Bug#621032: linux-2.6: please add armhf support
Hi,
2011/4/18 Ben Hutchings <ben@decadent.org.uk>:
> On Sun, 2011-04-17 at 11:39 +0200, Hector Oron wrote:
>> ##
>> ## file: arch/arm/Kconfig
>> ##
>> ## choice: ARM system type
>> # CONFIG_ARCH_REALVIEW is not set
>> ## end choice
>> CONFIG_AEABI=y
>> CONFIG_OABI_COMPAT=y
> Is OABI_COMPAT still important?
Not important at all. But kept as it was on armel default config.
> [...]
>> ##
>> ## file: net/econet/Kconfig
>> ##
>> CONFIG_ECONET_AUNUDP=y
>> CONFIG_ECONET_NATIVE=y
> We already disabled ECONET; this is just cruft in the armel config.
This comes from armel generic config, I have no idea, why it is
selected. We could probably drop it for armhf.
> [...]
>> ##
>> ## file: "kernel/time/Kconfig"
>> ##
>> #. Need to check if this are needed or selected by other symbols.
>> #. CONFIG_NO_HZ=y
>> #. CONFIG_HIGH_RES_TIMERS=y
> These are already set in the master config (debian/config/config).
Great, drop them from armhf then.
>> #. Need to cross-check with rtp if we need split regions.
>> CONFIG_VMSPLIT_3G=y
>> CONFIG_PAGE_OFFSET=0xC0000000
> Why? And you'd better enable HIGHMEM unless these machines are
> definitely limited to 1GB RAM.
At the moment devices are limited to 512MB RAM
We only enable HIGHMEM on Debian buildd kernels (mv78xx0) which it is
not supported by official debian kernel (yet?).
We'll enable HIGHMEM once devices need it.
Those symbols can probably be dropped, as are selected for ARM devices
according to Arnaud Patard.
>> #. At top level CONFIG_HZ_250=y found
>> #. CONFIG_HZ=100
> So why mention it?
Agreed, it can be dropped
>> #. Force console from kernel cmdline
>> CONFIG_CMDLINE="console=ttymxc0,115200 console=tty1"
> No, leave it to the boot loader.
Agreed. Bootloader uses a boot.scr script for booting, in case there
is some error I was thinking it would be helpful to code that in the
CMDLINE, but probably that is not needed.
>> ##
>> ## file: "drivers/Kconfig"
>> ##
>> ## file: "drivers/base/Kconfig"
>> #. CONFIG_DEVTMPFS is not set
> Why?
It is set on toplevel config, nothing wrong to respect that, that it
is left-over from work-in-progress. It can be dropped.
>> #. CONFIG_STANDALONE is not set
>> #. CONFIG_PREVENT_FIRMWARE_BUILD is not set
> Why? We strip all the firmware so there is no way firmware could be
> built.
Efika MX Smartbook uses SDMA for some tasks, hence it uses
sdma-firmware. We should live it out in Debian world I would suspect,
but I leave this one to Arnaud Patard. He is more knowledgeable on
what is going on there.
> [...]
>> ## file: "drivers/mtd/Kconfig"
>> #. Do we need this?
>> #. CONFIG_MTD=y
>> #. CONFIG_MTD_CONCAT=y
>> #. CONFIG_MTD_PARTITIONS=y
>> #. CONFIG_MTD_CMDLINE_PARTS=y
>> #. CONFIG_MTD_CHAR=y
>> #. CONFIG_MTD_BLOCK=y
> If most systems are going to need MTD then you should make MTD and
> MTD_CHAR built-in. Otherwise, they'll be included in the initramfs if
> necessary.
For know, possible targets are Freescales mx5 and maybe omaps, none of
the popular platforms (efikamx, efikasb, beagle, panda) make use of
MTD afaik, then I'll configure it as modules until we support a
platform with real MTD.
>> ## file: "drivers/ide/Kconfig"
>> #. CONFIG_HAVE_IDE=y
> What?
To be removed.
>> ## file: "drivers/scsi/Kconfig"
>> #. XXX Needed to promote to main config?
>> CONFIG_SCSI_MOD=y
> Never set SCSI_MOD directly; only set SCSI.
Agreed.
> [...]
>> ## file: "drivers/i2c/Kconfig"
>> #. XXX It need to check whether is needed
>> #. CONFIG_I2C=y
>> #. CONFIG_I2C_BOARDINFO=y
>> #. CONFIG_I2C_CHARDEV=y
>> #. CONFIG_I2C_ALGOBIT=y
>> #. CONFIG_I2C_ALGOPCF=y
>> #. CONFIG_I2C_ALGOPCA=y
> [...]
> I2C and the various algorithm modules should be selected by other
> drivers as necessary.
Then I'll drop them.
> Ben.
Thanks very much for the review, I'll submit a patch with changes later on.
> Also note you can define armhf flavours to include armel/config and then
> override only a few options. We already share some config files between
> i386 and amd64, powerpc and ppc64, sparc and sparc64, mips and mipsel.
Thanks, I'll have a look to see how that has been implemented.
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: