Bug 34165

Created attachment 43237 [details]
Xorg log file

Xorg log file of booting HP 5320m (Arrandale) with KMS and "drm.debug=0xe" enabled

Created attachment 43238 [details]
dmesg output booting with KMS and "drm.debug=0xe"

Created attachment 43240 [details]
lspci of HP5320m iCore5 (Arrandale)

Created attachment 43290 [details] [review]
Probe the eDP device on init

You hardware is very confused. It claims to be using both an eDP and and LVDS connection to the same panel.

One of those is a ghost, presumably the eDP since we have more severe sanity checking for false claims of an LVDS detection. Hopefully the attached should dissuade the claims of having an eDP.

This isn't an HP Probook 5320m, by any chance?

(In reply to comment #4)
> Created an attachment (id=43290) [details]
> Probe the eDP device on init
> 
> You hardware is very confused. It claims to be using both an eDP and and LVDS
> connection to the same panel.
> 
> One of those is a ghost, presumably the eDP since we have more severe sanity
> checking for false claims of an LVDS detection. Hopefully the attached should
> dissuade the claims of having an eDP.
> 
> This isn't an HP Probook 5320m, by any chance?

Oh yes, it is a HP 5320m Pro Book.

(In reply to comment #4)
> Created an attachment (id=43290) [details]
> Probe the eDP device on init
> 
> You hardware is very confused. It claims to be using both an eDP and and LVDS
> connection to the same panel.
> 
> One of those is a ghost, presumably the eDP since we have more severe sanity
> checking for false claims of an LVDS detection. Hopefully the attached should
> dissuade the claims of having an eDP.
> 
> This isn't an HP Probook 5320m, by any chance?

I have added your patch to my kernel, on top of 2.6.37.9 (provided by openSuSE kernel/stable/11.3 repository). But the system crashes now at startup, just after loading i915, KMS and friends. 
I have attached 2 screen shots of the crash, as it fills 2 pages of my frambuffer console:  ("chwilson_patch_kernel_panic_at_boot_page1.jpg", "chwilson_patch_kernel_panic_at_boot_page2.jpg")

Created attachment 43314 [details]
Screenshot of system crashing with eDP patch applied (1of2)

Created attachment 43315 [details]
Screenshot of system crashing with eDP patch applied (2of2)

Retested using kernel 2.6.37.1-1 (provided by openSUSE kernel/stable): Same results, heavy flicker and display distortion using KMS.

Also retried the eDP-Patch on this kernel 2.6.37.1; system panicked just after leaving GRUB.

Created attachment 43558 [details]
kernel 2.6.37.1 + "drm.debug=0xe", dmesg output

Chris,

Can you tell what's so special (or bad) about the HP 5320m ProBook ?

Thomas

Can you try this patch instead?  I think we were trying to read the DPCD too early in the other one.

diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
index 7ffb324..e1b4c88 100644
--- a/drivers/gpu/drm/i915/intel_dp.c
+++ b/drivers/gpu/drm/i915/intel_dp.c
@@ -1949,6 +1949,18 @@ intel_dp_init(struct drm_device *dev, int output_reg)
                ironlake_edp_panel_vdd_off(intel_dp);
        }
 
+       /* confirm the existence of the device with a simple probe */
+       if (intel_encoder->type == INTEL_OUTPUT_EDP &&
+           intel_dp_aux_native_read(intel_dp,
+                                    0x000, intel_dp->dpcd,
+                                    sizeof (intel_dp->dpcd))
+           != sizeof(intel_dp->dpcd)) {
+               kfree(intel_dp);
+               kfree(intel_connector);
+               return;
+       }
+
+
        intel_encoder->hot_plug = intel_dp_hot_plug;
 
        if (is_edp(intel_dp)) {

(In reply to comment #12)
> Can you try this patch instead?  I think we were trying to read the DPCD too
> early in the other one.
> 
> diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
> index 7ffb324..e1b4c88 100644
> --- a/drivers/gpu/drm/i915/intel_dp.c
> +++ b/drivers/gpu/drm/i915/intel_dp.c
> @@ -1949,6 +1949,18 @@ intel_dp_init(struct drm_device *dev, int output_reg)
>                 ironlake_edp_panel_vdd_off(intel_dp);
>         }
> 
> +       /* confirm the existence of the device with a simple probe */
> +       if (intel_encoder->type == INTEL_OUTPUT_EDP &&
> +           intel_dp_aux_native_read(intel_dp,
> +                                    0x000, intel_dp->dpcd,
> +                                    sizeof (intel_dp->dpcd))
> +           != sizeof(intel_dp->dpcd)) {
> +               kfree(intel_dp);
> +               kfree(intel_connector);
> +               return;
> +       }
> +
> +
>         intel_encoder->hot_plug = intel_dp_hot_plug;
> 
>         if (is_edp(intel_dp)) {

Sorry, the patch does not apply: Is this against 2.6.37.1 ? Because I cannot find the "ironlake_edp_panel_vdd_off()" in my drivers/gpu/drm/i915/intel_dp.c

Created attachment 43700 [details]
Patch rejection eDP patch applied against 2.6.37.1

Retried with kernel 2.6.37.2; same behaviour:
uname -a
Linux Kevin 2.6.37.2-1-desktop #1 SMP PREEMPT 2011-02-25 22:30:06 +0100 x86_64 x86_64 x86_64 GNU/Linux

You should be able to apply that code in the same place, the previous block with the vdd_off call won't be there though, but you can just put it above the assignment of intel_encoder->hot_plug = intel_dp_hot_plug;

(In reply to comment #16)
> You should be able to apply that code in the same place, the previous block
> with the vdd_off call won't be there though, but you can just put it above the
> assignment of intel_encoder->hot_plug = intel_dp_hot_plug;

I did this, but again the kernel panic'd shortly after starting from GRUB:

drm:intel_dp_init *ERROR* failed to retrieve link info
general protection fault: 0000 [#1] PREEMPT SMP

CPU 3
Modules linked in: i915(+) drm_kms_helper drm.....
RIP intel_modeset_init +0x576/0x1250 [i915]

Sorry, no camera to take a picture nearby...

Thomas

Created attachment 43987 [details]
Patched intel_dp-c source file, kernel 2.6.37.2

kernel commit 3d3dc149eda48566619d165f6b34e5eeca00edf1
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Sat Feb 12 10:33:12 2011 +0000

    drm/i915/dp: Sanity check eDP existence
    
    Some hardware claims to have both an LVDS panel and an eDP output.
    Whilst this may be true in a rare case, more often it is just broken
    hardware. If we see an eDP device we know that it must be connected and
    so we can confirm its existence with a simple probe.
    
    Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=34165
    Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=24822
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>