RPI boots to black screen with only the pointer.
Chris Archer (1887) 2 posts |
Hi, At this point I got over excited and tried dragging !Sr2000 to Aemulor, but all I got was a frozen Rpi at a white screen. No pointer or cursor. I power cycled the Rpi, and the boot process started normally with the Riscos banner and the text of the predesk tasks running underneath. At the point were the desktop should appear the screen goes black and just the mouse pointer appears (and can be moved). Nothing else happens or works. Is there a way to recover from this (stop it loading all the predesk stuff)? I presume it is something it loads that crashes it? Has anyone encountered a similar issue before? Could trying to run the game have corrupted the OS? Any help or suggestions would be greatly appreciated. I know I can easily re-image and start a fresh, but i would like to understand how not to break it and also recover it if it does. Thanks |
John Rickman (71) 631 posts |
Yes I have had a similar experience. SD cards are very easily corrupted on the Pi. I have broken 5 or 6 SDs on mine. As a result I have switched from keeping everything on a large SD (16GB) to using small ones(4GB) and keeping everything else on a fast USB pen drive. |
Rick Murray (539) 13430 posts |
Check your power supply and lead. And never ever switch off immediately after shutting down, count to five first. I had the same problems. Better power fixed it. The counting is paranoia… |
John Sandgrounder (1650) 574 posts |
Agreed (with both points).
Now that is a new point for me. I will try that, although my primary RISC OS systems now use Solid State drives for the RISCOS files. The SD cards only have the FAT formatted Pi boot partition and hence are much less liable to suffer corruption. |
Jeffrey Lee (213) 6046 posts |
Hard to say whether it’s filesystem corruption or just a bad configuration setting (e.g. bad CMOS setting, or something bad in the boot sequence). If you hammer the escape key while it’s doing the predesk stuff you might be able to get to the desktop and be able to check for any obvious problems. Running the screen setup plugin and re-selecting the mode it’s using for the desktop might be all that’s needed to fix the problem. |
Chris Archer (1887) 2 posts |
Thanks for all the ideas. Thank-you all for the suggestions and help. |
John Rickman (71) 631 posts |
And never ever switch off immediately after shutting down, count to five first. But especially don’t switch off and on again immediately. I did this to one of the first two IBM PCs to enter the UK. It started a fire in the power supply. Unlikely to happen to a raspberry pi as the power supply is external and soes not have large capacitors, but I have always counted to ten since then. |
Rick Murray (539) 13430 posts |
And never ever switch off immediately after shutting down, count to five first. The logic behind this (and possibly useful for most modern media) is that devices don’t tend to write directly to media. They write to a cache which is then written to media. Or maybe written at the same time (write-through vs write-back). I would imagine that Flash media would, in general, like to try to group writes (with a reasonable level of timeout) because we may see the filesystem as a number of blocks that are 512 bytes, 1024 bytes, etc. The actual Flash blocks may be 64K in size (which means writing one 1K sector could imply an erase and reprogram of an entire 64K block, so it would make sense to use the cache to minimise the number of block writes for sequential filesystem writes. Add to this, RISC OS is not able to inform the media when it is to be dismounted. If you have Flash media with a little light on it, you’ll probably have noticed that dismounting the device in Windows causes the light to go out. It doesn’t do this on RISC OS (nor Linux, that I could see). This is likely a set of requests for the device to flush all data to media, clear the caches, and prepare for removal. The light will go out when it’s all been done, and Windows will tell you that it’s now safe to remove the device. With all of this in mind, it won’t hurt to count to ten. It might hurt more to not do so.
Are you sure it was the power switch and not the 110V switch? :-) Really a PSU shouldn’t catch fire just by a single power cycle. You’d probably blow the capacitors if you did a bunch of on/off cycles, but the PC (was it the 5150?) should NOT have started up the second time as the power supply needs to be off for at least five seconds to generate the “POWER GOOD” signal. Without that, the machine will probably try to switch itself off.
Useful for any inductive load – vacuum cleaner, strip lights, etc. Probably why micro-blackouts (lasting maybe a second tops) can cause more damage than one would expect. |
John Rickman (71) 631 posts |
I had forgotten it was 110V. It was running off 110 volt transformer which I bought from RadioShack. The transformer was in a cardboard box on the floor (no PAT tests then). |