We are starting to see more arm64 vs armhf installation situations.
by adding a line 6: HW_ARCH to the dietpi-obtain_hw_model script, we can change the lines that look like this:
if (( ( $HW_MODEL == 12 ) || ( $HW_MODEL >= 40 && $HW_MODEL < 50 ) )); then
to this:
if (( $HW_ARCH == 3 )); then
proposed:
# - Line 6 -
# HW_ARCH 4 i386 (or VM)??
# HW_ARCH 3 arm64
# HW_ARCH 2 armhf
# HW_ARCH 1 armel
# HW_ARCH 0 unknown
Seeking @Fourdee input, so I didn't automatically assign this to me....
@rhkean Yep, was thinking about this the other day. Would be handy to have, rather than the current HW_MODEL we use now.
Dont think we need armel, all SBC's I know of these days have hardfloat support.
I'd probably suggest something like the following:
# - Line 6 -
# HW_ARCH 21 x86_64 (VM's)
# HW_ARCH 20 x86_32 (you never know :))
# HW_ARCH 10 arm64 (c2) (10 gives us room for any new arm64 types that may come out, and, allows for arm9 arm10 below)
# HW_ARCH 3 arm8 hf (RPi 3)
# HW_ARCH 2 arm7 hf (RPi 2)
# HW_ARCH 1 arm6 hf (RPi 1)
# HW_ARCH 0 unknown
I knew that you'd have better insight!!! LOL
Scrape/parsing notes:
root@DietPi:~# cat /proc/cpuinfo | grep -m1 'architecture' | awk '{print $3}'
AArch64
root@DietPi:~# arch
aarch64
root@DietPi:~# dpkg --print-architecture
arm64
Ok lets go with uname -m: https://github.com/Fourdee/DietPi/issues/341#issuecomment-220879616
We need to get a list going:
arm64 = aarch64
amd64 = x86_64
armv8 = armv8l
armv7 = armv7l
armv6 = armv6l
I assume you're referring to the output the the arch command
RPi A+ : armv6l
Odroid Xu4: armv7l
pineA64+: aarch64
virtualbox image: x86_64
although, we should probably use
uname -m
for portability
according to http://www.gnu.org/software/coreutils/arch , arch is not installed by default
@rhkean
uname -maccording to http://www.gnu.org/software/coreutils/arch , arch is not installed by default
Excellent :+1:
Ok, this is done, wanted to get it in so we can start using HW_ARCH in our code: :+1: https://github.com/Fourdee/DietPi/commit/e44ac7e31d757191c4dc6a628f0ed51503b2c734
Marking as closed.
@Fourdee
NICE!!!
sorry for my lack of contributions for the last week.... swamped at work.
@rhkean
Yep, HW_ARCH should help clean up the code, and, make it easier in the long run, so was a good idea :+1:
No worries Rob :+1: Do what we can, when we can :)
Most helpful comment
@rhkean Yep, was thinking about this the other day. Would be handy to have, rather than the current
HW_MODELwe use now.Dont think we need armel, all SBC's I know of these days have hardfloat support.
I'd probably suggest something like the following: