? And, is your battery
charged? Its not normal to shutdown on errors.
I think your phone shutdown because of the low battery, and thats why
you get the -ETIMEDOUT(110) error.
--
Daniel Ribeiro
Em Seg, 2009-04-20 às 18:56 -0300, Daniel Ribeiro escreveu:
Em Dom, 2009-04-19 às 10:58 +0300, Ilya Petrov escreveu:
Looks like it`s broken: http://pastebin.com/f24492489
IRQs on GPIO0/1 are broken on mainline linux.
I tracked the problem, and reported it to LAKML, but i dont know how
a pointer on how to fix this properly from the
PXA maintainer.
Please use an older version until a proper fix is available.
--
Daniel Ribeiro
, or report your findings to the ML. :)
--
Daniel Ribeiro
will apply it and test on all phones i have here, so we figure
out the remaining issues we discussed on IRC.
--
Daniel Ribeiro
on boot_usb. :)
I understand but I don't want to change my laptop ;-)
I was not suggesting you that! We need to know the source of the problem
if we want a definitive solution. (the sleep is not a proper solution).
--
Daniel Ribeiro
for mci .parent too, as
RMK requests that this is done on set_mci_info and not with
set_mci_parent... And i will probably do the same for set_pxa_fb_info to
keep it consistent with mci and udc... happy me. ;)
--
Daniel Ribeiro
perfectly.
Can you please update boot_usb from svn, then uncomment the
#define DEBUG line, remove the usleep, and send the output of a
session that freezes for me?
--
Daniel Ribeiro
If DMA is enabled, any spi_sync call after suspend/resume would block forever,
because DRCMR is lost on suspend. This patch restores DRCMR to the same values
set by probe.
Signed-off-by: Daniel Ribeiro drw...@gmail.com
---
drivers/spi/pxa2xx_spi.c |7 +++
1 files changed, 7 insertions
The card may be powered by a SPI or I2C connected PMIC or voltage regulator.
This patch provides a set_mci_parent function to assert proper suspend/resume
ordering.
Signed-off-by: Daniel Ribeiro drw...@gmail.com
---
arch/arm/mach-pxa/devices.c |5 +
arch/arm/mach-pxa/include
Em Ter, 2009-04-07 às 22:11 +0200, Alain2210 escreveu:
I have tried to build gen-blob as written on
http://wiki.openezx.org/Dual_boot
But I get a file of size 309248 and the genblob on your page : 65536. Is it
normal ?
No.
--
Daniel Ribeiro
-bp.c and move the
bp-init() call from handshake()(step 3) to ezxbp_probe(), just before
the call to handshake()... This will fix gen1 and probably break
gen2. :)
--
Daniel Ribeiro
the bootloader patch on http://people.openezx.org/wyrm/patch_blob
(needed only for deep sleep), and the latest (not the latest anymore)
git code.
Check if you have the dma resume fix on pxa2xx-spi.c applied.
--
Daniel Ribeiro
very broken :( I still dont know what is needed
to fix it.
--
Daniel Ribeiro
Em Qui, 2009-04-02 às 15:04 +0300, Ilya Petrov escreveu:
This one with platform device
Looks good! Thanks!
--
Daniel Ribeiro
, but at least on gen1 BP will still be on flash mode.
--
Daniel Ribeiro
Em Qua, 2009-04-01 às 09:54 +0300, Ilya Petrov escreveu:
2009/4/1 Daniel Ribeiro drw...@gmail.com:
If you have some time to spare, please change the .bpp field on the fb
configuration on our kernel to 24 (packed mode), and see if qtopia still
works with correct colors/alignment. :)
give
Em Qua, 2009-04-01 às 10:57 +0300, Ilya Petrov escreveu:
looks like /sys/devices/platform/pxa2xx-spi.1/spi1.0/adc_chargerid
always returns 1023
The names are meaningless, cat adc_* and you will see that one of them
will change depending on the kind of accessory that is connected.
--
Daniel
will submit a workaround to get this working as soon as
possible.
with normal write speed or as it was before?
Dont know. We need to test it. :)
--
Daniel Ribeiro
on git merges, and for a
good looking topgit dependency tree).
Also, we need a Signed-off-by: line from you to submit this driver
upstream.
Thanks for your contribution!
--
Daniel Ribeiro
armv5tel unknown
w...@brutus:~/devel/boot_usb$ ./boot_usb
$Id: boot_usb.c 2456 2009-03-29 23:46:33Z wyrm $
--
Daniel Ribeiro
Em Seg, 2009-03-30 às 15:23 +0300, Ilya Petrov escreveu:
2009/3/30 Daniel Ribeiro drw...@gmail.com:
Em Seg, 2009-03-30 às 13:08 +0300, Ilya Petrov escreveu:
Same question about sleep mode.
_Sleep_ mode works just fine with:
echo standby /sys/power/state
What is the difference
here it is: http://ezxdev.org/qtopia/temp/blob-1
Please dont change the script!
I will check the file and assert that its ok to patch it and update the
script after being sure it will not harm your phone. :)
--
Daniel Ribeiro
Em Seg, 2009-03-30 às 16:46 +0300, Ilya Petrov escreveu:
2009/3/30 Daniel Ribeiro drw...@gmail.com:
Please dont change the script!
I leave low-level magic to you ;)
Hi!
My conclusion on this is that you are using a self-compiled _A1200E_ lbl
on your phone, without
released kernel sources for A1200E, its a good starting
point to look for differences between the 2 versions.
--
Daniel Ribeiro
_previous_ authorization from its suppliers to
disclose the source code as required by the GPL.
I still have hope that someday motorola will sort these GPL compliance
issues, but the a1600 case is a _regression_ when compared with earlier
source code releases.
Have a nice day!
--
Daniel Ribeiro
.
--
Daniel Ribeiro
, RUN THIS SCRIPT BEFORE PUSHING CHANGES OR RUNNING tg remote
--populate ON OPENEZX.GIT AGAIN! THANKS!
--
Daniel Ribeiro
openezx-git-cleanup.sh
Description: application/shellscript
Brown later to check if it has any chance of being accepted.
I will (re)start some ezx-pcap.c work, we can try the .31 merge window
for it.
--
Daniel Ribeiro
with the
original chip FW.
--
Daniel Ribeiro
Em Qua, 2009-03-11 às 15:05 +0200, Ilya Petrov escreveu:
2009/3/11 Daniel Ribeiro drw...@gmail.com:
Serial port configuration is wrong or the chip is not powered. I think
that something is wrong on your code to power on the chip, maybe a small
delay after the ioctl is enough.
i tried
open root device mmca2 or 02:00', then
the builtin cmdline is being used.
Its possible to set the cmdline with boot_usb if you use gen-blob (check
atags addr at kernel source and boot_usb code), this will help you a lot
on testing. :)
--
Daniel Ribeiro
:)
btw. could someone update http://wiki.openezx.org/Project_Status :)
You are welcome to update the wiki.
--
Daniel Ribeiro
Em Dom, 2009-02-22 às 15:42 +0100, Fred . escreveu:
OK (new version is called blob if you downloaded the binary
recently, in Wyrm's folder.
My mistake. I removed the old version and placed the new one as
gen-blob.
--
Daniel Ribeiro
offset
$ cat offset gen-blob gen-blob-for-gen2
$ boot_usb flash 0xA gen-blob-for-gen2
$ boot_usb flash 0xC kernel-to-flash
# gen2 rootfs address may vary depending on phone and firmware version
$ boot_usb flash XXX backup-of-2.4-rootfs
--
Daniel Ribeiro
0xA kernel # correct
is fatal.
I already have reports of 2 bricked phones with this mistake, so please
update boot_usb or be _very_ cautious.
--
Daniel Ribeiro
is broken after an sysupdate !? -.- )
Please dont. I will delete this file from my folder. And PLEASE update
boot_usb, see the other e-mail i sent today, you've been warned. :)
--
Daniel Ribeiro
charger -.-
Its definitely not worth on gen1. On gen2 a dirty implementation would
be possible with a week of work, on gen1 it would take way too much
time.
--
Daniel Ribeiro
.
--
Daniel Ribeiro
on E680, so
you will need gen-blob flashed to the phone to get it working.
I'm glad to have X with the touchscreen, and usbnet working at least.
I would really like to get bluetooth and power management working
next. It will if I build a kernel with OpenEmbedded, right?
No.
--
Daniel Ribeiro
have no clue on whats wrong.
--
Daniel Ribeiro
backup-of-motorola-kernel
--
Daniel Ribeiro
with the kernel flash as it is on
the first 32mb.
--
Daniel Ribeiro
exploit to defeat the module hash checks?)
My phone has Telnet and FTP access already going.
Hi,
On ezx we can read the firmware with /dev/mem.
On magx, if you cant access the flash with /dev/mem or /dev/mtd* then
you will need to write a kernel module.
--
Daniel Ribeiro
Em Seg, 2009-02-09 às 16:01 +0100, Knut escreveu:
We got usb-host working on the a780. Up to now we successfully tested
usb keyboards, a d-link dwl-122 wifi adaptor and a usb to ethernet
adaptor.
We will release the respective patches soon.
Great! Thanks.
--
Daniel Ribeiro
-blob would not work when sent via boot_usb to the
original bootloader. Also, we would lose precious watchdog time on boot
(we need to start talking to BP as soon as possible after the handshake
is done).
--
Daniel Ribeiro
of the closed-source bootloader is legal. But I dont see any
big issue on publishing information on how to do the patching, or even
some software that does the patching automatically.
--
Daniel Ribeiro
they are interested at the time. What usually
works is attracting a lot of attention to the project by
publishing/advertising it in an accessible way, this will not only
attract new developers but also reward the developers with knowledge
gain and community recognition for their work.
--
Daniel Ribeiro
pressed instead of the VR key to enter gen-blob's usb mode instead of
the MBM flash mode.
Just consider the vol+ trick when flashing the first time as a way to
allow the bootloader to flash itself without changing it's own code. ;)
--
Daniel Ribeiro
at ram address
boot_usb off
power off the phone
--
Daniel Ribeiro
never allow the BP to sleep (by
sending a message every second).
Something is wrong with the resume logic, if you allow BP to sleep, it
will raise WDI as soon as you try to awake it, signalling that
something went wrong on resume.
--
Daniel Ribeiro
CONFIG_MTD_ devices that are not configured.
The mapping driver was never ported from svn to git. See [1], it should
point you how to get mtd working on openezx.git.
[1]http://svn.openezx.org/branches/kernel-2.6.25.x-patches/ezx-mtd-map.patch
--
Daniel Ribeiro
openezx (same with me), and thats why you should focus the
no patches out of mainline approach. My goal is to let users
download the ezx kernel source code from www.kernel.org, not from
git.openezx.org :)
--
Daniel Ribeiro
did was a 32bit write to _RAM_ followed by a
reboot. This should be safe! ;)
--
Daniel Ribeiro
of it (there is an AT command to get time
from the PCAP rtc).
--
Daniel Ribeiro
change the header file.
--
Daniel Ribeiro
pressed, you will see
USB ready on the screen, now send the kernel with boot_usb zImage
1742, the pxa27x-udc driver will not panic the kernel, gen-blob will
turns off the udc before giving control to kernel.
--
Daniel Ribeiro
just make sure that you configure the
kernel tree for arm, the kbuild will link asm-arm to asm, and it will
build.
* The other files add autoconf support.
I will apply these, thanks!
--
Daniel Ribeiro
Em Ter, 2008-11-25 às 21:29 -0200, Daniel Ribeiro escreveu:
* The other files add autoconf support.
I will apply these, thanks!
Applied with the following change:
--- trunk/src/host/boot_usb/configure.ac2008-11-25 23:37:17 UTC (rev
2363)
+++ trunk/src/host/boot_usb/configure.ac
new commands to boot-usb and gen-blob
to make it easy to manipulate flash data.
I also think we should move some code from boot_modes.c to separate
file and call it using parse_command()
Which code? Patch?
--
Daniel Ribeiro
to be the kernel partition.
And, as a final side-note, on gen1 gen-blob itself is loaded to
KERNEL_RAM_BASE(0xa020), so, on gen1 your code copies the kernel
over gen-blob while it is running causing unpredictable results.
--
Daniel Ribeiro
)).
--
Daniel Ribeiro
. see note about asm :)
Yes, it is possible to determine on which phone we are running by
looking at the flash contents, but its hard to write something 100% fail
proof because people can change the flash. :)
--
Daniel Ribeiro
(0xA030)
Too many different blobs. :)
But i still think that we should exploit blob/mbm loading, and not load
the kernel again from flash.
--
Daniel Ribeiro
Em Qui, 2008-11-20 às 06:54 +0200, Ilya Petrov escreveu:
2008/11/20 Daniel Ribeiro [EMAIL PROTECTED]:
hmm, i`l try place it after blob for now.
after blob(0xa0de+1MB) it will work better, but you will still be
limited to a little over 16MB (memory ends at 0xa1ff
11
ezx/pcap_rtc.diff 12
ezx/mach/pcap_rtc.diff 13
ezx/defconfig.diff 14
This order is fine regarding code dependency, and i think its a bit more
logic for review.
--
Daniel Ribeiro
(gen1_pin_config));
pxa2xx_mfp_config(ARRAY_AND_SIZE(a780_pin_config));
--
Daniel Ribeiro
are just on top of other patches in the linear patchstack on
ezx.c means we could think about moving them down a bit and submit them also.
ezx/pcap_rtc
ezx/mach/pcap_rtc
ezx/eoc
ezx/bp
ezx/mach/bp
motoq/mach
motoq/machtype
Not ready.
These are not ready for upstream.
--
Daniel Ribeiro
Em Dom, 2008-11-09 às 20:45 +0100, Bart Visscher escreveu:
Queued, thanks! :)
--
Daniel Ribeiro
I missing some EE/crypto magic in this process?
Its probably easier to do some disassembling/hex editing to f*ck with
this bootloader protection, as long as it is on a writable area of the
flash.
--
Daniel Ribeiro
a security flaw somewhere else and still boot your
freed firmware. All security hardened devices get hacked some day,
its just a matter of time.
--
Daniel Ribeiro
seconds to do your tests and the watchdog will turn the phone
off.
To prevent the watchdog we need to port the ts0710_mux driver, and
this is much harder ;)
--
Daniel Ribeiro
that it will not work with BP on flash mode.
--
Daniel Ribeiro
integrate it on ezx.c.
Timothy, if you like I can prepare a developer account for you. Just send me
your public ssh key and I'll do the rest.
[Timothy added to CC as i dont know if he is subscribed to the ML]
--
Daniel Ribeiro
and lsub).
On device display i see usb0 and ip config for it.
Probably some missing bits for eoc, i will fix it when i have time. :)
--
Daniel Ribeiro
. For your convenience, there is a
default config file on arch/arm/configs/ezx_defconfig, and a ready to
use kernel on http://people.openezx.org/wyrm/images/
Also, assert that you are on the ezx/current branch (git checkout
ezx/current) before building the kernel.
--
Daniel Ribeiro
, and
if
we like it we can switch to that tree.
Topgit seems great! I like it. It looks messy on gitk but it made
everything easier.
To update from our patchset from upstream, i just did:
$ git checkout master
$ git pull
$ git checkout ezx/current
$ tg update
$ git push openezx
--
Daniel
for
(value3).
--
Daniel Ribeiro
handshake, so it may be the
issue).
--
Daniel Ribeiro
on the other thread?
Also, is there any change on your tree i should be aware of?
Particularly code to finish the BP handshake?
--
Daniel Ribeiro
/devices/platform/pxa2xx-spi/spi1.0/adc_* even
when touching the screen (locking seems ok).
--
Daniel Ribeiro
to get the pcap
event handler right, i think that the TS will be as smooth as it was
when running on irq context.
--
Daniel Ribeiro
patchset.
Again, i think that pcap_ts is easy. ezx-pcap is what worries me more,
it may still need some more review rounds to get in shape.
--
Daniel Ribeiro
expose all adc channels and port controls
on sysfs, and let userspace deal with the logic.
Thanks.
--
Daniel Ribeiro
it
unusable. I will try a dedicated workqueue, and if its not enough, a
kthread. Do you have any other suggestion to decrease the latency?
--
Daniel Ribeiro
with framebuffer now, as with normal 32-bit one or not?
No.
Its 18bpp, each pixel is:
XXRR GGBB
You can also change the .bits_per_pixel field to 24, so the pixel will
have the same format as the 2.4 motorola kernel:
XXRR GGBB
--
Daniel Ribeiro
are welcome.
--
Daniel Ribeiro
think about changing the design to treat PCAP_REG_ISR as
events, and provide some sort of interface to register callbacks, and
run all the
code on task context, instead of interrupt context?
--
Daniel Ribeiro
A big thank you to Fred (sabrod on #openezx).
I just arrived home and there was an a910 waiting for me.
My plans are to work on transflash support for it on the
next weekend.
--
Daniel Ribeiro
submission, in
this branch we use guilt and update the patchset from time to time from
the development branch.
This way we would have a fast-forward development branch, much
easier for colaboration, and a patchset for upstream submission.
--
Daniel Ribeiro
Daniel Ribeiro escreveu:
Today me and Antonio did some tests on the git/guilt workflow, but
we do not figure out a way to work with it sharing the code with other
developers. So we decided to use _git_only_.
I think that we can create a branch for upstream submission
process only these ones.
How other projects handle these things? I haven't find clear docs about
that.
As long as the commiter doesnt mix the guilt patches
git-log --name-status is enough to figure out to which patch the commit
belongs.
--
Daniel Ribeiro
pull \
http://ftp.arm.linux.org.uk/pub/armlinux/kernel/git-cur/linux-2.6-arm.git/ \
pxa
$ git push -f
$ guilt pop -a git pull ... guilt push -a git push -f
Let's coordinate some test commits with updates by other devs.
Yes, we need some tests.
--
Daniel Ribeiro
=0df478d8dd5552a828bc9d7d967b027bcb65404a
--
Daniel Ribeiro
commit access, there is no need to apply the
OpenEZX patchset anymore, after the git clone the tree will have
all the OpenEZX patches already applied.
--
Daniel Ribeiro
to fit on 'unsigned short' vars.
Is there any reason to this limit? I am currently faking
the registers to always fit 16 bits, but this looks ugly.
Would a patch to soc-core/dapm to extend this limit to
32 bits registers be accepted?
Thanks!
--
Daniel Ribeiro
and is worth on the long-term.
--
Daniel Ribeiro
to you, use it, instead of
keeping a 2.6.25 which sucks, doesnt work with your micro-sd, and is
slow to flash mtd, as your ??_recovery_purpose_?? kernel on your
??_SD_card_?? which, by the way, doesnt work.
--
Daniel Ribeiro
simply use:
.detect_delay = 15,
--
Daniel Ribeiro
Antonio Ospite escreveu:
Well, I would prefere /* msecs_to_jiffies(150) */ as a comment.
For us the value is 15 even when CONFIG_NO_HZ is used.
ack.
--
Daniel Ribeiro
101 - 200 of 352 matches
Mail list logo
