Comment 51 for bug 686388

Revision history for this message
Bryce Harrington (bryce) wrote :

Upstream thinks the following two kernel patches (currently on drm-intel-staging) might resolve the issue, however they require testing before they'll commit to them:

commit e6793fa5504ac5c09a8f22f907c2b5f4543af7d9
Author: Chris Wilson <email address hidden>
Date: Tue Mar 29 10:40:27 2011 +0100

    drm/i915: Disable all outputs early, before KMS takeover

    If the outputs are active and continuing to access the GATT when we
    teardown the PTEs, then there is a potential for us to hang the GPU.
    The hang tends to be a PGTBL_ER with either an invalid host access or
    an invalid display plane fetch.

    Reported-by: Pekka Enberg <email address hidden>
    Signed-off-by: Chris Wilson <email address hidden>
    Tested-by: Daniel Vetter <email address hidden> (855GM)

commit b023d74ad16336ea07fb237b52899df6df63e4b2
Author: Chris Wilson <email address hidden>
Date: Tue Mar 29 13:19:09 2011 +0100

    drm/i915: Move the irq wait queue initialisation into the ring init

    Required so that we don't obliterate the queue if initialising the
    rings after the global IRQ handler is installed.

    Signed-off-by: Chris Wilson <email address hidden>