The size correction in lmc happens at the end, so my best guess is that it's
a qemu issue. Id put a bug in against that, and we should probably release
note it as well.
On Nov 9, 2010 6:58 PM, "john stultz" <johnstul(a)us.ibm.com> wrote:
On Tue, 2010-11-09 at 17:47 -0600, Paul Larson wrote:
> On Tue, Nov 9, 2010 at 5:12 PM, john stultz ...
Sure, if the bug below doesn't warrant removing it.
> Further, after providing an omap3 hwpack to build the image,
Yep. bzr pulled today. I'm on rev 165.
I've just gone back testing older images and hwpacks I've used in the
past and am seeing the same issue there, so it does seem like its
connected to the linaro-media-create scripts.
Here's the command I'm using, just as listed on the wiki:
#linaro-media-create --image_file beagle_sd.img --image_size 4G --swap_file
200 --dev beagle --binary
Per our irc discussion, I tried reproducing this by omitting the
"--image_size 4G" portion, and indeed it seems that worked and the
resulting image boots fine. Similarly it works fine using 8G. Something
just seems off using 4G.
Looking at linaro-media-create to see if I can narrow anything down.
Due to the Linaro 10.11 final release happening tomorrow and
blueprints still being
ironed out, the next two weeks of release meetings have been cancelled. Normal
release meetings will commence on 2010-11-25. This date is a US holiday but I
expect our non-US friends to join us.
Linaro Release Manager
I've put a little time into generating a hyper small footprint linaro
headless image. From UDS the two sizes to try and fit into as I
understand are 64M and 128M installed. So with that in mind, via live
helper I created a headless image where none of the recommended
packages were installed. This brings the installed size of the root fs
down to 165Meg installed (excluding hwpacks)
With a little script-foo which I've created to be part of the
live-helper build I can get the rootfs installed size down to
82,240,000. This is without turning to the usual things to slim things
down like busybox as well as without going through /bin /sbin /usr/bin
/usr/lib /lib and pulling out specific function that isn't needed.
What I have done is gone after the low hanging fruit. That is i18n
locales, documentation, zoneinfo, charmaps and some X11 & firefox
things that obviously aren't needed since those things aren't
installed in headless anyway. All of that gets us down to `125Meg. To
get the next approx 40Meg I have to take out /var/lib/dpkg and
/var/lib/apt. This could make a certain amount of sense for a demo
system running on out NAND flash. This is somewhat problematic however
in so much if you want to use linaro-media-create with hwpacks, well
you need apt to work and that comes with the as mentioned 40 Meg price
Anyway this seemed to be a good Friday night over a beverage type of
hack to gain a little data just to see the effort required.
All in all I do think this is a reasonable approach to get to a "nano"
sized system that would fit into flash. To get that last mile tho, I
think we'd need to turn to busybox, audit the libs and binaries,
adjust the kernel build such that modules are kept to a minimum and
only function actually needed is built in the kernel as well as
working out the last bit of install via something
linaro-media-create-like so that one could at least get things onto an
SD card without too much hackery for test yet not gain the full 123
meg noted below.
du of the 82 Meg system, hwpack not yet installed
du of the 125 Meg system (without /var/lib/apt and /var/lib/dpkg
chopped out) hwpack not installed yet
du of installed system (based on the 125Meg image) after hwpack
installed and system is booted
65312 ./usr <--- grew by 4 Meg
75816 ./var <-- var is ~31 Meg larger after hwpack install!!
86524 ./lib <-- lib after hwpack install is ~79 Meg larger!!
10400 ./boot <-- grew by 10 Meg
248484 . <-- grew by ~123Meg
For those interested in following in my footsteps, my live helper
maverick config is located in here:
du of original 165 Meg headless image without recommended packages.
The following patches are a bunch of OMAP-related fixes.
In summary these patches:
- allow omap3_defconfig and omap_4430sdp_defconfig to build and boot
- get MMC (somewhat) working on Panda board
- fix a bug that caused all 3630s to be reported as ES1.2 (highly
unlikely that these are available in the wild)
Except for Catalin's patch, all of them were cherry-picked from mainline.
I hope I'm not too late to get these merged.
The following changes since commit
Nicolas Pitre (1):
Revert "Revert "Versatile Express: add support for local timers on
are available in the git repository at:
Anand Gadiyar (1):
USB: musb: Kill board specific pinmux from driver file
Benoit Cousson (1):
omap4 hsmmc: Fix the init if CONFIG_MMC_OMAP_HS is not set
Catalin Marinas (1):
ARM: Do not enable SWP emulation if CPU_V6 && CPU_V7
David Anders (1):
omap4: pandaboard: Fix the init if CONFIG_MMC_OMAP_HS is not set
Madhusudhan Chikkature (1):
OMAP4 ES2: HSMMC soft reset change
Menon, Nishanth (1):
omap4: pandaboard: fix up mmc card detect logic
Nishanth Menon (1):
omap3: id: fix 3630 rev detection
Tony Lindgren (2):
omap: Use CONFIG_SMP for test_for_ipi and test_for_ltirq
omap: Fix sev instruction usage for multi-omap
kishore kadiyala (1):
mfd: Adding twl6030 mmc card detect support for MMC1
arch/arm/mach-omap2/Makefile | 1 +
arch/arm/mach-omap2/board-4430sdp.c | 15 ++++-
arch/arm/mach-omap2/board-omap4panda.c | 24 +++++++-
arch/arm/mach-omap2/hsmmc.c | 3 +
arch/arm/mach-omap2/id.c | 2 +-
arch/arm/mach-omap2/include/mach/entry-macro.S | 6 +-
arch/arm/mach-omap2/omap-smp.c | 3 +-
arch/arm/mm/Kconfig | 2 +-
arch/arm/plat-omap/include/plat/mmc.h | 1 +
arch/arm/plat-omap/include/plat/smp.h | 7 --
drivers/mfd/twl6030-irq.c | 73
drivers/mmc/host/omap_hsmmc.c | 15 ++++-
drivers/usb/musb/omap2430.c | 6 --
include/linux/i2c/twl.h | 31 ++++++++++
14 files changed, 162 insertions(+), 27 deletions(-)
Enclosed you'll find a link to the agenda, notes and actions from the
Linaro User Platforms Weekly Status meeting dated November 3rd held in
#linaro-meeting on irc.freenode.net at 13:00 UTC.
- new starter Jian Li on multimedia
- prototype for video playback of 720p mpeg-4 and 30 fps file finished
for ux500 based on linaro netbook-launcher-efl
- multimedia and graphics WG discussions and planning at UDS
- ELC Europe talk by BenjiG (Benjamin Gaignard) on "Android using
User Platforms Team
"We want great men who, when fortune frowns will not be discouraged."
- Colonel Henry Knox
w) tom.gall att linaro.org
w) tom_gall att vnet.ibm.com
h) tom_gall att mac.com