Re: Added 915G to intelfb of 2.6.10-kernel-sources

Re: Added 915G to intelfb of 2.6.10-kernel-sources

[Sylvain Meyer]

        Hi Axel,

    Thanks for your patch but could you format it as described in this 
document  
http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt

and resend it.

The best is to use these tools to maintain and send kernel patches
http://www.zip.com.au/~akpm/linux/patches/

Also you should subscribe and send fbdev-related mails to 
linux-fbdev-devel@lists.sourceforge.net

Regards
Sylvain

Buttchereit, Axel (XLsigned) a écrit:

> Hi,
>
> though I'm quite sure that the attached patch (intelfb-...) is far 
> from complete
> I think that someone/you (guessing from the comments that you are 
> maintaining
> the code) might be interested in my work.
>
> Having no technical documentation (from Intel) I had to "develop" the 
> patch by
> looking at the xfree/xorg-code that already supports I915G and by using
> trial-and-error (My ThinkCentre's internal graphics device is an I915G
> and that made me start working on this subject).
>
> During development and testing I've discovered that the way "intelfb" 
> and maybe
> other framebuffer-drivers are using "modedb" does not work, if 
> compiled as a
> module. The initial "modedb entries" are lost (overwritten?) when 
> "fb_find_mode"
> is called from the module. Linking "intelfb" into the kernel works as 
> expected.
>
> ByTheWay: Why isn't the framebuffer module (intelfb) automatically 
> loaded, if
> "video=intelfb:..." is specified as kernel-parameter?
>
> I have attached an additional patch (modedb-...) that re-initialize a
> minimal-modedb (my_mini_modedb), if MY_PREFERRED_MODE (1024x768-32@73)
> is used in a call to "fb_find_mode", to show that loading "intelfb" as a
> module works in principle.
>
> Just for completeness I've attached a third (trivial) patch that adds 
> support for
> the external graphics adapter (radeon X600, RV380) of my ThinkCentre, 
> by treating
> it as an RV350.
>
> Sorry for sending direct email.
>
> --Axel
>
>
>
>
>
>
>

------------------------------------------------------------------------

diff -ruN linux/drivers/video/intelfb.orig/intelfbdrv.c 
linux/drivers/video/intelfb/intelfbdrv.c
--- linux/drivers/video/intelfb.orig/intelfbdrv.c    2005-01-26 
18:51:13.162054107 +0100
+++ linux/drivers/video/intelfb/intelfbdrv.c    2005-01-27 
23:37:21.273059171 +0100
@@ -1,7 +1,8 @@
 /*
  * intelfb
  *
- * Linux framebuffer driver for Intel(R) 830M/845G/852GM/855GM/865G
+ * *** XLsigned-2005-01-27: Added 915G ***
+ * Linux framebuffer driver for Intel(R) 830M/845G/852GM/855GM/865G/915G
  * integrated graphics chips.
  *
  * Copyright © 2002, 2003 David Dawes <dawes@xfree86.org>
@@ -153,6 +154,8 @@
     { PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_845G, PCI_ANY_ID, 
PCI_ANY_ID, PCI_CLASS_DISPLAY_VGA << 8, INTELFB_CLASS_MASK, INTEL_845G },
     { PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_85XGM, PCI_ANY_ID, 
PCI_ANY_ID, PCI_CLASS_DISPLAY_VGA << 8, INTELFB_CLASS_MASK, INTEL_85XGM },
     { PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_865G, PCI_ANY_ID, 
PCI_ANY_ID, PCI_CLASS_DISPLAY_VGA << 8, INTELFB_CLASS_MASK, INTEL_865G },
+    /* XLsigned-2005-01-27: Added 915G */
+    { PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_915G, PCI_ANY_ID, 
PCI_ANY_ID, PCI_CLASS_DISPLAY_VGA << 8, INTELFB_CLASS_MASK, INTEL_915G },
     { 0, }
 };
 
@@ -516,13 +519,25 @@
     }
 
     /* Set base addresses. */
-    dinfo->aperture.physical = pci_resource_start(pdev, 0);
-    dinfo->aperture.size     = pci_resource_len(pdev, 0);
-    dinfo->mmio_base_phys    = pci_resource_start(pdev, 1);
-
-    DBG_MSG("fb aperture: 0x%lx/0x%lx, MMIO region: 0x%lx/0x%lx\n",
-        pci_resource_start(pdev, 0), pci_resource_len(pdev, 0),
-        pci_resource_start(pdev, 1), pci_resource_len(pdev, 1));
+    /* XLsigned-2005-01-27: Added 915G */
+    if (ent->device == PCI_DEVICE_ID_INTEL_915G) {
+        DBG_MSG("XLsigned-Debug: IS_915G\n");
+        dinfo->aperture.physical = pci_resource_start(pdev, 2);
+        dinfo->aperture.size     = pci_resource_len(pdev, 2);
+        dinfo->mmio_base_phys    = pci_resource_start(pdev, 0);
+       
+        DBG_MSG("fb aperture: 0x%lx/0x%lx, MMIO region: 0x%lx/0x%lx\n",
+            pci_resource_start(pdev, 2), pci_resource_len(pdev, 2),
+            pci_resource_start(pdev, 0), pci_resource_len(pdev, 0));
+    } else {
+        dinfo->aperture.physical = pci_resource_start(pdev, 0);
+        dinfo->aperture.size     = pci_resource_len(pdev, 0);
+        dinfo->mmio_base_phys    = pci_resource_start(pdev, 1);
+
+        DBG_MSG("fb aperture: 0x%lx/0x%lx, MMIO region: 0x%lx/0x%lx\n",
+            pci_resource_start(pdev, 0), pci_resource_len(pdev, 0),
+            pci_resource_start(pdev, 1), pci_resource_len(pdev, 1));
+    }
 
     /* Reserve the fb and MMIO regions */
     if (!request_mem_region(dinfo->aperture.physical, dinfo->aperture.size,
@@ -756,6 +771,8 @@
         printk(").  Disabling mode switching.\n");
     }
 
+    DBG_MSG("XLsigned-Debug: Value of bailearly: %d\n",bailearly);
+   
     if (bailearly == 1)
         bailout(dinfo);
 
@@ -785,7 +802,7 @@
         bailout(dinfo);
 
 
-    if (intelfb_set_fbinfo(dinfo)) {
+     if (intelfb_set_fbinfo(dinfo)) {
         cleanup(dinfo);
         return -ENODEV;
     }
@@ -983,17 +1000,20 @@
 
     var = &dinfo->info->var;
     if (FIXED_MODE(dinfo)) {
+        DBG_MSG("XLsigned-Debug: Setting FIXED_MODE\n");
             memcpy(var, &dinfo->initial_var,
                sizeof(struct fb_var_screeninfo));
         msrc = 5;
     } else {
         if (mode) {
+            DBG_MSG("XLsignedi-Debug: finding given mode: %s\n",mode);
             msrc = fb_find_mode(var, dinfo->info, mode,
                         NULL, 0, NULL, 0);
             if (msrc)
                 msrc |= 8;
         }
         if (!msrc) {
+            DBG_MSG("XLsigned-Debug: finding PREFERRED_MODE: 
%s\n",PREFERRED_MODE);
             msrc = fb_find_mode(var, dinfo->info, PREFERRED_MODE,
                         NULL, 0, NULL, 0);
         }
diff -ruN linux/drivers/video/intelfb.orig/intelfbdrv.h 
linux/drivers/video/intelfb/intelfbdrv.h
--- linux/drivers/video/intelfb.orig/intelfbdrv.h    2005-01-26 
18:51:13.162054107 +0100
+++ linux/drivers/video/intelfb/intelfbdrv.h    2005-01-27 
13:23:55.811192427 +0100
@@ -5,7 +5,8 @@
  
******************************************************************************
  * intelfb
  *
- * Linux framebuffer driver for Intel(R) 830M/845G/852GM/855GM/865G
+ * *** XLsigned-2005-01-27: Added 915G ***
+ * Linux framebuffer driver for Intel(R) 830M/845G/852GM/855GM/865G/915G
  * integrated graphics chips.
  *
  * Copyright © 2004 Sylvain Meyer
diff -ruN linux/drivers/video/intelfb.orig/intelfb.h 
linux/drivers/video/intelfb/intelfb.h
--- linux/drivers/video/intelfb.orig/intelfb.h    2005-01-26 
18:51:13.162054107 +0100
+++ linux/drivers/video/intelfb/intelfb.h    2005-01-28 
03:59:58.630904165 +0100
@@ -10,7 +10,8 @@
 /*** Version/name ***/
 #define INTELFB_VERSION            "0.9.2"
 #define INTELFB_MODULE_NAME        "intelfb"
-#define SUPPORTED_CHIPSETS        "830M/845G/852GM/855GM/865G"
+/* XLsigned-2005-01-27: Added 915G */
+#define SUPPORTED_CHIPSETS        "830M/845G/852GM/855GM/865G/915G"
 
 
 /*** Debug/feature defines ***/
@@ -35,10 +36,14 @@
 #define ALLOCATE_FOR_PANNING        1
 #endif
 
+/* XLsigned-2005-01-27: Make this a match of the (one-and-)only entry 
of "my_mini_modedb" */
+#define PREFERRED_MODE            "1024x768-32@73"
+
 #ifndef PREFERRED_MODE
 #define PREFERRED_MODE            "1024x768-16@60"
 #endif
 
+
 /*** hw-related values ***/
 
 /* PCI ids for supported devices */
@@ -46,6 +51,8 @@
 #define PCI_DEVICE_ID_INTEL_845G    0x2562
 #define PCI_DEVICE_ID_INTEL_85XGM    0x3582
 #define PCI_DEVICE_ID_INTEL_865G    0x2572
+/* XLsigned-2005-01-27: Added 915G */
+#define PCI_DEVICE_ID_INTEL_915G    0x2582
 
 /* Size of MMIO region */
 #define INTEL_REG_SIZE            0x80000
@@ -117,7 +124,9 @@
     INTEL_852GME,
     INTEL_855GM,
     INTEL_855GME,
-    INTEL_865G
+    /* XLsigned-2005-01-27: Added 915G */
+    INTEL_865G,
+    INTEL_915G
 };
 
 struct intelfb_hwstate {
diff -ruN linux/drivers/video/intelfb.orig/intelfbhw.c 
linux/drivers/video/intelfb/intelfbhw.c
--- linux/drivers/video/intelfb.orig/intelfbhw.c    2005-01-26 
18:51:13.162054107 +0100
+++ linux/drivers/video/intelfb/intelfbhw.c    2005-01-27 
20:27:12.672227462 +0100
@@ -98,6 +98,12 @@
         *chipset = INTEL_865G;
         *mobile = 0;
         return 0;
+    /* XLsigned-2005-01-27: Added 915G */
+    case PCI_DEVICE_ID_INTEL_915G:
+        *name = "Intel(R) 915G";
+        *chipset = INTEL_915G;
+        *mobile = 0;
+        return 0;
     default:
         return 1;
     }
@@ -169,6 +175,13 @@
         case INTEL_855_GMCH_GMS_STOLEN_32M:
             *stolen_size = MB(32) - KB(132);
             return 0;
+        /* XLsigned-2005-01-27: Added 915G */
+        case INTEL_915G_GMCH_GMS_STOLEN_48M:
+            *stolen_size = MB(48) - KB(132);
+            return 0;
+        case INTEL_915G_GMCH_GMS_STOLEN_64M:
+            *stolen_size = MB(64) - KB(132);
+            return 0;
         case INTEL_855_GMCH_GMS_DISABLED:
             ERR_MSG("video memory is disabled\n");
             return 0;
@@ -390,7 +403,7 @@
 #if VERBOSE > 0
     DBG_MSG("intelfbhw_read_hw_state\n");
 #endif
-
+   
     if (!hw || !dinfo)
         return -1;
 
diff -ruN linux/drivers/video/intelfb.orig/intelfbhw.h 
linux/drivers/video/intelfb/intelfbhw.h
--- linux/drivers/video/intelfb.orig/intelfbhw.h    2005-01-26 
18:51:13.163053957 +0100
+++ linux/drivers/video/intelfb/intelfbhw.h    2005-01-27 
14:37:30.365140440 +0100
@@ -46,6 +46,11 @@
 #define INTEL_855_GMCH_GMS_STOLEN_16M    (0x4 << 4)
 #define INTEL_855_GMCH_GMS_STOLEN_32M    (0x5 << 4)
 
+/* XLsigned-2005-01-27: Added 915G */
+#define INTEL_915G_GMCH_GMS_STOLEN_48M    (0x6 << 4)
+#define INTEL_915G_GMCH_GMS_STOLEN_64M    (0x7 << 4)
+
+
 /* HW registers */
 
 /* Fence registers */


------------------------------------------------------------------------

diff -ru linux/drivers/video/aty.orig/ati_ids.h 
linux/drivers/video/aty/ati_ids.h
--- linux/drivers/video/aty.orig/ati_ids.h    2005-01-26 
05:50:20.235665432 +0100
+++ linux/drivers/video/aty/ati_ids.h    2005-01-26 04:11:46.279111468 +0100
@@ -4,6 +4,9 @@
  * radeonfb
  */
 
+/* XLsigned-2004-11-16i: added RV380 (X600) */
+#define PCI_CHIP_RV380_3E50        0x3E50
+
 #define PCI_CHIP_RS100_4136        0x4136
 #define PCI_CHIP_RS200_4137        0x4137
 #define PCI_CHIP_R300_AD        0x4144
diff -ru linux/drivers/video/aty.orig/radeon_base.c 
linux/drivers/video/aty/radeon_base.c
--- linux/drivers/video/aty.orig/radeon_base.c    2005-01-26 
05:48:50.965236600 +0100
+++ linux/drivers/video/aty/radeon_base.c    2005-01-26 
06:21:33.388589496 +0100
@@ -107,6 +107,13 @@
     { PCI_VENDOR_ID_ATI, id, PCI_ANY_ID, PCI_ANY_ID, 0, 0, (flags) | 
(CHIP_FAMILY_##family) }
 
 static struct pci_device_id radeonfb_pci_table[] = {
+
+        /* XLsigned-2004-11-16: added RV380 (X600)
+         * XLsigned-2005-01-26: added CHIP_IS_MOBILITY flag to force 
reading
+         * of PLL config from VBIOS (BIOS mapped to RAM) and not from 
BIOS-ROM
+         */
+    CHIP_DEF(PCI_CHIP_RV380_3E50,    RV350,    CHIP_HAS_CRTC2 | 
CHIP_IS_MOBILITY),
+   
     /* Mobility M6 */
     CHIP_DEF(PCI_CHIP_RADEON_LY,     RV100,    CHIP_HAS_CRTC2 | 
CHIP_IS_MOBILITY),
     CHIP_DEF(PCI_CHIP_RADEON_LZ,    RV100,    CHIP_HAS_CRTC2 | 
CHIP_IS_MOBILITY),


------------------------------------------------------------------------

--- linux/drivers/video/modedb.c.orig    2005-01-28 04:19:28.710375507 +0100
+++ linux/drivers/video/modedb.c    2005-01-28 04:20:33.595040284 +0100
@@ -18,6 +18,9 @@
 
 #undef DEBUG
 
+/* XLsigned-2005-01-27: Added #define DEBUG 1 */
+#define DEBUG 1
+
 #define name_matches(v, s, l) \
     ((v).name && !strncmp((s), (v).name, (l)) && strlen((v).name) == (l))
 #define res_matches(v, x, y) \
@@ -39,6 +42,19 @@
 
 #define DEFAULT_MODEDB_INDEX    0
 
+/* XLsigned-2005-01-27: Added my_mini_modedb */
+#define MY_PREFERRED_MODE "1024x768-32@73"
+#define IS_MY_PREFERRED_MODE(m) \
+    ((m) && \
+     !strncmp((m),MY_PREFERRED_MODE,strlen(MY_PREFERRED_MODE)) && \
+     (strlen((m)) == strlen(MY_PREFERRED_MODE)))
+static struct fb_videomode my_mini_modedb[] = {
+    {
+    MY_PREFERRED_MODE, 70, 1024, 768, 13333, 180, 24, 29, 3, 136, 6,
+    0, FB_VMODE_NONINTERLACED, 0
+    }
+};
+
 static const __init struct fb_videomode modedb[] = {
     {
     /* 640x400 @ 70 Hz, 31.5 kHz hsync */
@@ -480,7 +496,42 @@
     if (!db) {
     db = modedb;
     dbsize = sizeof(modedb)/sizeof(*modedb);
+    /* XLsigned: 2005-01-27: Added my_mini_modedb
+     * ANY static "modedb" is obviously destroyed when this function
+     * is called from a loadable (framebuffer-)module (like intelfb)
+     */
+    if (mode_option) {
+      DPRINTK("XLsigned-Debug: mode_option = %s\n", mode_option);
+      DPRINTK("XLsigned-Debug: MY_PREFERRED_MODE = %s\n", 
MY_PREFERRED_MODE);
+    }
+    else {
+      DPRINTK("XLsigned-Debug: No mode_option specified\n");
+    }
+    if (IS_MY_PREFERRED_MODE(mode_option)) {
+        my_mini_modedb[0].name = MY_PREFERRED_MODE;
+        my_mini_modedb[0].refresh = 70;
+        my_mini_modedb[0].xres = 1024;
+        my_mini_modedb[0].yres = 768;
+        my_mini_modedb[0].pixclock = 13333;
+        my_mini_modedb[0].left_margin = 180;
+        my_mini_modedb[0].right_margin = 24;
+        my_mini_modedb[0].upper_margin = 29;
+        my_mini_modedb[0].lower_margin = 3;
+        my_mini_modedb[0].hsync_len = 136;
+        my_mini_modedb[0].vsync_len = 6;
+        my_mini_modedb[0].sync = 0;
+        my_mini_modedb[0].vmode = FB_VMODE_NONINTERLACED;
+        my_mini_modedb[0].flag = 0;
+
+        db = my_mini_modedb;
+        dbsize = sizeof(my_mini_modedb)/sizeof(*my_mini_modedb);
+        DPRINTK("XLsigned-Debug: using my_mini_modedb: dbsize = %i\n", 
dbsize);
+    }
+    else {
+        DPRINTK("XLsigned-Debug: using builtin modedb: dbsize = %i\n", 
dbsize);
+    }
     }
+   
     if (!default_mode)
     default_mode = &modedb[DEFAULT_MODEDB_INDEX];
     if (!default_bpp)






-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Re: Added 915G to intelfb of 2.6.10-kernel-sources

[Antonino A. Daplas]

On Saturday 29 January 2005 00:42, Sylvain Meyer wrote:
>         Hi Axel,
>
>     Thanks for your patch but could you format it as described in this
> document
> http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt
>
> and resend it.
>
> The best is to use these tools to maintain and send kernel patches
> http://www.zip.com.au/~akpm/linux/patches/
>
> Also you should subscribe and send fbdev-related mails to
> linux-fbdev-devel@lists.sourceforge.net
>
> Regards
> Sylvain
>
> Buttchereit, Axel (XLsigned) a écrit:
> > Hi,
> >
> > though I'm quite sure that the attached patch (intelfb-...) is far
> > from complete
> > I think that someone/you (guessing from the comments that you are
> > maintaining
> > the code) might be interested in my work.

Can you resend the patch that touches only intelfb?  Also, there are a lot
of XLSigned with the printk's, what is that?

> >
> > Having no technical documentation (from Intel) I had to "develop" the
> > patch by
> > looking at the xfree/xorg-code that already supports I915G and by using
> > trial-and-error (My ThinkCentre's internal graphics device is an I915G
> > and that made me start working on this subject).
> >
> > During development and testing I've discovered that the way "intelfb"
> > and maybe
> > other framebuffer-drivers are using "modedb" does not work, if
> > compiled as a
> > module. The initial "modedb entries" are lost (overwritten?) when
> > "fb_find_mode"
> > is called from the module. Linking "intelfb" into the kernel works as

Because the modedb array is marked __init.  You can however pass
the vesa_modes array instead to fb_find_mode().
 
> > expected.
> >
> > ByTheWay: Why isn't the framebuffer module (intelfb) automatically
> > loaded, if
> > "video=intelfb:..." is specified as kernel-parameter?

If you want it loaded automatically, compile the driver statically. If you don't
want to load a compiled-in driver, use:

video=intelfb:off

Tony




-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Re: Added 915G to intelfb of 2.6.10-kernel-sources

[Buttchereit, Axel (XLsigned)]

Antonino A. Daplas wrote:
> Can you resend the patch that touches only intelfb?  Also, there are a lot
> of XLSigned with the printk's, what is that?
Yes, I will do so.
I've marked all the places where I added/changed something with a
comment that contains my domainname (for finding them easier
during editing). I'll remove this before resending.

> Because the modedb array is marked __init.  You can however pass
> the vesa_modes array instead to fb_find_mode().
Ah, yes!
At the moment "intelfb" does not pass any "modedb" in its call(s) to
"fb_find_mode()" which results in using the "__init modedb" in case
of "static link" and using no modedb (actually using a not-initialized one)
at all, otherwise.
Passing "vesa_modes" would make things better when loading
"intelfb" as a module, but would change "well known behaviour",
when linked statically, though I could perfectly live with an "initfb"
that is always using "vesa_modedb".

Any suggestions?

Bumping into driver development for the first time it's quite
clear that I do not know many important things that I should know.
So please be "forgiving", if I ask stupid questions like this one:

How do I detect, if something (intelfb) is loaded as a module or
statically linked into the kernel?

>>>ByTheWay: Why isn't the framebuffer module (intelfb) automatically
>>>loaded, if
>>>"video=intelfb:..." is specified as kernel-parameter?
> 
> 
> If you want it loaded automatically, compile the driver statically. If you don't
> want to load a compiled-in driver, use:
> 
> video=intelfb:off
This is not exactly what I was asking for. :-)
It would be nice, if the module would automatically be loaded in
"early boot stage", when specified as a kernel-parameter. But this is
not really important (for me), because using "udev" and "hotplug/coldplug"
(on gentoo) finally loads it anyway.

Thanks
--Axel


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl

Re: Re: Added 915G to intelfb of 2.6.10-kernel-sources

[Antonino A. Daplas]

On Saturday 29 January 2005 06:41, Buttchereit, Axel (XLsigned) wrote:
> Antonino A. Daplas wrote:
> > Because the modedb array is marked __init.  You can however pass
> > the vesa_modes array instead to fb_find_mode().
>
> Ah, yes!
> At the moment "intelfb" does not pass any "modedb" in its call(s) to
> "fb_find_mode()" which results in using the "__init modedb" in case
> of "static link" and using no modedb (actually using a not-initialized one)
> at all, otherwise.
> Passing "vesa_modes" would make things better when loading
> "intelfb" as a module, but would change "well known behaviour",
> when linked statically, though I could perfectly live with an "initfb"
> that is always using "vesa_modedb".
>
> Any suggestions?

If you really must, you can pass modedb if compiled statically, and
vesa_modedb if compiled as a module.

>
> Bumping into driver development for the first time it's quite
> clear that I do not know many important things that I should know.
> So please be "forgiving", if I ask stupid questions like this one:
>
> How do I detect, if something (intelfb) is loaded as a module or
> statically linked into the kernel?

What do you mean, detect by whom? Detect by another module
in the kernel or by userspace?

You can do several things:

1. Use #ifdef MODULE/#endif
2. Check if CONFIG_FB_INTELFB_MODULE is defined
3. Check struct fb_info.flags if FBINFO_MODULE is set
4. Use symbol_get()/symbol_put() and check the return value before calling
any of the functions of the driver.

> >>>loaded, if
> >>>"video=intelfb:..." is specified as kernel-parameter?
> >
> > If you want it loaded automatically, compile the driver statically. If
> > you don't want to load a compiled-in driver, use:
> >
> > video=intelfb:off
>
> This is not exactly what I was asking for. :-)
> It would be nice, if the module would automatically be loaded in
> "early boot stage", when specified as a kernel-parameter. But this is

Because at this stage, userspace hasn't been initialized yet, so you cannot
automatically load modules (which are stored in the filesystem) without help
from userspace.

> not really important (for me), because using "udev" and "hotplug/coldplug"
> (on gentoo) finally loads it anyway.
>

Yes, that is a better solution.

Tony




-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl