2011/11/14/#ubuntu-arm.txt

=== ericm|ubuntu is now known as ericm-afk
=== ericm-afk is now known as ericm
[13:50] <suihkulokki> oneiric firefox, when being compiled on a machine with linux 3.0 kernel, will fail to compile with:
[13:50] <suihkulokki> ../coreconf/config.mk:71: ../coreconf/Linux3.0.mk: No such file or directory
[13:50] <suihkulokki> fun :)
[13:50] <ogra_> oh, could you file a bug, our buildds all use old kernels
[13:51] <ogra_> (yet)
[13:51] <suihkulokki> looking at it right now
[13:52] <ogra_> we have some pandas but that are currently all assigned to armhf porting iirc
[13:52] <ogra_> so trhey havent seen any higher level builds like desktop packages yet
[13:52] <ogra_> it is expected that we drop all but panda soon though
[14:10] <punxos> Hi
[14:15] <punxos> How can "remove" a jump to a tag in a macro ??I to change this code --> http://pastebin.com/CrkXZZJK (tag already defined obvius)
[14:20] <dmart> punxos: what are you trying to do exactly?  I don't understand from your code
[14:21] <punxos> this code is inside a macro, so I can't create a tag because the compiler fail due to tag is redefined
[14:21] <brendand> is it still correct to dd the 11.10 pandaboard image to the SD card? or is a different method used now?
[14:23] <ogra_> brendand, neither the images nor the method to write tham changed
[14:23] <ogra_> *them
[14:23] <ogra_> its the same as 11.04
[14:24] <brendand> ogra_ - ok, thanks
[14:24] <dmart> punxos: so, where you say "tag", you mean "label
[14:26] <dmart> punxos: Is this as assembler macro, or are you working with C inline assembler?
[14:27] <punxos> assembler macro
[14:28] <punxos> dmart: I found the solution: add $ to label
[14:28] <punxos> thx anyway
[14:30] <dmart> punxos: "dollar local labels" don't work on arm -- though if you have some code where they do appear to work, I'd like to see it
[14:30] <dmart> 1$
[14:30] <dmart> ->
[14:30] <dmart> Error: junk at end of line, first unrecognized character is `1'
[14:31] <punxos> dmart: beq $label
[14:31] <punxos> $label: some code
[14:32] <punxos> dmart: sorry not work :(
[14:32] <dmart> $ is allowed in symbol names on arm -- you haven't really solved the problem there; you just have a different label name when you ass $
[14:32] <punxos> I dint' do enough checks
[14:33] <punxos> yes yes
[14:33] <punxos> sorry
[14:33] <ogra_> .oO( arm is a british company, try a pound sign instead of a dollar one ;) )
[14:33] <punxos> xD
[14:33] <punxos> dmart: this example code-> http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.kui0100a/armasm_cegceijg.htm
[14:34]  * dmart doesn't know if UTF-8 is allows in symbol names for £ :P
[14:34] <ogra_> hehe
[14:35] <punxos> :D
[14:35] <dmart> punxos: are you working with the ARM RVCT assembler (armasm) or GNU as?  The syntax for macros is completely different; the example is for armasm
[14:35] <punxos> arm-none-eabi-
[14:36] <dmart> OK, so GNU as then
[14:39] <dmart> punxos: porting that example to GNU as might look something like this: http://pastebin.ubuntu.com/738238/
[14:40] <punxos> dmart: thanks
[14:40] <dmart> But I suspect that doesn't quite solve your problem.  Can you paste your actual source code so I can see what you're trying to do?
[14:41] <punxos> yes 1 sec
[14:42] <punxos> dmart:
[14:43] <punxos> http://pastebin.ubuntu.com/738241/
[14:43] <punxos> just update a register if r0 is 0
[14:47] <punxos> dmart: but this macro is inside other macro due to that your example not work
[14:50] <Neko> hello ubuntu guys!
[14:50] <Neko> does anyone know of a particular brand/model of SD card that actually supports erase/secure erase?
[14:53] <gildean> yes, a drill will securely erase any sd-card
[14:53] <gildean> the bigger the drillbit the better
[14:58] <Neko> not useful :D
[15:03] <dmart> punxos: in this case, you could avoid the branch using conditional assembly
[15:04] <punxos> mm I dont' know how
[15:04] <punxos> a short sample please ?
[15:08] <dmart> punxos: here are some examples of things you can do:
[15:08] <dmart> http://pastebin.ubuntu.com/738263/
[15:08] <dmart> The GNU assembler documentation is a bit vague about some acpects of macros, but it is worth reading
[15:09] <dmart> http://sourceware.org/binutils/docs-2.21/as/Macro.html#Macro
[15:09] <punxos> dmart: oooo thanks very much ! great job
[15:12] <dmart> hope it helps
[15:12] <punxos> yes it's perfect :)
[15:12] <punxos> works perfectly
[15:13] <dmart> which approach did you choose?
[15:13] <punxos> beq 1f
[15:13] <punxos> 1:
[15:14] <punxos> it's more visual for me
[15:15] <dmart> That's a good simple solution, and works in a lot of cases.  If your code is performance-critical, you might still want to use conditional instructions; though it really depends on how reliably that branch can be predicted (and the sophistication of branch prediction on the CPU(s) you're targeting)
[15:23] <robclark> note that if you are building thumb2 (or you want to support thumb2) then see IT instruction instead..
=== Quintasan_ is now known as Quintasan
[15:39] <antoniodariush> Hi
[15:40] <antoniodariush> I have a beagleboard, can I install ubuntu on it ?
[15:41] <antoniodariush> well i guess so :)
[15:42] <ogra_> antoniodariush, depends what version of beagle and which flavour of ubuntu
[15:44] <antoniodariush> I have a project but running on nanonote so i need to cross compile and use toolchains everytime to compile the software
[15:44] <ogra_> (you cant easily run ubuntu-desktop on a beagle C4 ... not enough ram, but it will run ubuntu-server fine for example)
[15:45] <antoniodariush> as long I can avoid cross compilation I am fine with any flavour of ubuntu
[15:45] <ogra_> right, use -server for this
[15:46] <antoniodariush> I have a beagleboard-xM
[15:47] <antoniodariush> ok i'll have a look thanks
[15:48] <ogra_> even with Xm you want -server unless you actually want to run and use desktop apps ... if its for compiling, -server is the best chouce to not waste resources
[15:48] <ogra_> *choice
[15:51] <antoniodariush> yeh that's true
[15:51] <punxos> dmart: thanks for the info
[15:53] <dmart> punxos: no problem -- glad it was helpful
[17:46] <punxos> If a irq is raised when the user is in destination instruction jump. if you do "subs pc.lr,#4" maybe is not true (if the las jump was done). Any can explain me that ?
[17:57] <slangasek> ogra_, NCommander: https://blueprints.launchpad.net/ubuntu/+spec/ubuntu-arm-p-optional-initrd had work items, and now it doesn't - what's the story there? :)
[17:58] <ogra_> slangasek, NCommander just blindly pasted notes in, while i'm working on the spec atm
[17:58] <ogra_> (I'm the drafter, ask him why he blindly pastes stuff without talking to me)
[17:59] <ogra_> slangasek, apw, tim and me had another long session in the hallway later, actually looking at code ... the kernel side changed a bit
[17:59] <ogra_> so i'm rewriting the stuff atm
[18:01] <slangasek> hmm
[18:01] <ogra_> slangasek, i'll finish it by tomorrow, do you want any specific person assigned to the foundations WIs ?
[18:02] <ogra_> ro should i  just put the team in and leave it to you
[18:02] <ogra_> *or
[18:02] <slangasek> the latter is fine
[18:03] <slangasek> many of these are probably taken by me personally
[18:03] <ogra_> k
[18:11]  * ogra_ smiles seeing libpam-sshauth in the archive 
[18:29] <punxos> Hi
[19:11] <int_ua> Was there an answer to the OMAP3 question? Was it dropped for 12.04?
[19:11] <ogra_> no, it wasnt
[19:11] <ogra_> (what was the question though ?)
[19:17] <GrueMaster> int_ua: The only real question is whether or not to declare no-support for ubuntu-desktop on omap3 based systems with <512M memory.  Currently, running ubuntu-desktop on a Beagle C4 is extremely painful due to swappiness.
[19:18] <GrueMaster> It takes me over an hour to do any release testing on it alone (other platforms can be smoke tested in 20 minutes).
[19:26] <int_ua> GrueMaster: I'm running Ubuntu Core on the N900 from class 6 microSD and I wouldn't say that swappiness is a serious problem.
[19:27] <GrueMaster> Ubuntu-core is not Ubuntu-Desktop (with unity-2d, etc).
[19:27] <int_ua> GrueMaster: are there daily releases of the Core?
[19:28] <GrueMaster> Only when core packages change afaik.
[19:28] <GrueMaster> We may have scaled back image building to 3/week due to armHF bring up.
[19:29] <int_ua> Where are they published? There are no ARM: http://cdimage.ubuntu.com/ubuntu-core/daily/current/
[19:30] <GrueMaster> looks like daily/20111112 is the most recent for armel.  We'll need to make sure that they don't get deleted.
[19:30] <int_ua> GrueMaster: I don't really care getting it _daily_. I care about their existing :)
[19:30] <GrueMaster> Yes, agreed.
[19:31] <GrueMaster> infinity: ^^^
[19:31] <GrueMaster> Any comments?
[19:32] <infinity> Hrm.  Curious.
[19:32] <infinity> May be an accidental oops from the cdimage cutover.
[19:32] <infinity> Or something.
[19:32] <int_ua> P.S. I've posted a question also: http://askubuntu.com/questions/78875/was-omap3-support-dropped-for-12-04
[19:32] <infinity> Although, I see armel on the 12th.
[19:32] <GrueMaster> int_ua: omap3 will not be dropped.
[19:32] <infinity> http://cdimage.ubuntu.com/ubuntu-core/daily/20111112/
[19:32] <int_ua> 20111112 is the last daily armel on every palce
[19:33] <int_ua> *place
[19:33] <GrueMaster> We went over the question of omap3.
[19:33] <int_ua> GrueMaster: Thanks :)
[19:33] <GrueMaster> I will still be actively testing on beagleXM (512M), and am looking to get ubuntu-desktop running on the Nook Color this cycle.
[19:34] <infinity> int_ua: So it is.  We need to hunt that down, thanks.
[19:34] <GrueMaster> I would guess possible pool-churn (again).
[19:37] <infinity> GrueMaster: No.
[19:37] <infinity> GrueMaster: There aren't even failed build attempts.
[19:37] <infinity> GrueMaster: I'll look into it.
[19:37] <GrueMaster> hrm.  That is odd.
[19:39] <infinity> And more curiously, there are builds on the livefs builders...
[19:40] <GrueMaster> cdimage full (again)?
[19:42] <infinity> no.
[19:42] <infinity> New machine.
[19:42] <infinity> Also, ubuntu-server built.
[19:42] <infinity> And ubuntu.
[19:42] <infinity> Nevermind my claims that nothing was building.
[19:43] <infinity> Looks like only core is missing.  And logs are missing, but that's a different issue entirely.
[19:44] <infinity> Oh!
[19:44] <infinity> annonaceae is dead.
[19:44] <infinity> I think.
[19:44] <int_ua> infinity: kubuntu-netbook and kubuntu-mobile are missing too
[19:45] <infinity> int_ua: kubuntu-netbook is dead.
[19:45] <GrueMaster> Hrm.  That is a beagleXM buildd.
[19:45] <infinity> int_ua: kubuntu-mobile is half there.
[19:45] <infinity> int_ua: (basically, all omap3, mx5, and the -core images are missing, because their buildd went kaput)
[19:45] <int_ua> infinity: s/kubuntu-netbook/kubuntu/
[19:45] <GrueMaster> All the *ceae buildds are beagleXMs from linaro iirc.
[19:45]  * infinity chases up with IS now.
[19:52] <infinity> Okay, annonaceaeaaeaaeee fixed.  Dailies should look more normal tomorrow.
[20:01] <GrueMaster> heh
[20:02] <int_ua> Thanks a lot :)
[20:04] <int_ua> Is Annonaceae a server name?)
[20:05] <GrueMaster> No, it is one of many buildds in our pool.  It is a beagleXM platform.
[20:05] <GrueMaster> We build the entire armel pool (arch=armel/armhf) on actual hardware.
[20:23] <infinity> int_ua: "server" is a pretty liberal use of the word, but it's a hostname, yes.
=== sauerbraten_ is now known as sauerbraten