Re: [Angstrom-devel] hx4700 touchscreen fix for linux-2.6.31
On 09:44 Wed 14 Oct , Paul Parsons wrote:
For your consideration is a fix for the hx4700 touchscreen on linux-2.6.31.
I found that touchscreen events were being ignored in ads7846_rx() because
the received pressure values - typically about 550 to 560 - exceeded the
pressure_max value of 512.
So I doubled pressure_max to 1024 and the touchscreen started working.
Finally found some time to quick test this patch - and yes, looks like it
helps on my hx4700 too. As it's mainline kernel issue, can you please
resend this patch with proper Signed-off-by to LAKML
linux-arm-ker...@lists.infradead.org with Cc:
to Philipp Zabel philipp.za...@gmail.com (HX4700 maintainer) and
Eric Miao eric.y.m...@gmail.com (PXA sub-arch maintainer).
Thanks.
diff -c -r linux-2.6.31/arch/arm/mach-pxa/hx4700.c
linux-2.6.32-rc3/arch/arm/mach-pxa/hx4700.c
*** linux-2.6.31/arch/arm/mach-pxa/hx4700.c 2009-10-14 17:15:53.357930822
+0100
--- linux-2.6.32-rc3/arch/arm/mach-pxa/hx4700.c 2009-10-13 21:20:46.421139176
+0100
***
*** 604,610
static const struct ads7846_platform_data tsc2046_info = {
.model= 7846,
.vref_delay_usecs = 100,
! .pressure_max = 512,
.debounce_max = 10,
.debounce_tol = 3,
.debounce_rep = 1,
--- 604,610
static const struct ads7846_platform_data tsc2046_info = {
.model= 7846,
.vref_delay_usecs = 100,
! .pressure_max = 1024,
.debounce_max = 10,
.debounce_tol = 3,
.debounce_rep = 1,
Regards.
___
Angstrom-distro-devel mailing list
Angstrom-distro-devel@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
--
Best regards,
Dmitry MAD Artamonow
___
Angstrom-distro-devel mailing list
Angstrom-distro-devel@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] linux-2.6.31 config for hx4700 ?
On 22:41 Thu 17 Sep , Paul Parsons wrote: The main word here is _basic_ - there's still a lot of missing features. Touchscreen for example. OK, I understand. That might explain why nothing appears on screen! Hmm, there's no support for touchscreen, but framebuffer itself should work. Check if you have CONFIG_FB_W100=y in .config and console=tty0 in kernel parameters line (that's in startup.txt if you are booting via HaRET). FYI, here's the config for hx4700 I used last time I bothered to try this new kernel: http://mad.is-a-geek.org/tmp/config.hx4700.2.6.31 -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] linux-2.6.31 config for hx4700 ?
On 11:59 Thu 17 Sep , Paul Parsons wrote: The linux-2.6.31 release added basic support for HP iPAQ hx4700 PDAs. Hooray. The main word here is _basic_ - there's still a lot of missing features. Touchscreen for example. Can anyone point me to a config specifically for the hx4700? Or is the included arch/arm/configs/magician_defconfig the recommended starting point. Yes arch/arm/configs/magician_defconfig is the one you should start with. After coming close to bricking my hx4700 with 2.6.31 yesterday I would prefer to start from a known working config. I should probably explore booting it from CF too. IIRC, there's no support for CF on hx4700 in mainline 2.6.31. But if you familiar with git, you may try to build kernel from Philipp Zabel's git repo (dev branch): http://git.linuxtogo.org/?p=ph5/kernel.git;a=shortlog;h=refs/heads/dev -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] hx4700, 2.6.28 upgrade with acx-mac80211
On 22:56 Tue 14 Apr , Oliver wrote: Hello, Since some time I have started working on an 2.6.28 kernel upgrade for the hx4700 - driven by curiosity and also to check it out myself a bit ... . The main purpose was in fact to get the acx100 with mac80211, which would result in WPA wifi support. My plan would be to get the WPA working, since I have my teeth in it, and then do an rebase on 2.6.29 (or 30?) and continue with the other hx features. Nice to hear that someone is working on a kernel for hx4700! I have this PDA, but sadly don't have enough time to work on the kernel for it :( BTW, you may want to look at Philipp Zabel's work here[1] - there's some support for hx4700 in mag-29-rc3 branch (as name implies based on 2.6.29-rc3). Perhaps you may want to contact Philipp to check the current status of this work. [1] http://git.linuxtogo.org/?p=ph5/kernel.git;a=summary -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] h3600 kernel
On 14:42 Sat 21 Mar , flameman mayer wrote: hi, i applied your patch to vanilla 2.6.28: it works good, so i can switch this kernel as the working line kernel Well, in that case you might be interested in another update I just uploaded to http://mad.is-a-geek.org/h3600/ Changes: * Some more patches were pushed upstream, so I put in tarball final versions of them cherry-picked from linux-next tree * All patches now checkpatch clean (and even that pile of junk - sa1100_udc now has just three warnings) * Micro driver converted to use MFD cell API * Suspend finally fixed with couple of hacks I wrote one dirty hack to workaround bootldr bug and also added Thomas Kunze PCMCIA hack from OE I think I'll try to look into preparing a patch for OE next week. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] Booting angstrom on iPaq H3630
On 09:57 Thu 19 Mar , Vitus Jensen wrote: Hello all! I'm trying to get my iPaq H3630 to boot angstrom but the kernel can't mount to root device on /dev/mtdblock1. I've verified that the rootfs is readable and at the correct index: boot partition show argv[1]=show npartitions=0002 bootldr base: size: 0004 end: 0004 flags: 0002 root base: 0004 size: 00FC end: 0100 flags: 0018 boot show Compaq OHH BootLoader, Rev 2-18-1 [BIG_KERNEL] [MD5] [MONO] Last link date: Tue Jan 22 17:46:27 EST 2002 Contact: boot...@handhelds.org boot_type= jffs2 linuxargs= noinitrd root=/dev/mtdblock1 init=/linuxrc console=ttySA0 mach_type=22 kernel_partition= root kernel_filename= boot/zImage ... What I've noticed is that there is no output about MTD partitions during the boot, like it's done here http://osdir.com/ml/handhelds.linux.h5400/2005-05/msg00028.html In the linux-handhelds-2.6_2.6.21-hh20.bb recipe H5000 get's CONFIG_MTD_PARTITIONS=y set but H3630 is missing that parameter. I'm a on the right track when setting CONFIG_MTD_PARTITIONS=y for my iPaq too? I think so. And you'll need CONFIG_MTD_SA1100 too. BTW, if I remember correctly, handhelds.org 2.6.21 kernel have a strange default partitioning scheme (something like 4 or 5 different partitions, and not just 2 as bootldr thinks), so maybe you'll need to patch a kernel or tweak bootldr's params to pass right partitioning scheme via command line (don't forget to turn on CONFIG_MTD_CMDLINE_PARTS in that case). -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] h3600 kernel
On 23:16 Thu 12 Mar , Bernhard Guillon wrote: Dmitry Artamonow wrote: Thanks for offering, but I don't think it's worth the effort. At this moment my patchset consists of unfinished cleanup of vanilla tree (which has much less features even than hh.org kernel) plus some random proof-of-concept patches, that need some more work to become useful. In my opinion it will be a good base to start :). Restart was something which really helped the simpad. Can you create a patch for oe? It would make adding and picking the good patches easier for me. Errm, do you mean a patch against OE, that will add my patchset and defconfig and do appropriate changes to some linux_2.6.2x recipe ? Or just some consistent patchset against kernel that will be suitable to including into OE? In either cases, I guess I'll wait for 2.6.29 to be released, but in the meantime I'll try to sync all the patches I have, so they can be applied simultaneously. I'll announce when I have the updated version. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] h3600 kernel
On 13:08 Fri 13 Mar , flameman mayer wrote: umm i tested vanilla 2.6.29-rc1...rc7 with your patch set well, it seems your patches need some more tune up in order to work, cause applying the current is working, but a lot of stuff is included in the misc patch subtree, which is not applying to vanilla: for example i have no back light support cause it is included in the misc subtree. Patches in misc are quite random (that's why I called it 'misc' :), and they probably won't apply on top of 'current' branch, which is focused on the h3600 machine file cleanup and movement towards gpiolib conversion. However, I'm working on making all these patches applicable in one batch. Stay tuned. so could you provide us a git access in order to git clone it ? it should help me great Sorry, no. At least not now. Anyway, you can clone Linus' tree, make branch and apply patches with git-am. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] h3600 kernel
On 23:59 Fri 06 Mar , Bernhard Guillon wrote: Alessandro Briosi wrote: If it's of any interest there is [1] which was a fork from hh.org kernel tree. We should test this fork and maybe switch to it. Is anyone already using this fork for a machine? It's as dead as the original hh.org tree. There's only two branches related to hh.org code: 'hh.org' - mirrors hh.org CVS (2.6.21) and 'work' - hh.org rebased to 2.6.23 and partially fixed. Still 'work' branch lacks all of the features of original 2.6.21 hh.org kernel because of breakage, and there was no work on it since last summer. I know that I'm talking about because last commits to it were mine. BTW, AFAIK there was no h3600-related commits in this branch, and probably code was broken during rebase because hh.org code largely differs from vanilla one. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] ANGSTROM_MODE renamed to ANGSTROMLIBC
On 11:39 Mon 08 Dec , Koen Kooi wrote: Hi, Due to changes in how bitbake handles environment vars ANGSTROM_MODE has been renamed to ANGSTROMLIBC: http://cgit.openembedded.net/cgit.cgi?url=openembedded/commit/id=325476d19b821623dbe151d7026c20dfd3754432 Please update your private buildscripts accordingly, the public angstrom ones have already been updated. I think Building Ångström page [1] should be updated as well - it mentions ANGSTROM_MODE in the couple of places. regards, Koen [1] http://www.angstrom-distribution.org/building-%C3%A5ngstr%C3%B6m -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] xserver-kdrive 1.4 investigations
On 16:58 Wed 24 Sep , Koen Kooi wrote: Koen Kooi wrote: Dmitry Artamonow wrote: Koen, what was the motivation behind that kdrive-1.4-fixes-fix.patch ? I'm not sure, since I left out the why in the commit[1], IIRC it breaks things when you have an external (e.g. USB) keyboard. Placing it outside the case statement apparently broke things. Well, if it brokes external keyboards, then it will do so, only for some machines which is not Zaurus/iPAQ/Simpad/etc. Let's see. without kdrive-1.4-fixes-fix.patch order was: * apply default keymap * apply machine-specific keymap overrides for known machines now with kdrive-1.4-fixes-fix.patch it's: * apply machine-specific keymaps for known machines * if machine isn't known, then apply default keymap I think that the first approach was generally right, just maybe it needed some tuning in part of what should be in default keymap and what in machine-specific. Someone should see if xkeyboard-config[2] fixes the keymap issues I didn't tried, but I doubt that it will help. At least not with our kdrive. As stated in [4] The project is targeted to XKB-based systems, and if you look into packages/xorg-xserver/xserver-kdrive-common.inc in OE you'll see --disable-xkb in EXTRA_OECONF. Actually, there might be an easier short-term solution: Remove all empty assignments from the keymap so xmodmap doesn't overwrite those, like the mamemo keymap[3] does. Any volunteers for trying that? Does the attached patch looks ok? I've compiled default.xmodmap out of existing patches, removed empty keycodes and placed it into separate file for easier maintaining. Sadly, I don't have any machines with external keyboards here, but I can confirm that it works on h2200 as expected. [1] http://gitweb.openembedded.net/?p=org.openembedded.dev.git;a=history;f=packages/xserver-common;hb=f7e14b297624d0ac3d350772abb8614004ab7a1d [2] http://svn.o-hand.com/view/poky?rev=5212view=rev [3] http://dev.openbossa.org/trac/mamona/browser/packages/xserver-kdrive-common/xserver-kdrive-common/mamona/keymap.xmodmap?rev=e1a27a129b9ae3c921a41bc76b14ed5b500252b7 [4] http://freedesktop.org/wiki/Software/XKeyboardConfig -- Best regards, Dmitry MAD Artamonow From f17344a6730ced68bc107988b02c61e095de6833 Mon Sep 17 00:00:00 2001 From: mad [EMAIL PROTECTED] Date: Fri, 26 Sep 2008 11:43:22 +0400 Subject: [PATCH] xserver-common: rework default keymap * outfactored default.xmodmap out of kdrive-1.4-fixes.patch and fixes-fix-fix.patch for easier maintaining * empty keycodes removed from default.xmodmap * removed kdrive-1.4-fixes-fix.patch, so now default.xmodmap applied before all machine-specific xmodmaps again --- packages/xserver-common/files/default.xmodmap | 123 + packages/xserver-common/files/fixes-fix-fix.patch | 20 -- .../files/kdrive-1.4-fixes-fix.patch | 22 -- .../xserver-common/files/kdrive-1.4-fixes.patch| 264 packages/xserver-common/xserver-common_1.22.bb | 10 +- 5 files changed, 129 insertions(+), 310 deletions(-) create mode 100644 packages/xserver-common/files/default.xmodmap delete mode 100644 packages/xserver-common/files/fixes-fix-fix.patch delete mode 100644 packages/xserver-common/files/kdrive-1.4-fixes-fix.patch diff --git a/packages/xserver-common/files/default.xmodmap b/packages/xserver-common/files/default.xmodmap new file mode 100644 index 000..005257d --- /dev/null +++ b/packages/xserver-common/files/default.xmodmap @@ -0,0 +1,123 @@ +keycode 9 = Escape +keycode 10 = 1 exclam +keycode 11 = 2 at +keycode 12 = 3 numbersign +keycode 13 = 4 dollar +keycode 14 = 5 percent +keycode 15 = 6 asciicircum +keycode 16 = 7 ampersand braceleft +keycode 17 = 8 asterisk bracketleft +keycode 18 = 9 parenleft bracketright +keycode 19 = 0 parenright braceright +keycode 20 = minus underscore backslash +keycode 21 = equal plus +keycode 22 = BackSpace +keycode 23 = Tab +keycode 24 = q Q q +keycode 25 = w W w +keycode 26 = e E +keycode 27 = r R r +keycode 28 = t T t +keycode 29 = y Y y +keycode 30 = u U u +keycode 31 = i I i +keycode 32 = o O o +keycode 33 = p P p +keycode 34 = bracketleft braceleft +keycode 35 = bracketright braceright asciitilde +keycode 36 = Return +keycode 37 = Control_L +keycode 38 = a A +keycode 39 = s S s +keycode 40 = d D +keycode 41 = f F +keycode 42 = g G g +keycode 43 = h H h +keycode 44 = j J j +keycode 45 = k K k +keycode 46 = l L l +keycode 47 = semicolon colon +keycode 48 = apostrophe quotedbl +keycode 49 = grave asciitilde +keycode 50 = Shift_L +keycode 51 = backslash bar +keycode 52 = z Z z +keycode 53 = x X x +keycode 54 = c C +keycode 55 = v V v +keycode 56 = b B +keycode 57 = n N n +keycode 58 = m M m +keycode 59 = comma less +keycode 60 = period greater +keycode 61 = slash question +keycode 62 = Shift_R +keycode 63 = KP_Multiply +keycode 64 = Alt_L +keycode 65 = space +keycode 66 = Caps_Lock +keycode 67 = F1
[Angstrom-devel] xserver-kdrive 1.4 investigations
Today I've looked into infamous kdrive-1.4 problem.
For those who doesn't know - kdrive starting from version 1.4 doesn't
ship any default keyboard mapping, that renders keyboards (both virtual
and hardware) unusable. And because 1.4 is the default version in .dev,
we have no luck :(
After some looking into /etc/X11/ on target machine and into
packages/xserver-common recipe in OE, I found that there is already a
workaround for that - only it's disabled for all machines, which have
their own xmodmap additions (like mapping for app buttons on
keyboardless devices, etc).
Here's the patch that makes my h2200 happy again with kdrive-1.4
(and it looks like it doesn't break kdrive-1.3 either).
Sorry, can't test it on any other hardware now, but I think it should
help with other devices too, because problem is generic.
-
diff --git a/packages/xserver-common/files/kdrive-1.4-fixes-fix.patch
b/packages/xserver-common/files/kdrive-1.4-fixes-fix.patch
deleted file mode 100644
index 14b77e2..000
--- a/packages/xserver-common/files/kdrive-1.4-fixes-fix.patch
+++ /dev/null
@@ -1,22 +0,0 @@
/tmp/12keymap 2008-06-07 12:24:24.425525517 +0200
-+++ xserver-common-1.22/X11/Xinit.d/12keymap 2008-06-07 12:27:35.152175933
+0200
-@@ -2,9 +2,6 @@
-
- MACHINE=`awk 'BEGIN { FS=: } /Hardware/ { print $2 } ' /proc/cpuinfo`
-
--# since kdrive 1.4 there is no default keymap in server
--xmodmap - /etc/X11/default.xmodmap
--
- case $MACHINE in
- SHARP Shepherd | SHARP Husky | SHARP Corgi)
- xmodmap - /etc/X11/shepherd.xmodmap
-@@ -28,5 +25,9 @@
- Simpad | HP iPAQ* | Asus MyPal*)
- xmodmap - /etc/X11/keyboardless.xmodmap
- ;;
-+ *)
-+ # since kdrive 1.4 there is no default keymap in server
-+ xmodmap - /etc/X11/default.xmodmap
-+ ;;
- esac
-
diff --git a/packages/xserver-common/xserver-common_1.22.bb
b/packages/xserver-common/xserver-common_1.22.bb
index 9ef4e45..b2df433 100644
--- a/packages/xserver-common/xserver-common_1.22.bb
+++ b/packages/xserver-common/xserver-common_1.22.bb
@@ -2,7 +2,7 @@ DESCRIPTION = Common X11 scripts and support files
LICENSE = GPL
SECTION = x11
RDEPENDS_${PN} = xmodmap xrandr xdpyinfo
-PR = r6
+PR = r7
PACKAGE_ARCH = all
@@ -11,7 +11,6 @@ inherit gpe
SRC_URI_append = file://setDPI.sh
SRC_URI_append_angstrom = file://kdrive-1.4-fixes.patch;patch=1 \
-file://kdrive-1.4-fixes-fix.patch;patch=1 \
file://fixes-fix-fix.patch;patch=1 \
-
Koen, what was the motivation behind that kdrive-1.4-fixes-fix.patch ?
--
Best regards,
Dmitry MAD Artamonow
___
Angstrom-distro-devel mailing list
Angstrom-distro-devel@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] xserver-kdrive 1.4 investigations
Forgot to say - looks like Mamona guys have similar local workaround for xserver-kdrive-common recipe. See this link: http://dev.openbossa.org/trac/mamona/changeset?new=e1a27a129b9ae3c921a41bc76b14ed5b500252b7%40packages%2Fxserver-kdrive-commonold=9ef7499f51f5e81ac44d7148365291fc6779a07b%40packages%2Fxserver-kdrive-common -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] Does .stable branch updates via monotone?
On Mon, Sep 22, 2008 at 10:08:41AM +0200, Koen Kooi wrote: Dmitry Artamonow wrote: Playing with .stable branch recently I found that it doesn't updates for me anymore and I don't get recent commits via monotone. Doing mtn pull; mtn update shows this: [EMAIL PROTECTED] ~/OE/org.openembedded.stable $ mtn pull ; mtn update mtn: doing anonymous pull; use -kKEYNAME if you need authentication mtn: connecting to monotone.openembedded.org mtn: finding items to synchronize: mtn: certificates | keys | revisions mtn:78772 | 76 | 25951 mtn: bytes in | bytes out | certs in | revs in mtn: 20.7 k |15.4 k |12/12 | 3/3 mtn: successful exchange with monotone.openembedded.org mtn: updating along branch 'org.openembedded.stable' mtn: already up to date at 97f6800e7228874b1f13f700a43a88bb5c10a4bb Looks for me like something was broken while migrating to git. There has been no migration to git yet, everything is still done in monotone. Oops, just fetched .stable from scratch using instructions on http://www.angstrom-distribution.org/building-ångström page and looks like everything in place. So probably it was something with my local copy. Sorry for the noise and thanks for clearing up git migration status. There's no problem for me to switch to git either, but my main concern is about Angstrom autobuilder scripts - was they adapted to use git already? If not, then we may get -r19 release looking exactly like -r18 :) Also, if translating .stable branch from git to mtn won't get fixed, Building Angstrom page also will need update. -- best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] Angstrom Stable Changelog page
On 00:51 Mon 22 Sep , Junqian Gordon Xu wrote:
On 09/21/2008 03:54 PM, Dmitry Artamonow wrote:
On 05:16 Sun 21 Sep , Junqian Gordon Xu wrote:
On 09/20/2008 05:30 AM, Dmitry Artamonow wrote:
On 17:36 Wed 17 Sep , Junqian Gordon Xu wrote:
[]
Good idea, especially in part of package requests for next stable
release. I've added leafpad to list of tested programs - I'm using it
actively for about a half year on hx4700 and c700, so version present
in .stable (0.8.9-r1) is ready for using.
I've updated it to 0.8.15. Please test in dev
Ok, I tested it on h2200 with .dev and with .stable on hx4700.
All works ok, except for file and font choosers - both are hardly
usable on QVGA (at least in .dev - thanks to Paul Sokolovsky in .stable
we have more usable filechooser). But it's problem with gtk, not
leafpad.
There is nothing we can do for now until somebody with time and skill
hackes the gtk code again. The usability could be better on VGA screen?
Yes, on VGA all is ok, and even on QVGA you still can choose files,
it's just harder, because only 1-2 letters from filenames fit into
window.
Concerning gtk+ dialogs I don't see the reason for spending time to
trying fix them in .devel branch now, because .dev is a moving
target. Let's just stick to our patched gtk+ version in stable branch
and who knows, maybe by the time next .stable branch will be created,
upstream offer some solution, so we won't need patching anymore?
BTW, recipe can be cleaned up some more - we could use icon and .desktop
file provided by leafpad itself and get rid of custom OE replacements
and corresponding do_install_append section in recipe. Also it would be
nice to package leafpad's icon set separately - it doesn't really needed
for leafpad operation, but costs about 30kb disk space (don't forget
that we have machines with 32 and even 16megs of flash, so every byte
is precious). Here's my resulting .bb file:
Good idea
--
DESCRIPTION = GTK+ based simple text editor
HOMEPAGE = http://tarot.freeshell.org/leafpad;
AUTHOR = Tarot Osuji [EMAIL PROTECTED]
SECTION = x11/applications
LICENSE = GPLv2
DEPENDS = gtk+
SRC_URI = http://savannah.nongnu.org/download/${PN}/${PN}-${PV}.tar.gz;
PR = r1
inherit autotools pkgconfig
EXTRA_OECONF = --enable-chooser --enable-emacs --disable-print
PACKAGES += ${PN}-icons
PACKAGES =+, otherwise ${PN} precedes ${PN}-icons and takes over the
${datadir}
Oh, thanks for explanation! I never tried to split packages before :)
FILES_${PN}-icons = ${datadir}/icons
--
Even with this recipe there's still room for improvement - for example,
we can investigate why leafpad needs icon in /usr/share/pixmaps in 2
formats - .png and .xpm. Also we can tune description of ${PN}-icons
package.
But it's a late night already and I prefer to have some sleep instead.
I'm not sure what leafpad.xpm is for. I would leave it in for now.
Attached is an mtn diff from my local tree, please review
Hmm, should it be DESCRIPTION_${PN}-icons instead of
DESCRIPTION_${PN}_icons ? Otherwise looks good.
BTW, seems like opkg in .dev doesn't show package descriptions at all
(try to issue 'opkg list_installed' for example)
--
Best regards,
Dmitry MAD Artamonow
___
Angstrom-distro-devel mailing list
Angstrom-distro-devel@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] Angstrom Stable Changelog page
On 05:16 Sun 21 Sep , Junqian Gordon Xu wrote:
On 09/20/2008 05:30 AM, Dmitry Artamonow wrote:
On 17:36 Wed 17 Sep , Junqian Gordon Xu wrote:
[]
Good idea, especially in part of package requests for next stable
release. I've added leafpad to list of tested programs - I'm using it
actively for about a half year on hx4700 and c700, so version present
in .stable (0.8.9-r1) is ready for using.
I've updated it to 0.8.15. Please test in dev
Ok, I tested it on h2200 with .dev and with .stable on hx4700.
All works ok, except for file and font choosers - both are hardly
usable on QVGA (at least in .dev - thanks to Paul Sokolovsky in .stable
we have more usable filechooser). But it's problem with gtk, not
leafpad.
BTW, recipe can be cleaned up some more - we could use icon and .desktop
file provided by leafpad itself and get rid of custom OE replacements
and corresponding do_install_append section in recipe. Also it would be
nice to package leafpad's icon set separately - it doesn't really needed
for leafpad operation, but costs about 30kb disk space (don't forget
that we have machines with 32 and even 16megs of flash, so every byte
is precious). Here's my resulting .bb file:
--
DESCRIPTION = GTK+ based simple text editor
HOMEPAGE = http://tarot.freeshell.org/leafpad;
AUTHOR = Tarot Osuji [EMAIL PROTECTED]
SECTION = x11/applications
LICENSE = GPLv2
DEPENDS = gtk+
SRC_URI = http://savannah.nongnu.org/download/${PN}/${PN}-${PV}.tar.gz;
PR = r1
inherit autotools pkgconfig
EXTRA_OECONF = --enable-chooser --enable-emacs --disable-print
PACKAGES += ${PN}-icons
FILES_${PN}-icons = ${datadir}/icons
--
Even with this recipe there's still room for improvement - for example,
we can investigate why leafpad needs icon in /usr/share/pixmaps in 2
formats - .png and .xpm. Also we can tune description of ${PN}-icons package.
But it's a late night already and I prefer to have some sleep instead.
The only thing that confuses
me, what do you mean by 'recipe QA checked' and is it the mandatory for
getting packages into feeds?
Yes it is. e.g. problems in leafpad_0.8.9.bb
* missing fields: DESCRIPTION, HOMEPAGE, AUTHOR, *SECTION* (affecting feed
sorting)
It's a good opportunity to weed out problems like this before adding
packages to the feed.
Well, if you'll help with all backporting troubles, why not.
Otherwise, I would prefer to have working software in feeds built by
ugly recipes, than to have nothing at all.
I ran leafpad_0.8.9.bb through contrib/oe-stylize.py and it showed
couple of warnings about whitespace (tabs instead of spaces), otherwise
recipe is good.
See above.
Whoa. Seems oe-stylise needs some tuning - e.g. ability to warn about
missing variables, etc. In fact I would like to see something like
kernel's checkpatch.pl - script that will show recipe's errors.
Hmm, guess it's time for me to learn python.
BTW, links to closed OE bugs (if applicable) would be nice addition to
changelog.
Done.
Thanks!
Regards
Gordon
--
Best regards,
Dmitry MAD Artamonow
___
Angstrom-distro-devel mailing list
Angstrom-distro-devel@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
[Angstrom-devel] Does .stable branch updates via monotone?
Playing with .stable branch recently I found that it doesn't updates for me anymore and I don't get recent commits via monotone. Doing mtn pull; mtn update shows this: [EMAIL PROTECTED] ~/OE/org.openembedded.stable $ mtn pull ; mtn update mtn: doing anonymous pull; use -kKEYNAME if you need authentication mtn: connecting to monotone.openembedded.org mtn: finding items to synchronize: mtn: certificates | keys | revisions mtn:78772 | 76 | 25951 mtn: bytes in | bytes out | certs in | revs in mtn: 20.7 k |15.4 k |12/12 | 3/3 mtn: successful exchange with monotone.openembedded.org mtn: updating along branch 'org.openembedded.stable' mtn: already up to date at 97f6800e7228874b1f13f700a43a88bb5c10a4bb Looks for me like something was broken while migrating to git. There's no problem for me to switch to git either, but my main concern is about Angstrom autobuilder scripts - was they adapted to use git already? If not, then we may get -r19 release looking exactly like -r18 :) Also, if translating .stable branch from git to mtn won't get fixed, Building Angstrom page also will need update. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] Current .dev branch status?
On 13:04 Tue 19 Aug , colin wrote: Dmitry Artamonow [EMAIL PROTECTED] writes: Hi, I found that correcting the line in /etc/X11/Xinit.d/99gpe-login from: /usr/X11R6/bin/xmodmap to: /usr/X11R6/bin/xmodmap made my keyboard work. This is on Akita Did you mean to /usr/bin/xmodmap ? Yes, it helps (tested on h2200), though I don't understand why - it just defines keycode 135 as Mode_switch. Virtual keyboard now works correctly - at least after log in. On logon screen now it seems to generate something wrong - not all keys give a '*' in password field and backspace doesn't work. Anyway, that's a lot of progress - thank you for the hint! -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] cf gps on h2200 w/ angstrom 2007.12
On 13:19 Sat 09 Aug , Udo van den Heuvel wrote: So there is some overlap/conflict between pxa2xx-uart, 8250 and serial_cs? Could stuff work with pxa2xx-uart set for 5 ports? Or what is the right solution? While googlin' around I found: http://lists.infradead.org/pipermail/linux-pcmcia/2005-July/002383.html (thread starts at http://lists.infradead.org/pipermail/linux-pcmcia/2005-July/002359.html) Got it, and ah. This is where it gets impossible. The PXA driver re-uses the 8250 drivers namespace (major 4 minor 64 upwards, /dev/ttyS0 to ttyS63). Is this still valid? Is there a way to use the serial port on the CF GPS card together with the serial ports on the ipaq h2210? Yes it's still valid and, just FYI, mainline kernel guys still didn't sort out this problem. The problem with Angstrom 2007.12 in fact is two separate problems: 1. Missing parts in defconfigs (not only for h2200, but for other machines with CF slot, which use linux-handhelds kernel) 2. infamous PXA-vs-8250 serial clash. To workaround it linux-handhelds kernel used to have in-tree hack, similar to packages/linux/linux-rp-2.6.24/pxa-serial-hack.patch which Zauruses use, but somewhere between 2.6.21-hh17 and -hh18 it was dropped from the tree, leaving all machines helpless against problem. Soon after that linux-handhelds development cease, and now we have what we have. I already rised attention to the problem back in january: http://projects.linuxtogo.org/pipermail/angstrom-distro-users/2008-January/001349.html Of the 3 choices mentioned (revert to -hh17, reapply hack, use PXA-serial namespace renaming patch to ttySAx) I personally use second. Though it hackish, it has good sides too - no need to fix any software, which depends on PXA-serial ports naming (blueprobe comes to mind, maybe something else?) and it's consistent with zaurus-kernels. See attached patch (made against .stable) for example how I did it for hx4700. If there's no objections from developers, I can expand this patch by adding more machines defconfigs for inclusion into stable branch. And I guess we should push similar patch in .dev too - until mainline kernel guys fix the problem, giving us a direction to go. -- Best regards, Dmitry MAD Artamonow # # old_revision [97f6800e7228874b1f13f700a43a88bb5c10a4bb] # # add_file packages/linux/linux-handhelds-2.6-2.6.21/pxa-serial-hack.patch # content [d2dcca459b4eedfbfed830b558faa47ddac3d212] # # patch packages/linux/linux-handhelds-2.6-2.6.21/hx4700/defconfig # from [50ce9c68d453f6faec44e7d9d1913bb06f76f319] #to [bf3d39a240fa3ffae16248c96cf1d4ceb8a67a84] # # patch packages/linux/linux-handhelds-2.6_2.6.21-hh20.bb # from [b4e447c3efe1d3a370f7b31c7aa9e0af749909a1] #to [0e1c95b9e426f2384d008bd646fa8bc5578a25cb] # --- packages/linux/linux-handhelds-2.6-2.6.21/pxa-serial-hack.patch d2dcca459b4eedfbfed830b558faa47ddac3d212 +++ packages/linux/linux-handhelds-2.6-2.6.21/pxa-serial-hack.patch d2dcca459b4eedfbfed830b558faa47ddac3d212 @@ -0,0 +1,32 @@ +Index: drivers/serial/8250.c +=== +RCS file: /cvs/linux/kernel26/drivers/serial/8250.c,v +retrieving revision 1.28 +retrieving revision 1.27 +diff -u -r1.28 -r1.27 +--- a/drivers/serial/8250.c25 Oct 2007 19:35:43 - 1.28 b/drivers/serial/8250.c7 May 2007 01:37:28 - 1.27 +@@ -86,6 +86,12 @@ + #define CONFIG_SERIAL_MANY_PORTS 1 + #endif + ++#ifdef CONFIG_SERIAL_PXA ++#define SERIAL_HW_SKIP_COUNT CONFIG_SERIAL_PXA_COUNT ++#else ++#define SERIAL_HW_SKIP_COUNT 0 ++#endif ++ + /* + * HUB6 is always on. This will be removed once the header + * files have been cleaned. +@@ -2479,8 +2485,8 @@ + .driver_name= serial, + .dev_name = ttyS, + .major = TTY_MAJOR, +- .minor = 64, +- .name_base = 0, ++ .minor = 64 + SERIAL_HW_SKIP_COUNT,/* skip host internal uarts */ ++ .name_base = SERIAL_HW_SKIP_COUNT, + .nr = UART_NR, + .cons = SERIAL8250_CONSOLE, + }; --- packages/linux/linux-handhelds-2.6-2.6.21/hx4700/defconfig 50ce9c68d453f6faec44e7d9d1913bb06f76f319 +++ packages/linux/linux-handhelds-2.6-2.6.21/hx4700/defconfig bf3d39a240fa3ffae16248c96cf1d4ceb8a67a84 @@ -958,8 +958,8 @@ CONFIG_HW_CONSOLE=y # # Serial drivers # -# CONFIG_SERIAL_8250 is not set - +CONFIG_SERIAL_8250=y +CONFIG_SERIAL_8250_CS=y # # Non-8250 serial port support # --- packages/linux/linux-handhelds-2.6_2.6.21-hh20.bb b4e447c3efe1d3a370f7b31c7aa9e0af749909a1 +++ packages/linux/linux-handhelds-2.6_2.6.21-hh20.bb 0e1c95b9e426f2384d008bd646fa8bc5578a25cb @@ -1,7 +1,7 @@ LICENSE = GPL SECTION =
Re: [Angstrom-devel] suspend and backlight control
On 17:20 Sun 08 Jun , Yan Seiner wrote: Is there any documentation for corgi-bl? I need to be able to set backlight dim and suspend intervals from the command line and I can't find any docs for any of the stuff in /sys/class/backlight and I can't find any way to control the suspend interval via CLI. Can anyone shed some light on this? Hmm, AFAIK auto-dim and auto-turn-off of backlight has nothing to do with kernel. It's the feature of ipaq-sleep - at least when I disabled ipaq-sleep on my C700 (by removing /etc/X11/Xsession.d/69ipaq-sleep), backlight never went off since. I think via /sys/class/backlight you can only set brightness and turn backlight on/off manually. There's also command line utility called 'bl' which can be used for same purposes. -- Best regards, Dmitry MAD Artamonov ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
Re: [Angstrom-devel] [Familiar] Gpsdrive on iPAQ
On 13:41 Mon 17 Dec , Paul Sokolovsky wrote: Hello Andrew, Sunday, December 16, 2007, 5:45:13 PM, you wrote: Even though gpsdrive has been written with a PDA mode in mind there are a number of problems with it that I found annoying. This was mainly due to the fact that it is slow on iPAQs due to the amount of floating point maths that it performs. The user interface also doesn't really fit on to the screen well with the status tab in particular being unreadable in places. The program also contains a large amount of options which don't make sense on a PDA. Eventually these problems became too much for me so I tackled the gpsdrive source code and removed or fixed what I saw as the problems with the PDA mode as well as adding some new features. The program now uses a lot less CPU time (far fewer floating point operations), consumes less memory (executable size is halved) and is easier to use (GUI spread across more tabs). (Since I made these changes a new version of gpsdrive has been released. I have not tried this in PDA mode, but it doesn't claim to solve any of these problems.) Since this is a GPL program and other people may also want to try out my improvements I have made them available at the address below. There is original source code, modified source code and a familiar ipkg compiled with Familiar 0.8.4 on my 3800 series iPAQ. http://www.gedanken.demon.co.uk/gpsdrive-ipaq/ [ skip ] That said, I cross-post this to a more active distro's list and to people I know working on mapping solutions, maybe they will be able to make some use of your work. After looking at screenshots I realised that this program is something I longed for. Sadly, there was only ipk for Familiar-0.8.4 (that old distro with OABI and non-working keys on h2200) on project homepage. So I started fiddling with OE and bitbake for making this beast compile for Ångström, and after a lot of tries finally made my first ever working bitbake recipe. It may be not very good, as I'm still learning, but it WorksForMe(tm). It runs fine on my h2200, though I not tested it in real life yet :) I submitted recipe and needed patch to OE bugzilla: http://bugs.openembedded.org/show_bug.cgi?id=3679 Hope it will be useful for someone. -- Best regards, Dmitry 'MAD' Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
[Angstrom-devel] Problems with screen blanking on h2200
Sorry for probably off-topic, but [EMAIL PROTECTED] seems to be long time dead, and I don't know where else I can post kernel-related problems. I would be glad if someone can clarify current situation with handhelds kernel developement. Anyway, here's my story. Recently I've been playing with Gentoo on my h2200. After compiling xorg-server-1.4 (full-blown, not the kdrive version), I tried to start it - but without success. Screen flickered, turned black, then flickered again and slowly turned white instead of showing X cursor. X server itself didn't show any error in console or logs and worked until ^C was pressed. But even after ^C screen never returned to normal state until reset. I tried to run X under strace and noticed this calls: ioctl(6, FBIOBLANK, 0x1)= 0 and then near the end of strace log: ioctl(6, FBIOBLANK, 0) = 0 This seemed to be the cause, because I already experienced similar problems with blanking in plain console and was pressed to workaround it with setterm -blank 0. Simple testcase program in C proved it, and also I found that bug can be triggered from command line by following statements: echo 1 /sys/class/graphics/fb0/blank echo 0 /sys/class/graphics/fb0/blank For assuring myself what it's not the fault of my custom kernel config, I checked also with LiveRam Angstrom image (20070922) and successfully triggered this bug from command line. Then I dug into kernel code, but since I'm not a kernel hacker at all, the only thing I tried was commenting out mq1100fb_power call from mq1100fb_blank function in drivers/video/mq1100fb.c : -- diff -u -r1.45 mq1100fb.c --- drivers/video/mq1100fb.c18 Jul 2007 19:21:42 - 1.45 +++ drivers/video/mq1100fb.c21 Oct 2007 08:25:30 - @@ -318,7 +318,7 @@ *blank_mode == 4: powerdown */ - mq1100fb_power (info, blank_mode); +/* mq1100fb_power (info, blank_mode); */ return 0; } -- Suddenly it turned out to be a solution - screen was blanking anyway (seemed that it's somehow handled by drivers/video/backlight/h2200_lcd.c driver (and maybe backlight driver too)), but on unblanking it has all of its contents intact and X finally worked. Still I can't understand how it works and how screen blanking differs from device suspending which seemed to call the same mq1100fb_power without any problems. -- Best regards, Dmitry MAD Artamonow ___ Angstrom-distro-devel mailing list Angstrom-distro-devel@linuxtogo.org http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-devel
