define SRC_24BPP		0x500
 #define SRC_32BPP		0x600
+#define SRC_MASK		0xF00
+#define DP_HOST_TRIPLE_EN	0x2000
 #define HOST_1BPP		0x00000
 #define HOST_4BPP		0x10000
 #define HOST_8BPP		0x20000
@@ -1004,8 +1007,10 @@
 #define HOST_16BPP		0x40000
 #define HOST_24BPP		0x50000
 #define HOST_32BPP		0x60000
+#define HOST_MASK		0xF0000
 #define BYTE_ORDER_MSB_TO_LSB	0
 #define BYTE_ORDER_LSB_TO_MSB	0x1000000
+#define BYTE_ORDER_MASK		0x1000000
 
 /* DP_MIX register constants */
 #define BKGD_MIX_NOT_D			0

Confused about atyfb status.

[Richard Smith]

I'm trying to use the atyfb on a Rage Mobility M1 but I'm confused on 
which source tree has the best chance of getting to work.

Grepping through both the 2.4.24 and 2.6.1rc3 tree I find that the 
2.4.24 tree has several references to a mobility M1 in the comments.

The 2.6.1rc3 tree does not.

Neither tree however has an entry for my device.  Neither will detect my 
device which has an id of 0x4c52.

If I copy one of the "3D RAGE Mobility" id lines and change the clocks 
to what I think they should be the device gets detected but various 
specifications (like ammount and type of RAM) are incorrect.

So if I am about to start hacking on a tree to figure out whats wrong 
what tree should I use?  Where is the most advanced tree for the atyfb 
located?


-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Confused about atyfb status.

[Jon Smirl]

There is a fairly advanced one here: http://www.saftware.de/
but I don't know it's current status.

--- Richard Smith <rsmith@bitworks.com> wrote:
> I'm trying to use the atyfb on a Rage Mobility M1 but I'm confused on 
> which source tree has the best chance of getting to work.
> 
> Grepping through both the 2.4.24 and 2.6.1rc3 tree I find that the 
> 2.4.24 tree has several references to a mobility M1 in the comments.
> 
> The 2.6.1rc3 tree does not.
> 
> Neither tree however has an entry for my device.  Neither will detect my 
> device which has an id of 0x4c52.
> 
> If I copy one of the "3D RAGE Mobility" id lines and change the clocks 
> to what I think they should be the device gets detected but various 
> specifications (like ammount and type of RAM) are incorrect.
> 
> So if I am about to start hacking on a tree to figure out whats wrong 
> what tree should I use?  Where is the most advanced tree for the atyfb 
> located?
> 
> 
> -- 
> Richard A. Smith
> rsmith@bitworks.com
> 
> 
> 
> 
> -------------------------------------------------------
> The SF.Net email is sponsored by EclipseCon 2004
> Premiere Conference on Open Tools Development and Integration
> See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> http://www.eclipsecon.org/osdn
> _______________________________________________
> Linux-fbdev-devel mailing list
> Linux-fbdev-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-fbdev-devel


=====
Jon Smirl
jonsmirl@yahoo.com

__________________________________
Do you Yahoo!?
Yahoo! Hotjobs: Enter the "Signing Bonus" Sweepstakes
http://hotjobs.sweepstakes.yahoo.com/signingbonus


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Confused about atyfb status.

[Richard Smith]

Jon Smirl wrote:

> There is a fairly advanced one here: http://www.saftware.de/
> but I don't know it's current status.

Thanks.  This is a rage128 fb though and I was under the impression that 
the Rage Mobility M1 is mach64 based.  None of the device ids in this 
driver match either and no mention of anything "Mobility M1"

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Confused about atyfb status.

[Alexander Kern]

Am Samstag, 17. Januar 2004 07:38 schrieb Jon Smirl:
> There is a fairly advanced one here: http://www.saftware.de/
> but I don't know it's current status.
>
> --- Richard Smith <rsmith@bitworks.com> wrote:
> > I'm trying to use the atyfb on a Rage Mobility M1 but I'm confused on
> > which source tree has the best chance of getting to work.
> >
> > Grepping through both the 2.4.24 and 2.6.1rc3 tree I find that the
> > 2.4.24 tree has several references to a mobility M1 in the comments.
> >
> > The 2.6.1rc3 tree does not.
> >
> > Neither tree however has an entry for my device.  Neither will detect my
> > device which has an id of 0x4c52.
> >
> > If I copy one of the "3D RAGE Mobility" id lines and change the clocks
> > to what I think they should be the device gets detected but various
> > specifications (like ammount and type of RAM) are incorrect.
> >
> > So if I am about to start hacking on a tree to figure out whats wrong
> > what tree should I use?  Where is the most advanced tree for the atyfb
> > located?
> >
> >
> > --
> > Richard A. Smith
> > rsmith@bitworks.com

Mach64 MOBILITY M1 works by me for ages. I submitted all patches to James for 
months. They are in fbdev.diff.gz. But I dont think, they will be merged with 
vanilla very soon.

Regards
Alex Kern



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Confused about atyfb status.

[Geert Uytterhoeven]

On Fri, 16 Jan 2004, Richard Smith wrote:
> I'm trying to use the atyfb on a Rage Mobility M1 but I'm confused on
> which source tree has the best chance of getting to work.
>
> Grepping through both the 2.4.24 and 2.6.1rc3 tree I find that the
> 2.4.24 tree has several references to a mobility M1 in the comments.
>
> The 2.6.1rc3 tree does not.
>
> Neither tree however has an entry for my device.  Neither will detect my
> device which has an id of 0x4c52.
>
> If I copy one of the "3D RAGE Mobility" id lines and change the clocks
> to what I think they should be the device gets detected but various
> specifications (like ammount and type of RAM) are incorrect.
>
> So if I am about to start hacking on a tree to figure out whats wrong
> what tree should I use?  Where is the most advanced tree for the atyfb
> located?

2.4.24 has the Mobility M1 support, 2.6 hasn't.

BTW, I already forwarded your previous message to Daniël Mantione (who did the
M1 support), in case he missed your message on the list. Let's hope he knows
what's wrong.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Confused about atyfb status.

[Benjamin Herrenschmidt]

> 
> Mach64 MOBILITY M1 works by me for ages. I submitted all patches to James for 
> months. They are in fbdev.diff.gz. But I dont think, they will be merged with 
> vanilla very soon.

James stuffs will be merged soon, I hope. At this point, I need to
go over the lots of patches and break them in pieces suitable for
submission upstream (we can't get the whole lot merged at once, or
maybe we can but only after I have reviewed all of the stuff in
more details).

I don't know if it will make it in time for 2.6.2, but it will be
merged.

Anyway, I'm back from LCA now and plan to work on this next week.

Ben.




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

still ..Confused about atyfb and M1

[Richard Smith]

Benjamin Herrenschmidt wrote:

>>Mach64 MOBILITY M1 works by me for ages. I submitted all patches to James for 
>>months. They are in fbdev.diff.gz. But I dont think, they will be merged with 
>>vanilla very soon.
> 

   I grabbed fbdev.diff.gz from infradead (is this the canonical 
source?) and patched my 2.6.1 tree.  But I still don't see my 0x4c52 
device id listed in the tables.

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Richard Smith wrote:

> Benjamin Herrenschmidt wrote:
> 
>>> Mach64 MOBILITY M1 works by me for ages. I submitted all patches to 
>>> James for months. They are in fbdev.diff.gz. But I dont think, they 
>>> will be merged with vanilla very soon.
>>
>>
>   I grabbed fbdev.diff.gz from infradead (is this the canonical source?) 
> and patched my 2.6.1 tree.  But I still don't see my 0x4c52 device id 
> listed in the tables.
> 

And it still dosen't detect my device(s).

I never received an answer to this.  I don't see how the M1 can work 
supported when the device id isn't in the table.  Did I really get the 
latest patch?

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[James Simmons]

Can you submit a patch.


On Tue, 27 Jan 2004, Richard Smith wrote:

> Richard Smith wrote:
> 
> > Benjamin Herrenschmidt wrote:
> > 
> >>> Mach64 MOBILITY M1 works by me for ages. I submitted all patches to 
> >>> James for months. They are in fbdev.diff.gz. But I dont think, they 
> >>> will be merged with vanilla very soon.
> >>
> >>
> >   I grabbed fbdev.diff.gz from infradead (is this the canonical source?) 
> > and patched my 2.6.1 tree.  But I still don't see my 0x4c52 device id 
> > listed in the tables.
> > 
> 
> And it still dosen't detect my device(s).
> 
> I never received an answer to this.  I don't see how the M1 can work 
> supported when the device id isn't in the table.  Did I really get the 
> latest patch?
> 
> 



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Richard Smith]

James Simmons wrote:

> Can you submit a patch.
> 

Sure but the problem is that it dosen't work.  All I did was duplicate 
the 0x4c4d device (m64n_mob_p) and change the device Id to 0x4c52. 
It detects my devices but the type and  ammount of ram it finds and 
various other things are incorrect. The 0x4c4d matches the M1s clock 
speeds but I don't know what all those other flags specified mean and 
the text description for the m64_mob_p says "3D Rage Mobility.... P/M 
AGP 2x" and my device is on the PCI bus.  The Marketing descriptions 
describe the M1 as a "Rage Mobility" not a "3D Rage Mobility"

I suppose now would be the time to own up and mention that the above 
incorrect dection problems may not be the drivers fault at all.  I have 
a really special setup. (explained below) Howerver, since my device ID 
wasn't in the device table I don't know if the driver is _supposed_ to 
be able to get the right info.

Ok. Here's my crazy setup.  I have custom SBC that we re-designed with 2 
ATI M1's on it to replace 2 Assilliant 69030s which went end of life 
last year.  Both chips are on the PCI bus.

This board boots linuxBIOS rather than a normal BIOS.  LinuxBIOS has the 
ability to run videobios provided by the mfg but it only has limited 
bios int support.  The assilliant videobios must have had much fewer 
requirements than ATI's bios because it all worked fine and dandy.

The ATI videobios runs and will enable VSYNC but I don't ever get any 
video output.  So I suspect that there are a few int calls that it needs 
that LinuxBIOS dosen't have support for and its not setting some 
registers correctly or perhaps not at all.

lspci can read the pci config space and all the info looks good so right 
now I don't have any reason to believe that the hardware isn't hooked up 
to the pci bus right.

My hope was that the chip was up far enough that the framebuffer driver 
code would bring it to life.  But that didn't happen so thats why I 
wan't to know what the driver is supposed to be able to do.  So I have 
new untested hardware booting a half-ass videobios running an unknown 
driver.  Gee... wonder why it dosen't work? *grin*

And no.  I don't have a known good system as you can't buy video cards 
with an M1 on them. :( (At least I haven't found one yet)  Anybody know 
what laptop brands/models have an M1 in them?  That would be handy.

If someone can tell me what's supposed to work on a sane setup then I 
can begin to figure out why that dosen't happen on my setup.

I should also mention that in order to use these chips Bitworks had to 
enter into a NDA with ATI.  I have access to all the propritary info on 
the registers and so forth.

So I have to be careful with any patches I submit as ATI hasn't been 
very helpfull to me with what info I can and can't release.  I don't 
want to inadvertently release any NDA info and get us all into trouble. 
  (Anyone with experience in this area please feel free to contact me)

Obviously if the M1 is already supported by atyfb then ATI authorized 
that info to be released already and I don't have to worry quite so much 
about submitting patches to that code.

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Somehow this didn't make it to the list the first time... so I'm 
resending it.

James Simmons wrote:

 > Can you submit a patch.
 >

Sure but the problem is that it dosen't work.  All I did was duplicate 
the 0x4c4d device (m64n_mob_p) and change the device Id to 0x4c52. It 
detects my devices but the type and  ammount of ram it finds and various 
other things are incorrect. The 0x4c4d matches the M1s clock speeds but 
I don't know what all those other flags specified mean and the text 
description for the m64_mob_p says "3D Rage Mobility.... P/M AGP 2x" and 
my device is on the PCI bus.  The Marketing descriptions describe the M1 
as a "Rage Mobility" not a "3D Rage Mobility"

I suppose now would be the time to own up and mention that the above 
incorrect dection problems may not be the drivers fault at all.  I have 
a really special setup. (explained below) Howerver, since my device ID 
wasn't in the device table I don't know if the driver is _supposed_ to 
be able to get the right info.

Ok. Here's my crazy setup.  I have custom SBC that we re-designed with 2 
ATI M1's on it to replace 2 Assilliant 69030s which went end of life 
last year.  Both chips are on the PCI bus.

This board boots linuxBIOS rather than a normal BIOS.  LinuxBIOS has the 
ability to run videobios provided by the mfg but it only has limited 
bios int support.  The assilliant videobios must have had much fewer 
requirements than ATI's bios because it all worked fine and dandy.

The ATI videobios runs and will enable VSYNC but I don't ever get any 
video output.  So I suspect that there are a few int calls that it needs 
that LinuxBIOS dosen't have support for and its not setting some 
registers correctly or perhaps not at all.

lspci can read the pci config space and all the info looks good so right 
now I don't have any reason to believe that the hardware isn't hooked up 
to the pci bus right.

My hope was that the chip was up far enough that the framebuffer driver 
code would bring it to life.  But that didn't happen so thats why I 
wan't to know what the driver is supposed to be able to do.  So I have 
new untested hardware booting a half-ass videobios running an unknown 
driver.  Gee... wonder why it dosen't work? *grin*

And no.  I don't have a known good system as you can't buy video cards 
with an M1 on them. (At least I haven't found one yet)  Anybody know 
what laptop brands/models have an M1 in them?  That would be handy.

If someone can tell me what's supposed to work on a sane setup then I 
can begin to figure out why that dosen't happen on my setup.

I should also mention that in order to use these chips Bitworks had to 
enter into a NDA with ATI.  I have access to all the propritary info on 
the registers and so forth.

So I have to be careful with any patches I submit as ATI hasn't been 
very helpfull to me with what info I can and can't release.  I don't 
want to inadvertently release any NDA info and get us all into trouble. 
  (Anyone with experience in this area please feel free to contact me)

Obviously if the M1 is already supported by atyfb then ATI authorized 
that info to be released already and I don't have to worry quite so much 
about submitting patches to that code.


-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Tue, 3 Feb 2004, Richard Smith wrote:
> My hope was that the chip was up far enough that the framebuffer driver
> code would bring it to life.  But that didn't happen so thats why I
> wan't to know what the driver is supposed to be able to do.  So I have
> new untested hardware booting a half-ass videobios running an unknown
> driver.  Gee... wonder why it dosen't work? *grin*

Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
initialization code for RAGE XL, though.

> And no.  I don't have a known good system as you can't buy video cards
> with an M1 on them. :( (At least I haven't found one yet)  Anybody know
> what laptop brands/models have an M1 in them?  That would be handy.
>
> Obviously if the M1 is already supported by atyfb then ATI authorized
> that info to be released already and I don't have to worry quite so much
> about submitting patches to that code.

My Vaio Z600TEK (Z505 in US/JP) has a M1. It works with atyfb in recent 2.4,
but it took Daniël Mantione quite a bit of work to make it work on this
particular machine...

Let's say atyfb in 2.4 works on most M1s these days.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

> On Tue, 3 Feb 2004, Richard Smith wrote:
> 
>>new untested hardware booting a half-ass videobios running an unknown
>>driver.  Gee... wonder why it dosen't work? *grin*

 >Geert Uytterhoeven wrote:

> Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
> 
How much?  Are things like the ammount and type of RAM autodetected or 
just read from registers that the BIOS is supposed to set?

> 
 > initialization code for RAGE XL, though.

Any idea how different the RAGE XL is from the M1? I know they are both 
Mach64 but how consistent are the registers across all of the Mach64 chips?

> My Vaio Z600TEK (Z505 in US/JP) has a M1. It works with atyfb in recent 2.4,
> but it took Daniël Mantione quite a bit of work to make it work on this
> particular machine...

What device ID is your M1?

Any idea what he had to do?  Most of DM's code seems to deal with 
makeing it work properly with a LCD.  Right now I just have a CRT hooked 
up.  LCD will comes later.

> Let's say atyfb in 2.4 works on most M1s these days.

So which one should I work with? 2.4 or 2.6 with fbdiff?

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Richard Smith]

Geert Uytterhoeven wrote:

> Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
> initialization code for RAGE XL, though.

Does OpenFirmware support exist for the M1?

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Wed, 4 Feb 2004, Richard Smith wrote:
> > On Tue, 3 Feb 2004, Richard Smith wrote:
> >
> >>new untested hardware booting a half-ass videobios running an unknown
> >>driver.  Gee... wonder why it dosen't work? *grin*
>
>  >Geert Uytterhoeven wrote:
>
> > Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
> >
> How much?  Are things like the ammount and type of RAM autodetected or
> just read from registers that the BIOS is supposed to set?

Atyfb reads them from the registers.

> > My Vaio Z600TEK (Z505 in US/JP) has a M1. It works with atyfb in recent 2.4,
> > but it took Daniël Mantione quite a bit of work to make it work on this
> > particular machine...
>
> What device ID is your M1?

1002:4c4d

> Any idea what he had to do?  Most of DM's code seems to deal with

IIRC, programming an extra shadow register.

> makeing it work properly with a LCD.  Right now I just have a CRT hooked
> up.  LCD will comes later.
>
> > Let's say atyfb in 2.4 works on most M1s these days.
>
> So which one should I work with? 2.4 or 2.6 with fbdiff?

I think 2.4 is the safest. I don't know whether James' tree already contains
all Mobility support from 2.4.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Wed, 4 Feb 2004, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> > Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
> > initialization code for RAGE XL, though.
>
> Does OpenFirmware support exist for the M1?

I guess so, since the M1 is used in some iBooks, IIRC.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Benjamin Herrenschmidt]

On Thu, 2004-02-05 at 02:18, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> 
> > Indeed, atyfb needs BIOS (or OpenFirmware) initialization. It has
> > initialization code for RAGE XL, though.
> 
> Does OpenFirmware support exist for the M1?

Some iBooks ship with an M1, those have an OF driver

Ben.



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Geert Uytterhoeven wrote:

>>How much?  Are things like the ammount and type of RAM autodetected or
>>just read from registers that the BIOS is supposed to set?

> Atyfb reads them from the registers.

So this is probally why all the values are jacked up.  They probally 
aren't set correctly to begin with.  I'll try hardcodeing them in the 
driver with what I think they should be.

> 1002:4c4d

Wonder why your device ID is differnet than mine.  Can you send me a 
'lspci -vv' of your M1?

James:

If you read this from the list...  I get your mail but for some reason I 
can't send you mail.  I get:

   Attempting TCP connection to [phoenix.ipv6.infradead.org : 25]
   A-record resolution of [phoenix.ipv6.infradead.org] in progress...
   Name server has no records of the requested type for that domain.

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Thu, 5 Feb 2004, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> >>How much?  Are things like the ammount and type of RAM autodetected or
> >>just read from registers that the BIOS is supposed to set?
>
> > Atyfb reads them from the registers.
>
> So this is probally why all the values are jacked up.  They probally
> aren't set correctly to begin with.  I'll try hardcodeing them in the
> driver with what I think they should be.

You may have to program the memory control for the correct type of memory, too.

> > 1002:4c4d
>
> Wonder why your device ID is differnet than mine.  Can you send me a

Because mine is AGP and yours is PCI?

> 'lspci -vv' of your M1?

I can, but not right now. If the answer above doesn't satisfy you, just ask
again.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Geert Uytterhoeven wrote:

>>>1002:4c4d
>>
>>Wonder why your device ID is differnet than mine.  Can you send me a
>
> Because mine is AGP and yours is PCI?

Should not matter.  Chip has both interfaces on it and AGP is just PCI 
with some additional signals.

>>'lspci -vv' of your M1?
> I can, but not right now. If the answer above doesn't satisfy you, just ask
> again.
> 

No rush. Im messing with a different project right now anyway.  But yes 
I'd still like a -vv of your chip when you can.

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Benjamin Herrenschmidt]

On Fri, 2004-02-06 at 08:37, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> 
> >>>1002:4c4d
> >>
> >>Wonder why your device ID is differnet than mine.  Can you send me a
> >
> > Because mine is AGP and yours is PCI?
> 
> Should not matter.  Chip has both interfaces on it and AGP is just PCI 
> with some additional signals.

No, ATI used to have both versions of the chip with different IDs

> >>'lspci -vv' of your M1?
> > I can, but not right now. If the answer above doesn't satisfy you, just ask
> > again.
> > 
> 
> No rush. Im messing with a different project right now anyway.  But yes 
> I'd still like a -vv of your chip when you can.
-- 
Benjamin Herrenschmidt <benh@kernel.crashing.org>



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Benjamin Herrenschmidt wrote:

>>>>>1002:4c4d
>>>>
>>>>Wonder why your device ID is differnet than mine.  Can you send me a
>>>
>>>Because mine is AGP and yours is PCI?

> No, ATI used to have both versions of the chip with different IDs

Hmmmm.. Looking at the 2.4.24 aty code 0x4c4d uses the m64n_mob_p which 
is described as "3D Rage Mobility (PCI)".  0x4c4e is "3D Rage Mobility 
(AGP)"  Geert if your device is really an AGP device then the 
description in the driver is wrong.

Ah.. I see that in the 2.6 fbdiff both descriptions are now AGP.  Cool 
so that probally why my device id is different.  Is there anything 
different that you have to do to with using the chip between AGP and PCI 
then?

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Thu, 5 Feb 2004, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> >>>1002:4c4d
> >>
> >>Wonder why your device ID is differnet than mine.  Can you send me a
> >
> > Because mine is AGP and yours is PCI?
>
> Should not matter.  Chip has both interfaces on it and AGP is just PCI
> with some additional signals.
>
> >>'lspci -vv' of your M1?
> > I can, but not right now. If the answer above doesn't satisfy you, just ask
> > again.
>
> No rush. Im messing with a different project right now anyway.  But yes
> I'd still like a -vv of your chip when you can.

01:00.0 VGA compatible controller: ATI Technologies Inc Rage Mobility P/M AGP 2x (rev 64) (prog-if 00 [VGA])
        Subsystem: Sony Corporation: Unknown device 80c7
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B-
        Status: Cap+ 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
        Latency: 66 (2000ns min), Cache Line Size: 0x08 (32 bytes)
        Interrupt: pin A routed to IRQ 9
        Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
        Region 1: I/O ports at e800 [size=256]
        Region 2: Memory at febff000 (32-bit, non-prefetchable) [size=4K]
        Expansion ROM at <unassigned> [disabled] [size=128K]
        Capabilities: [50] AGP version 1.0
                Status: RQ=256 Iso- ArqSz=0 Cal=0 SBA+ ITACoh- GART64- HTrans- 64bit- FW- AGP3- Rate=x1,x2
                Command: RQ=1 ArqSz=0 Cal=0 SBA- AGP- GART64- 64bit- FW- Rate=<none>
        Capabilities: [5c] Power Management version 1
                Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0-,D1-,D2-,D3hot-,D3cold-)
                Status: D0 PME-Enable- DSel=0 DScale=0 PME-

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Geert Uytterhoeven wrote:

> 01:00.0 VGA compatible controller: ATI Technologies Inc Rage Mobility P/M AGP 2x (rev 64) (prog-if 00 [VGA])
>         Subsystem: Sony Corporation: Unknown device 80c7
>         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B-
>         Status: Cap+ 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
>         Latency: 66 (2000ns min), Cache Line Size: 0x08 (32 bytes)
>         Interrupt: pin A routed to IRQ 9
>         Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]

I don't think your chip is an M1.  The M1 is limited to 8 Megs of memory 
which is on chip and not expandable so I don't see how you can have 16Megs.




-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[James Simmons]

> James:
> 
> If you read this from the list...  I get your mail but for some reason I 
> can't send you mail.  I get:
> 
>    Attempting TCP connection to [phoenix.ipv6.infradead.org : 25]
>    A-record resolution of [phoenix.ipv6.infradead.org] in progress...
>    Name server has no records of the requested type for that domain.

Have no idea why. What email address are you sending it too?




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Richard Smith]

James Simmons wrote:

>>James:
>>
>>If you read this from the list...  I get your mail but for some reason I 
>>can't send you mail.  I get:
>>
>>   Attempting TCP connection to [phoenix.ipv6.infradead.org : 25]
>>   A-record resolution of [phoenix.ipv6.infradead.org] in progress...
>>   Name server has no records of the requested type for that domain.
> 
> 
> Have no idea why. What email address are you sending it too?

jsimmons@infradead.org.  Someone sent me mail showing that your primary 
MX resolves out to an IPV6 setup which our server dosen't know how to 
do.  And then for some reason it fails to use the next MX record.  So it 
looks like a problem with our MTA.

-- 
Richard A. Smith
rsmith@bitworks.com





-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Geert Uytterhoeven]

On Fri, 6 Feb 2004, Richard Smith wrote:
> Geert Uytterhoeven wrote:
> > 01:00.0 VGA compatible controller: ATI Technologies Inc Rage Mobility P/M AGP 2x (rev 64) (prog-if 00 [VGA])
> >         Subsystem: Sony Corporation: Unknown device 80c7
> >         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping+ SERR- FastB2B-
> >         Status: Cap+ 66Mhz- UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
> >         Latency: 66 (2000ns min), Cache Line Size: 0x08 (32 bytes)
> >         Interrupt: pin A routed to IRQ 9
> >         Region 0: Memory at fd000000 (32-bit, non-prefetchable) [size=16M]
>
> I don't think your chip is an M1.  The M1 is limited to 8 Megs of memory
> which is on chip and not expandable so I don't see how you can have 16Megs.

It has 8 MiB of RAM.

All Mach64 PCI/AGP variants show 16 MiB of PCI memory space, for up to 8 MiB of
video RAM. The second 8 MiB is the big endian aperture.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: Re: still ..Confused about atyfb and M1

[Alexander Kern]

Am Samstag, 7. Februar 2004 18:54 schrieb Geert Uytterhoeven:
> On Fri, 6 Feb 2004, Richard Smith wrote:
> > Geert Uytterhoeven wrote:
> > > 01:00.0 VGA compatible controller: ATI Technologies Inc Rage Mobility
> > > P/M AGP 2x (rev 64) (prog-if 00 [VGA]) Subsystem: Sony Corporation:
> > > Unknown device 80c7
> > >         Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop-
> > > ParErr- Stepping+ SERR- FastB2B- Status: Cap+ 66Mhz- UDF- FastB2B+
> > > ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR- Latency:
> > > 66 (2000ns min), Cache Line Size: 0x08 (32 bytes) Interrupt: pin A
> > > routed to IRQ 9
> > >         Region 0: Memory at fd000000 (32-bit, non-prefetchable)
> > > [size=16M]
> >
> > I don't think your chip is an M1.  The M1 is limited to 8 Megs of memory
> > which is on chip and not expandable so I don't see how you can have
> > 16Megs.
>
> It has 8 MiB of RAM.
>
> All Mach64 PCI/AGP variants show 16 MiB of PCI memory space, for up to 8
> MiB of video RAM. The second 8 MiB is the big endian aperture.
>
> Gr{oetje,eeting}s,
>
> 						Geert
Hallo all,

I'm not understand, what a problem is, who is problem reporter and why
he don't ask me?

I work on this driver for 2.6 kernel and it works good on my ARMADA E500
It's 15" 1400x1050 natural with
ATI Technologies Inc Rage Mobility P/M AGP 2x (rev 64)

good means all resolutions, from 320x200 to 1600x1200(that one only on CRT).

Regards Alex

P.S.
If everybody has a problem, please be sure, you use the last code from James's 
fbdev.diff.gz, than switch debug define in atyfb_base.c on and
send me a log.




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: still ..Confused about atyfb and M1

[Richard Smith]

Geert Uytterhoeven wrote:

 >>I don't think your chip is an M1.  The M1 is limited to 8 Megs of memory
 >>which is on chip and not expandable so I don't see how you can have 
16Megs.

 > It has 8 MiB of RAM.
 >
 > All Mach64 PCI/AGP variants show 16 MiB of PCI memory space, for up 
to 8 MiB of
 > video RAM. The second 8 MiB is the big endian aperture.

Ah my fbdev ignorance shows.  I messed with this lots this weekend and 
I'm pretty sure its a bios thing and not the driver.

In other news I gust got word today from ATI that I can release my code 
under GPL.  Woohoo!   I'm about to start porting over some dos code I 
have that is supposed to do a biosless init of the M1 for inclusion into 
LinuxBIOS.  I'll also try and add the same thing into atyfb when I have 
it all working.

Thanks for all your help.

-- 
Richard A. Smith
rsmith@bitworks.com




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn