Openmoko Bug #2217: Noise screen of death: Freerunner looses SDIO connection

[Openmoko Public Trac]

#2217: Noise screen of death: Freerunner looses SDIO connection
-----------------------------+----------------------------------------------
 Reporter:  xbaldauf         |          Owner:  openmoko-kernel
     Type:  defect           |         Status:  new            
 Priority:  normal           |      Milestone:                 
Component:  System Software  |        Version:                 
 Severity:  major            |       Keywords:                 
 Haspatch:  0                |      Blockedby:                 
Estimated:                   |    Patchreview:                 
 Blocking:                   |   Reproducible:                 
-----------------------------+----------------------------------------------

Comment(by xbaldauf):

 Replying to [comment:28 andy]:
 > Is the noise always appearing from the "top" when it comes?

 The overwrite direction is always from top to bottom.
 I think it happens almost always that it starts at the top.
 In rare cases (but I do not quite remember), it may be that it starts,
 stops, then starts again in a lower region, so there may be gaps.


 Maybe I should add something to the system setup:
 > That randomish noise is not that easy to generate.
 My complete filesystem is encrypted. That is, I run ext3 over dm-crypt
 over /dev/mmcblk0p2. So, in my case, the encrypted filesystem data is a
 good explanation of why the data visible is actually noisy.


 Triggering the bug may also have something to with the CPU load due to
 encryption or with effects of encryption onto timing (e.g. loading takes
 longer).

 > I wonder if what has happened is that too much data is written from the
 SD Card unit in the Glamo (which is done by local Glamo DMA), so that it
 blows through the allocation of Glamo internal memory and wraps into an
 alias that is the framebuffer.

 This looks like an interesting theory, but as I'm not a hardware hacker, I
 cannot really comment. However, I'm eager to compile some special kernels
 with debugging messages and the like in order to help you diagnose and
 fix.

 I may add that, when running from internal flash memory instead of SD-
 card, I almost never experienced such crashes. However, I cannot properly
 compare, because I cannot run dm-crypt over the internal flash memory,
 because the internal flash memory is not a normal Linux block device.

-- 
Ticket URL: <https://docs.openmoko.org/trac/ticket/2217#comment:29>
docs.openmoko.org <http://docs.openmoko.org/trac/>
openmoko trac