Bananian

SATA port multiplier support

107 61023
I have recompiled three times following your instructions. Build once takes over an hour, 2nd build that only does kernel after modifying that file and setting CONFIG_AHCI_SUNXI=m, creating the file to say pmp=1. Still no dice.

zgrep CONFIG_AHCI_SUNXI /proc/config.gz is now empty too.

Something is going wrong and I've got no idea what it is, is there anything else in your second build that is different? I even commented out the SATA line as you did the 2nd time.

tkaiser replied at Tue Mar 24, 2015 10:34
You will need a reboot for the new kernel to load.

I simply started on a x86 Ubuntu 14.04 install ...

Just pinging you so you can see my most recent issue, pretty close to giving up as I've compiled so many times

Even in my previous attempt where it showed PMP=Y it should still have worked just not as an easily switchable mode, right?

Any need to change kernel_config=n to y in the build.sh before rebuilding (which I also tried with same result)?

More ideas welcome.

tkaiser  
Edited by tkaiser at Thu Mar 26, 2015 02:58
blindpet replied at Wed Mar 25, 2015 15:38
Even in my previous attempt where it showed PMP=Y it should still have worked just not as an easily switchable mode, right?


When ahci_sunxi is built directly into kernel then SATA should work without PM support. And all you should tweak is the single line in linux-sunxi-next.config as outlined before. Igor's build script uses this file to build the kernel. But now I understand that I've been not precise enough. When you want the image to contain mainline kernel you have to edit build.sh as well and change BRANCH="default" into BRANCH="next"

tkaiser replied at Thu Mar 26, 2015 02:57
When ahci_sunxi is built directly into kernel then SATA should work without PM support. And all yo ...

OK now I have something new to try, are you saying it shouldn't have worked without mainline kernel?

tkaiser  
Edited by tkaiser at Thu Mar 26, 2015 04:47
blindpet replied at Thu Mar 26, 2015 03:17
are you saying it shouldn't have worked without mainline kernel?


When you built an image with BRANCH="default" then kernel 3.4.x is used and there the module approach and the simple switch between PM and non-PM mode does not work (this has been added to 3.18 or 3.19). And since the output of 'zgrep CONFIG_AHCI_SUNXI /proc/config.gz' was empty you used 3.4.x then.

Edited by blindpet at Thu Mar 26, 2015 13:39
tkaiser replied at Thu Mar 26, 2015 04:47
When you built an image with BRANCH="default" then kernel 3.4.x is used and there the module appro ...


Thank you for clarifying, that means in theory I should be able to use the build script for 3.19 and add the module option to get it working?

Do you know how to edit the build.sh so it compiles the 3.19 kernel, doesn't say how on Igor's git.

EDIT: Think I found it, will add BRANCH = "next"
KERNELTAG="v3.19" to build.sh

Edited by blindpet at Thu Mar 26, 2015 17:49
tkaiser replied at Thu Mar 26, 2015 04:47
When you built an image with BRANCH="default" then kernel 3.4.x is used and there the module appro ...


OK I did what I said I would do in the previous post and same result, no working port multiplier with 3.4 still and zgrep CONFIG_AHCI_SUNXI /proc/config.gz is still empty, fml.

tkaiser  
Edited by tkaiser at Fri Mar 27, 2015 00:13
blindpet replied at Thu Mar 26, 2015 17:48
OK I did what I said I would do in the previous post and same result, no working port multiplier


That's because "what you said" is different from what you've been advised to do.

You've been told to modify the line that reads
  1. BRANCH=default
Copy the Code
into
  1. BRANCH=next
Copy the Code
But you chose to do something completely different and add another line somewhere with illegal syntax:
  1. BRANCH = next
Copy the Code
(the spaces cause a syntax error and therefore $BRANCH stays "default" if defined before --> you'll end up with kernel 3.4 and PMP support not working)

And you still have to adjust the contents of one file in Igor's lib directory as outlined before to let ahxi_sunxi be build as a module.

tkaiser replied at Fri Mar 27, 2015 00:11
That's because "what you said" is different from what you've been advised to do.  

You've been  ...

Excellent, well done for spotting that. Unfortunately it still isn't working, now these errors occur during the compilation. I told Igor about it so hopefully he can take a look at it.

All I'm trying to do is build a ready made image that has PMP support, I can't believe this is taking me so long. Thanks again for all your help, I should buy you some beers.
  1. make[5]: *** [drivers/net/phy/swconfig.o] Error 1
  2. make[4]: *** [drivers/net/phy] Error 2
  3. make[3]: *** [drivers/net] Error 2
  4. make[2]: *** [drivers] Error 2
  5. make[1]: *** [deb-pkg] Error 2
  6. make: *** [deb-pkg] Error 2
  7. tar: *.deb: Cannot stat: No such file or directory
  8. tar: Exiting with failure status due to previous errors
  9. rm: cannot remove ‘*.deb’: No such file or directory
  10. ------ Install kernel
  11. dpkg: error processing /tmp/*image*.deb (--install):
  12. cannot access archive: No such file or directory
  13. Errors were encountered while processing:
  14. /tmp/*image*.deb
  15. dpkg: error processing /tmp/*headers*.deb (--install):
  16. cannot access archive: No such file or directory
  17. Errors were encountered while processing:
  18. /tmp/*headers*.deb
  19. dpkg: error processing /tmp/*dtb*.deb (--install):
  20. cannot access archive: No such file or directory
  21. Errors were encountered while processing:
  22. /tmp/*dtb*.deb
Copy the Code

tkaiser replied at Fri Mar 27, 2015 00:11
That's because "what you said" is different from what you've been advised to do.  

You've been  ...

omg it finally works! Thank you for all of your patience, I got in touch with Igor and he gave the last few clues and hallelujah!. i have documented how to get it working and will publish the how to later after I clean up the post.

Something worth noting. I am using that 4 x 2.5" enclosure that supports hot swapping, when I pull one out the other hard drive stays alive and when I pop the drive back in it pops up again. I'm not insinuating that it means they can be used for RAID but it could mean that when pushing data to 1 drive and the other dies in the enclosure it doesn't affect transfer to the other disks. I plan to test this later today after OMV is installed.

You have to log in before you can reply Login | Sign Up

Points Rules