[BUG] VGA palette corruption in vgacon.c for EFI machine

[BUG] VGA palette corruption in vgacon.c for EFI machine

[Yannick Heneault]

Hi,

It seems there is a small problem of VGA palette corruption on EFI 
machine. When the kernel initializes the architecture, it checks if the 
machine is a EFI machine and if the memory at 0xa0000 is not tagged as 
conventional memory, which should not be the case because this is VGA 
memory. In that case, the kernel assumes that a VGA console can exist. I 
would think that this assumption is correct.

When the kernel initialize the console in vgacon_startup it checks if it 
can really use the VGA console. I think this is where a check is 
missing. Currently, the function can fail if a VESA boot mode is 
detected but not if a EFI boot mode was used. I think the code should be 
modified like this :

--- a/drivers/video/console/vgacon.c
+++ b/drivers/video/console/vgacon.c
@@ -377,7 +377,8 @@ static const char *vgacon_startup(void)
         u16 saved1, saved2;
         volatile u16 *p;

-       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB) {
+       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB ||
+           screen_info.orig_video_isVGA == VIDEO_TYPE_EFI) {
               no_vga:
  #ifdef CONFIG_DUMMY_CONSOLE
                 conswitchp = &dummy_con;


In that case, the VGA console will fail and a dummy console will be 
temporary used until the EFI driver get loaded. At that point the 
console will be mapped to the EFI driver and graphics (splash screen and 
penguins) will appear correctly.

Currently, the function vgacon_startup doesn't fail and initialize the 
video card for a real VGA mode. This function change the first 16 
entries of the VGA palette. When the efifb driver kick-in, the palette 
is not restored to default ramp value, thus the 16 first entry remain in 
a modified state.

Thanks,

Yannick Heneault

[BUG] VGA palette corruption in vgacon.c for EFI machine

[Yannick Heneault]

Hi,

It seems there is a small problem of VGA palette corruption on EFI 
machine. When the kernel initializes the architecture, it checks if the 
machine is a EFI machine and if the memory at 0xa0000 is not tagged as 
conventional memory, which should not be the case because this is VGA 
memory. In that case, the kernel assumes that a VGA console can exist. I 
would think that this assumption is correct.

When the kernel initialize the console in vgacon_startup it checks if it 
can really use the VGA console. I think this is where a check is 
missing. Currently, the function can fail if a VESA boot mode is 
detected but not if a EFI boot mode was used. I think the code should be 
modified like this :

--- a/drivers/video/console/vgacon.c
+++ b/drivers/video/console/vgacon.c
@@ -377,7 +377,8 @@ static const char *vgacon_startup(void)
         u16 saved1, saved2;
         volatile u16 *p;

-       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB) {
+       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB ||
+           screen_info.orig_video_isVGA == VIDEO_TYPE_EFI) {
               no_vga:
  #ifdef CONFIG_DUMMY_CONSOLE
                 conswitchp = &dummy_con;


In that case, the VGA console will fail and a dummy console will be 
temporary used until the EFI driver get loaded. At that point the 
console will be mapped to the EFI driver and graphics (splash screen and 
penguins) will appear correctly.

Currently, the function vgacon_startup doesn't fail and initialize the 
video card for a real VGA mode. This function change the first 16 
entries of the VGA palette. When the efifb driver kick-in, the palette 
is not restored to default ramp value, thus the 16 first entry remain in 
a modified state.

Thanks,

Yannick Heneault

Re: [BUG] VGA palette corruption in vgacon.c for EFI machine

[Andrew Morton]

On Fri, 26 Nov 2010 09:44:51 -0500
Yannick Heneault <yannick_heneault@matrox.com> wrote:

> Hi,
> 
> It seems there is a small problem of VGA palette corruption on EFI 
> machine. When the kernel initializes the architecture, it checks if the 
> machine is a EFI machine and if the memory at 0xa0000 is not tagged as 
> conventional memory, which should not be the case because this is VGA 
> memory. In that case, the kernel assumes that a VGA console can exist. I 
> would think that this assumption is correct.
> 
> When the kernel initialize the console in vgacon_startup it checks if it 
> can really use the VGA console. I think this is where a check is 
> missing. Currently, the function can fail if a VESA boot mode is 
> detected but not if a EFI boot mode was used. I think the code should be 
> modified like this :
> 
> --- a/drivers/video/console/vgacon.c
> +++ b/drivers/video/console/vgacon.c
> @@ -377,7 +377,8 @@ static const char *vgacon_startup(void)
>          u16 saved1, saved2;
>          volatile u16 *p;
> 
> -       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB) {
> +       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB ||
> +           screen_info.orig_video_isVGA == VIDEO_TYPE_EFI) {
>                no_vga:
>   #ifdef CONFIG_DUMMY_CONSOLE
>                  conswitchp = &dummy_con;
> 
> 
> In that case, the VGA console will fail and a dummy console will be 
> temporary used until the EFI driver get loaded. At that point the 
> console will be mapped to the EFI driver and graphics (splash screen and 
> penguins) will appear correctly.
> 
> Currently, the function vgacon_startup doesn't fail and initialize the 
> video card for a real VGA mode. This function change the first 16 
> entries of the VGA palette. When the efifb driver kick-in, the palette 
> is not restored to default ramp value, thus the 16 first entry remain in 
> a modified state.
> 

If that fix worked for you then please send it as a formal, tested,
signed-off kernel patch and we'll put it in the pipeline and see what
emerges from the other end.

[BUG] VGA palette corruption in vgacon.c for EFI machine

[Yannick Heneault]

Hi,

It seems there is a small problem of VGA palette corruption on EFI 
machine. When the kernel initializes the architecture, it checks if the 
machine is a EFI machine and if the memory at 0xa0000 is not tagged as 
conventional memory, which should not be the case because this is VGA 
memory. In that case, the kernel assumes that a VGA console can exist. I 
would think that this assumption is correct.

When the kernel initialize the console in vgacon_startup it checks if it 
can really use the VGA console. I think this is where a check is 
missing. Currently, the function can fail if a VESA boot mode is 
detected but not if a EFI boot mode was used. I think the code should be 
modified like this :

--- a/drivers/video/console/vgacon.c
+++ b/drivers/video/console/vgacon.c
@@ -377,7 +377,8 @@ static const char *vgacon_startup(void)
         u16 saved1, saved2;
         volatile u16 *p;

-       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB) {
+       if (screen_info.orig_video_isVGA == VIDEO_TYPE_VLFB ||
+           screen_info.orig_video_isVGA == VIDEO_TYPE_EFI) {
               no_vga:
  #ifdef CONFIG_DUMMY_CONSOLE
                 conswitchp = &dummy_con;


In that case, the VGA console will fail and a dummy console will be 
temporary used until the EFI driver get loaded. At that point the 
console will be mapped to the EFI driver and graphics (splash screen and 
penguins) will appear correctly.

Currently, the function vgacon_startup doesn't fail and initialize the 
video card for a real VGA mode. This function change the first 16 
entries of the VGA palette. When the efifb driver kick-in, the palette 
is not restored to default ramp value, thus the 16 first entry remain in 
a modified state.

Thanks,

Yannick Heneault