[bisected] commit 176f28eb breaks my Sandybridge

[bisected] commit 176f28eb breaks my Sandybridge

[Carl Worth]

The commit below causes my Sandybridge system to fail at initial KMS
initialization:

Aug 27 18:58:04 livid kernel: [    9.099032] i915 0000:00:02.0: irq 42 for MSI/M
SI-X
Aug 27 18:58:04 livid kernel: [    9.099086] [drm:init_ring_common] *ERROR* render ring initialization failed ctl 00000000 head 00000000 tail 00000000 start 00000000
Aug 27 18:58:04 livid kernel: [    9.099266] [drm:i915_driver_load] *ERROR* failed to init modeset
Aug 27 18:58:04 livid kernel: [    9.110257] PM: Removing info for No Bus:card0
Aug 27 18:58:04 livid kernel: [    9.110304] PM: Removing info for No Bus:controlD64
Aug 27 18:58:04 livid kernel: [    9.110326] i915: probe of 0000:00:02.0 failed with error -5

[Ignore the date/time---that really was from a boot today.]

Chris, what can I do to help debug what the correct solution is here?

-Carl

commit 176f28ebf4303b4f7e3a5bd8be7842a8bbecd9c3
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Thu Oct 28 11:18:07 2010 +0100

    drm/i915/ringbuffer: Check that we setup the ringbuffer
    
    Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/inte
index 8eaa60c..e88214e 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -176,9 +176,10 @@ static int init_ring_common(struct intel_ring_buffer *ring)
                        ((ring->gem_object->size - PAGE_SIZE) & RING_NR_PAGES)
                        | RING_NO_REPORT | RING_VALID);
 
-       head = I915_READ_HEAD(ring) & HEAD_ADDR;
        /* If the head is still not zero, the ring is dead */
-       if (head != 0) {
+       if ((I915_READ_CTL(ring) & RING_VALID) == 0 ||
+           I915_READ_START(ring) != obj_priv->gtt_offset ||
+           (I915_READ_HEAD(ring) & HEAD_ADDR) != 0) {
                DRM_ERROR("%s initialization failed "
                                "ctl %08x head %08x tail %08x start %08x\n",
                                ring->name,

Re: [bisected] commit 176f28eb breaks my Sandybridge

[Chris Wilson]

On Thu, 04 Nov 2010 16:03:45 -0700, Carl Worth <cworth@cworth.org> wrote:
> The commit below causes my Sandybridge system to fail at initial KMS
> initialization:
> 
> Aug 27 18:58:04 livid kernel: [    9.099032] i915 0000:00:02.0: irq 42 for MSI/M
> SI-X
> Aug 27 18:58:04 livid kernel: [    9.099086] [drm:init_ring_common] *ERROR* render ring initialization failed ctl 00000000 head 00000000 tail 00000000 start 00000000
> 
> Chris, what can I do to help debug what the correct solution is here?

Find a hw engineer, and ask him nicely why his ringbuffer registers return
0 *most* of the time. You can comment out the check as we have a
workaround in place (auto-reporting of the HEAD) and under certain
circumstances /sys/kernel/debug/dri/0/i915_*ringbuffer_info do report the
correct values. I haven't worked out when they do report the correct
values, it appears to be only when the rings themselves are busy (but not
always). Having the error there was to make sure people noticed and I
could find out just how many SNB revisions failed. I presume you have a
rev 8?
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre

Re: [bisected] commit 176f28eb breaks my Sandybridge

[Carl Worth]

[-- Attachment #1.1: Type: text/plain, Size: 406 bytes --]

On Thu, 04 Nov 2010 23:41:45 +0000, Chris Wilson <chris@chris-wilson.co.uk> wrote:
> Find a hw engineer, and ask him nicely why his ringbuffer registers return
> 0 *most* of the time.

Ugh.

> Having the error there was to make sure people noticed and I
> could find out just how many SNB revisions failed. I presume you have a
> rev 8?

Yes, rev 8 here.

-Carl

-- 
carl.d.worth@intel.com

[-- Attachment #1.2: Type: application/pgp-signature, Size: 189 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [bisected] commit 176f28eb breaks my Sandybridge

[Zhenyu Wang]

[-- Attachment #1.1: Type: text/plain, Size: 441 bytes --]

On 2010.11.04 16:52:51 -0700, Carl Worth wrote:
> > Having the error there was to make sure people noticed and I
> > could find out just how many SNB revisions failed. I presume you have a
> > rev 8?
> 
> Yes, rev 8 here.
> 

FYI, rev 9 desktop sandybridge doesn't show this issue so far.
It just reads back correct values.

-- 
Open Source Technology Center, Intel ltd.

$gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827

[-- Attachment #1.2: Digital signature --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 159 bytes --]

_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: [bisected] commit 176f28eb breaks my Sandybridge

[Chris Wilson]

On Sun, 7 Nov 2010 21:09:16 +0800, Zhenyu Wang <zhenyuw@linux.intel.com> wrote:
> On 2010.11.04 16:52:51 -0700, Carl Worth wrote:
> > > Having the error there was to make sure people noticed and I
> > > could find out just how many SNB revisions failed. I presume you have a
> > > rev 8?
> > 
> > Yes, rev 8 here.
> > 
> 
> FYI, rev 9 desktop sandybridge doesn't show this issue so far.
> It just reads back correct values.

Thanks, I've pushed a patch to -next to not throw an error for SNB rev <= 8
so those of us with the reticent hw can get back to finding bugs.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre