If it helps, I am forwarding Tomi's pull request to Florian for 3.5 merge window. I haven't checked very closely, but looks like the fixes are in that pull request.
On Sat, Jun 2, 2012 at 8:19 AM, John Stultz johnstul@us.ibm.com wrote:
On 05/25/2012 07:24 AM, Tomi Valkeinen wrote:
On Fri, 2012-05-25 at 19:49 +0530, S, Venkatraman wrote:
On Thu, May 24, 2012 at 10:56 PM, John Stultzjohnstul@us.ibm.com wrote:
Hey Arnd, Rob, So after Arnd's help sorting a workaround for the mmc driver, I'm now trying to sort out the following HDMI failure I'm seeing w/ the current 3.5-rc with my Panda Board.
[ 2.973693] omapdss error: HPD IRQ request failed [ 2.978759] omapdss HDMI error: failed to power on device [ 2.982391] smsc95xx 1-1.1:1.0: eth0: register 'smsc95xx' at usb-ehci-omap.09 [ 2.996459] omapdss error: failed to power on [ 3.001037] omapfb omapfb: Failed to enable display 'hdmi' [ 3.006805] omapfb omapfb: failed to initialize default display [ 3.013183] Console: switching to colour dummy device 80x30 [ 3.022430] omapfb omapfb: failed to setup omapfb [ 3.027374] omapfb: probe of omapfb failed with error -5
Any guesses on this?
( This is just guesswork, as you asked :-) ) Apparently, request_threaded_irq requires that IRQF_ONESHOT be passed as a flag, when the handler function is NULL. Many drivers are facing this issue. Russell posted a patch [1] to fix 2 of them (see this patch log). Maybe DSS also requires this change.
[1] http://marc.info/?l=linux-arm-kernel&m=133785047227029&w=2
Yes, I have a fix for this in my pull request. The fbdev maintainer hasn't responded yet to the request.
Hey Tomi,
Sorry, looks I didn't reply-to-all with my earlier reply. This issue is still causing me trouble (well, as of yesterday). Do you have a link to your fix or your git tree so I can do some testing while we wait for it to be merged?
thanks -john