[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#612376: flash-kernel: please include efikamx support



Hello,

2011/4/4 Loïc Minier <lool@dooz.org>:

>  It would be nice if you could file a bug against linux to get the new
>  flavor; I personally would recommend -mx5.

That's fine with me. I'll wait for the bug report until we had proper
patches for kernel.

>  Technically, mx5 would only support mx51 right now, but would soon
>  allow supporting mx53 as well, notably for mx53loco / quickstart boards
>  which I bet a bunch of Debian ARM users will get.

I guess if we aim to unify, makes sense to use "mx5" subarchitecture.
I'll walk around debian-installer base code to fix subarches. I hope
it does not break previous ubuntu contributions.

>  I reviewed your commits, some notes:
>  - fa8fdbc9607ebd892eba1e1c3b4ad12f710c901f Add support for Efika devices:
>   please sort machine names alphabetically in the case statement
>  - 7e6bd315830ad89c11aadebbe45f1636dd686055 README: Add Efika support:
>   kernel uses "nettop" in cpuinfo, but vendors uses "Smarttop"; also,
>   vendor uses "Efika MX" not just "Efika".  I would prefer just the
>   vendor name in the README  but you could have both with "Efika MX
>   Smarttop (nettop)"; the kernel name is "set in stone" now, but it's
>   not too nice
>  - 4f873843d590b6d5475dcef7d863919c89609b4f
>   flash-kernel-installer.isinstallable: Add mx51 as subarchitecture:
>   I realize this comes from other parts of d-i, but would it make sense
>   to use arm*/mx5?  My preference would be to avoid subarches as long
>   as possible but use upstream names when required; arch/arm/mach-mx5
>   is the current name, so that's what I'd use.
>  - 5deea668506d7a26a8b79fea218101c7900b9752 changelog: add Efika support
>   (Closes: #612376):
>   please amend the currently UNRELEASED changelog entry
>
>  Othewise looked good; thanks!

Please recheck changes:
  http://git.debian.org/?p=d-i/flash-kernel.git;a=commitdiff;h=108de0c2f52301151dd866fe5a0bb7c5f34e5f57

>  I'm also in favor of a boot.scr.

Yes, I agree that flash-kernel needs to know about boot.scr, could we
steal ubuntu code/approach (you explained on IRC) or should I write a
patch for support it?

16:37 < lool> zumbi: well flash-kernel can do whatever is needed
16:38 < lool> zumbi: The main reason I'd use one would be to pass
cmdline args such as root=UUID=xyz to the kernel
              but it's an useful facility for later changes as well
16:38 < lool> zumbi: Up to you
16:39 < lool> zumbi: It's also the only way to persist your cmdline
flags when booting from SD, since you can't
              saveenv to SD
16:39 < lool> zumbi: There are two ways this could be supported:
generated at Debian install time, generated every
              time flash-kernel is run
16:48 < zumbi> lool: ok, let's say, we enable boot.scr support on
flash-kernel-installer, so it is just used at
               Debian install time, if later we find out we need it at
kernel upgrade time, we can always add the code
16:50 < lool> zumbi: If we add it later, it might break expectations
that the file is not overwritten on upgrade, and
              so it might overwrite user data
16:50 < lool> zumbi: In Linaro, we write a boot.txt near the boot.scr
with the contents before calling mkimage; I
              think Ubuntu's flash-kernel will generate boot.scr from
boot.txt if present
16:53 < zumbi> lool: I do not see why not follow Ubuntu approach
17:06 < lool-> zumbi: I'm not suggesting not to follow it either

>  Outside of Martin's remarks, we need to agree on whether this is meant
>  for installation of Debian to PATA, to SD card, or both and whether
>  this is meant to work with vendor's u-boot, upstream u-boot, or both.

I would say we want to install on which ever is mounted under /boot,
so if /boot/boot.txt is present, then boot.scr is regenerated
everytime kernel is installed.

Best regards,
-- 
 Héctor Orón  -.. . -... .. .- -.   -.. . ...- . .-.. --- .--. . .-.

"Our Sun unleashes tremendous flares expelling hot gas into the Solar
System, which one day will disconnect us."

-- Day DVB-T stop working nicely
Video flare: http://antwrp.gsfc.nasa.gov/apod/ap100510.html



Reply to: