Join the chat at beagleboard.org/chat
IRC Log for 2008-10-11
Timestamps are in UTC.
- [00:07:24] * docelic (n=docelic@78.134.202.188) Quit ("http://www.spinlocksolutions.com/")
- [00:20:15] * gduncan (n=spec@netblock-208-127-123-95.dslextreme.com) Quit ()
- [00:22:05] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [00:23:01] * scary (n=scary@h-74-1-188-242.snfccasy.covad.net) Quit (Remote closed the connection)
- [00:32:34] * jrmuizel (n=jrmuizel@CPE001f5be79d0f-CM0017ee62f8b0.cpe.net.cable.rogers.com) has joined #beagle
- [00:38:30] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Connection timed out)
- [00:39:37] * Fl1pFl0p1 (n=Fl1pFl0p@146.244.179.234) Quit ("Leaving.")
- [00:43:17] <mru>
anyone had any luck with gcc-csl 2008q3?
- [00:43:54] * TAK2004 (n=Administ@dslb-088-072-220-107.pools.arcor-ip.net) Quit (Read error: 104 (Connection reset by peer))
- [01:21:59] * jrmuizel (n=jrmuizel@CPE001f5be79d0f-CM0017ee62f8b0.cpe.net.cable.rogers.com) Quit ()
- [01:22:09] * BThompson (n=BThompso@cpe-76-185-93-11.tx.res.rr.com) Quit (Read error: 110 (Connection timed out))
- [01:24:29] * gcohler (n=gcohler@pool-72-93-108-183.bstnma.fios.verizon.net) has joined #beagle
- [01:27:38] * gcohler (n=gcohler@pool-72-93-108-183.bstnma.fios.verizon.net) Quit (Remote closed the connection)
- [01:42:33] * gcohler (n=gcohler@pool-72-93-108-183.bstnma.fios.verizon.net) has joined #beagle
- [01:46:30] * Openfree (n=df@222.65.132.11) Quit (Remote closed the connection)
- [01:49:54] * khasim (n=a0393720@192.163.20.231) has joined #beagle
- [01:52:47] * gcohler_ (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) has joined #beagle
- [01:52:53] * slothlove (n=srussell@174-146-140-238.area3.spcsdns.net) has joined #beagle
- [01:54:20] * gcohler_ (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) has left #beagle
- [01:55:27] * john280z (n=johnm@user-0c2h5av.cable.mindspring.com) Quit ("Ex-Chat")
- [02:04:25] * gcohler (n=gcohler@pool-72-93-108-183.bstnma.fios.verizon.net) Quit (Read error: 110 (Connection timed out))
- [02:14:06] * khasim (n=a0393720@192.163.20.231) Quit (Remote closed the connection)
- [02:20:44] * Olipro_ is now known as Olipro
- [02:53:46] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Read error: 104 (Connection reset by peer))
- [02:54:34] * Olipro (i=Olipro@uncyclopedia/Olipro) has joined #beagle
- [03:24:27] * slothlove (n=srussell@174-146-140-238.area3.spcsdns.net) Quit (Read error: 113 (No route to host))
- [03:26:46] * slothlove (n=srussell@174-146-140-238.area3.spcsdns.net) has joined #beagle
- [04:50:37] * slothlove (n=srussell@174-146-140-238.area3.spcsdns.net) Quit (Read error: 113 (No route to host))
- [05:26:14] * thephilosopher (n=chatzill@modemcable108.47-58-74.mc.videotron.ca) Quit (Read error: 110 (Connection timed out))
- [06:14:43] * felipec (n=felipec@a88-113-6-167.elisa-laajakaista.fi) Quit (Read error: 110 (Connection timed out))
- [06:33:28] * slothlove (n=srussell@ip98-169-198-218.dc.dc.cox.net) has joined #beagle
- [07:02:16] * slothlove (n=srussell@ip98-169-198-218.dc.dc.cox.net) Quit (Read error: 110 (Connection timed out))
- [07:15:32] * Xenion (n=robert@p579FCD68.dip.t-dialin.net) has joined #beagle
- [07:17:09] <Xenion>
Guuuuuuuuuuten Morgen !!! :-) | Goooooood Morning
- [07:21:14] * Openfree (n=df@222.65.132.11) has joined #beagle
- [07:23:28] * lukketto (n=luca@host77-6-dynamic.7-87-r.retail.telecomitalia.it) has joined #beagle
- [07:44:18] * lukketto (n=luca@host77-6-dynamic.7-87-r.retail.telecomitalia.it) Quit (No route to host)
- [07:45:21] * felipec (n=felipec@a88-113-6-167.elisa-laajakaista.fi) has joined #beagle
- [08:02:49] * kulve (i=kulve@xdsl-179-201.nblnetworks.fi) has joined #beagle
- [08:04:07] * hli (i=chaton@vig91-2-82-232-97-149.fbx.proxad.net) has joined #beagle
- [08:33:20] * methril|laptop (n=methril@224.72.220.87.dynamic.jazztel.es) has joined #beagle
- [08:34:58] * zedstar (n=john@fsf/member/zedstar) has joined #beagle
- [08:37:24] * Fl1pFl0p (n=Fl1pFl0p@ip68-8-230-62.sd.sd.cox.net) has joined #beagle
- [08:38:27] <[X]Spot>
how can I copy and install application on ANgstrom distro, after building it with : bitbake package (OE) ?
- [08:43:44] * methril|laptop (n=methril@224.72.220.87.dynamic.jazztel.es) Quit ("Ex-Chat")
- [08:47:14] <zedstar>
scp it over and opkg install it
- [09:17:22] * valhalla (n=valhalla@81-174-25-222.dynamic.ngi.it) has joined #beagle
- [09:23:45] * khasim (n=a0393720@192.163.20.231) has joined #beagle
- [09:28:18] <[X]Spot>
zedstar but where is the build package ?
- [09:30:52] * jes2 (n=jess@117.195.131.244) has joined #beagle
- [09:34:32] <zedstar>
xspot something like <whatever>/tmp/deploy/glibc/ipk/armv7a/ ?
- [09:35:52] <zedstar>
or do a find for the name*.ipk in your build area
- [09:40:35] <[X]Spot>
zedstar bitbake package.. : download, configure, compile and build the package
- [09:40:54] <[X]Spot>
how can I configure the package with my settings ?
- [09:41:47] <zedstar>
are you asking how to write a recipe?
- [09:54:44] * guillaum1 (n=Guillaum@AMontsouris-153-1-38-79.w90-2.abo.wanadoo.fr) has joined #beagle
- [09:56:57] * jes2 (n=jess@117.195.131.244) Quit ("Leaving.")
- [09:58:04] * Xenion (n=robert@p579FCD68.dip.t-dialin.net) Quit (Read error: 110 (Connection timed out))
- [09:58:24] * Xenion (n=robert@p579FC994.dip.t-dialin.net) has joined #beagle
- [10:09:30] * docelic (n=docelic@78.134.193.37) has joined #beagle
- [10:14:33] * Xenion (n=robert@p579FC994.dip.t-dialin.net) Quit ("Verlassend")
- [10:22:17] * khasim (n=a0393720@192.163.20.231) Quit (Remote closed the connection)
- [11:16:46] * lukketto (n=luca@host77-6-dynamic.7-87-r.retail.telecomitalia.it) has joined #beagle
- [11:23:34] <mru>
morning
- [11:30:45] <felipec>
morning mru
- [11:31:19] <mru>
I've been trying to use gcc-csl 2008q3
- [11:31:25] <mru>
so far, it miscompiles ffmpeg
- [11:37:33] <felipec>
mru: the neon optimized stuff?
- [11:37:37] * jbenedetto (n=jbenedet@c-24-131-229-153.hsd1.pa.comcast.net) has joined #beagle
- [11:37:42] <mru>
plain C code
- [11:37:54] <mru>
the neon optimisations are written in pure assembler
- [11:38:00] <mru>
so gcc doesn't get to mess with those
- [11:38:32] <felipec>
mru: hmm, weird, I haven't had any problems... but I haven't compiled much with it
- [11:38:43] <mru>
ffmpeg is notorious for breaking compilers
- [11:46:56] * lukketto (n=luca@host77-6-dynamic.7-87-r.retail.telecomitalia.it) Quit (No route to host)
- [11:47:34] * kulve (i=kulve@xdsl-179-201.nblnetworks.fi) Quit ("leaving")
- [11:56:43] * zuh (i=1a0b5dbf@emperor.pingu.fi) has joined #beagle
- [11:58:23] * feig (n=ejf3@c-76-118-153-30.hsd1.ma.comcast.net) has joined #beagle
- [11:58:44] * kulve (i=a729a748@emperor.pingu.fi) has joined #beagle
- [12:22:35] <mru>
oh dear, 2008q3 can't even compile a trivial copy loop correctly
- [12:23:04] <mru>
for (i = 0; i < len; i++) dst[i] = src[i];
- [12:23:56] <kulve>
their 2008qX toolchains haven't really succeeded..
- [12:24:00] * woski (n=dleme@65.182.51.67) Quit (Read error: 104 (Connection reset by peer))
- [12:24:16] <mru>
seems -ftree-vectorize is on by default
- [12:24:20] <mru>
and it's oh so broken
- [12:24:36] * woski (n=dleme@65.182.51.67) has joined #beagle
- [12:25:03] <mru>
it tries to use neon to load/store 8 bytes at a time
- [12:25:16] * Xenion (n=robert@p579FC994.dip.t-dialin.net) has joined #beagle
- [12:25:23] <mru>
this fails if src and dst are not both 8-byte aligned
- [12:25:29] <kulve>
"Improved support for NEON and, in particular, auto-vectorization using
- [12:25:29] <kulve>
NEON."
- [12:25:38] <mru>
rotfl
- [12:26:39] <mru>
it's back to 2007q3 here
- [12:27:36] <felipec>
mrc3: 2008q1 doesn't work correctly either?
- [12:27:42] <mru>
never tried it
- [12:27:43] <kulve>
yeah, it doesn't
- [12:27:47] <mru>
everybody said it was broken
- [12:28:01] <felipec>
that's what I've been using
- [12:28:19] <mru>
brave man, and lucky
- [12:28:27] <kulve>
felipec: people say that it compiles broken NEON
- [12:28:48] <mru>
if 2008q3 does broken neon, I wouldn't expect better from 2008q1
- [12:29:08] <felipec>
kulve: I'm not compiling NEON stuff
- [12:29:20] <felipec>
ah, maybe that's why I couldn't compile omapfbplay
- [12:29:22] <mru>
-mfpu=neon is enough to trigger it
- [12:30:51] * woski (n=dleme@65.182.51.67) Quit (Read error: 104 (Connection reset by peer))
- [12:31:01] * Xenion (n=robert@p579FC994.dip.t-dialin.net) Quit ("Verlassend")
- [12:31:18] * Xenion (n=robert@p579FC994.dip.t-dialin.net) has joined #beagle
- [12:31:24] * woski (n=dleme@65.182.51.67) has joined #beagle
- [12:32:06] <felipec>
mru: how about sending a mail to the cs ml?
- [12:35:00] <mru>
won't they just tell me to buy the supported version or shut up?
- [12:35:13] <pbrook>
probably :-)
- [12:35:52] <pbrook>
ICBW, but I though someone at ARM was supposed to be testing these things.
- [12:36:06] <mru>
I'll notify ARM
- [12:36:14] <mru>
they usually listen
- [12:36:36] <kulve>
mru: who do you notify at ARM?
- [12:36:59] <pbrook>
They (and all our paying customers) get a 1-month preview release so they can shout and get things fixed before the final public release goes out.
- [12:38:34] <mru>
kulve: where did you find that bit you quoted above?
- [12:38:45] <kulve>
http://www.codesourcery.com/archives/arm-gnu-announce/msg00024.html
- [12:39:00] <mru>
pbrook: are you with codesourcery?
- [12:39:06] <pbrook>
Yes
- [12:39:25] <kulve>
mru: and something similar in http://www.codesourcery.com/gnu_toolchains/arm
- [12:39:31] * pbrook was the release manager for all but the last release.
- [12:39:34] <kulve>
"Better NEON vectorization!"
- [12:39:41] <mru>
yeah, I saw that
- [12:39:53] <mru>
maybe it is better
- [12:39:58] <mru>
just not yet working
- [12:43:21] <pbrook>
If noone reports bugs, we can't fix it.
- [12:43:31] <mru>
consider it reported
- [12:43:47] <mru>
and people have been complaining about 2008q1 for months
- [12:43:49] <pbrook>
"ffmpeg broke" isn't what I consider to be a decent bugreport.
- [12:44:17] <pbrook>
We don't do updates to the lite releases. Most those bugs were fixed in our product toolcahins fairly quickly.
- [12:44:39] <kulve>
mru: back to old topic. Did you notice my question yesterday (15:54 UTC) about scaling with omapfb?
- [12:44:59] <mru>
pbrook: and you've still not fixed them in the lite release, it would seem
- [12:45:19] <mru>
what's the point in publishing known-broken compilers?
- [12:45:29] <mru>
do you think it will make people more willing to pay?
- [12:45:32] <pbrook>
mru: We aren't aware of any open bugs.
- [12:45:37] <mru>
now you are
- [12:45:40] <pbrook>
No.
- [12:45:47] <mru>
I just fucking told you
- [12:45:56] <pbrook>
You said that ffmpeg is busted.
- [12:45:58] <mru>
it can't compile a bloody copy loop
- [12:46:03] <mru>
scroll up a few pages
- [12:46:15] <mru>
ffmpeg is fine, the compiler is busted
- [12:46:25] <kulve>
15:25 < mru> it tries to use neon to load/store 8 bytes at a time
- [12:46:25] <kulve>
15:25 < mru> this fails if src and dst are not both 8-byte aligned
- [12:46:33] <pbrook>
Well, I'm pretty sure our test work ok.
- [12:46:44] <mru>
see, no point reporting bugs
- [12:46:49] <mru>
"our tests work"
- [12:46:50] <mru>
bah
- [12:47:06] <pbrook>
Any I'm pretty sure our tests will over the code you quoted.
- [12:47:06] <mru>
what did you test? produces any output at all?
- [12:47:24] <mru>
you must be wrong then
- [12:47:29] <mru>
because it *does* fail here
- [12:52:40] <felipec>
mru: not just cs is on the ml, if you can send code that doesn't compile correctly other people will be interested
- [12:55:06] <felipec>
and my guess is that cs would be interested in fixing it too
- [12:55:35] <pbrook>
mru: I just tried your code, and it looks fine. gcc aligns the source pointer, and uses vst1 for the destination store.
- [12:55:48] <mru>
it's the load that fails
- [12:56:01] <mru>
it uses vldm there
- [12:56:06] <pbrook>
Yes.
- [12:56:15] <mru>
and vldm requires strict alignment
- [12:56:22] <pbrook>
Yes.
- [12:56:27] <mru>
and it doesn't check it
- [12:57:14] <pbrook>
Mine certainly does.
- [12:57:34] <mru>
maybe you're using the pay-for version
- [12:57:39] <pbrook>
It peels sufficient iterations of the loop that the inner vector loop is always known to be aligned.
- [12:57:52] <pbrook>
No. I'm using the 2008q3 binaries.
- [13:01:02] <mru>
well, explain why it doesn't work
- [13:01:09] <mru>
did you try running the generated code?
- [13:01:14] <pbrook>
Not yet.
- [13:01:23] <mru>
I'll post a full example in a bit
- [13:07:33] * tulpe (n=ask@p54AF6376.dip.t-dialin.net) has joined #beagle
- [13:13:42] <tulpe>
hi.. i got angstrom running. what's a good wlan stick that runs with the board?
- [13:16:00] <mru>
pbrook: ok, here's a test case: http://pastebin.com/m186bc0a9 http://pastebin.com/m7f85a7a1
- [13:16:15] <mru>
save those in separate files, and compile/link it
- [13:17:24] <mru>
if they're in the same file, gcc knows too much, and optimises out the bad code
- [13:21:39] <felipec>
mru: which flags?
- [13:22:11] <mru>
-fmpu=neon -mfloat-abi=softfp -mcpu=cortex-a8 -O3
- [13:22:14] <pbrook>
Hum, that's interesting.
- [13:22:29] <pbrook>
Looks like it might be aligning the wrong argument.
- [13:22:34] <mru>
heh
- [13:22:53] <Crofton|work>
don't tell him he's right, he be impossible :)
- [13:23:03] <Crofton|work>
he'll be that is
- [13:34:56] * lukketto (n=luca@host86-192-dynamic.7-87-r.retail.telecomitalia.it) has joined #beagle
- [13:55:28] <mru>
pbrook: does it break for you too?
- [13:59:14] <jkridner>
good morning all
- [13:59:35] <mru>
morning jkridner
- [13:59:39] <Xenion>
hello jkridner
- [13:59:55] <mru>
we were just discussing gcc-csl, and how it's still broken
- [14:00:11] <pbrook>
Yeah, I just submitted an internal bugreport.
- [14:00:31] <mru>
pbrook: thanks
- [14:00:37] <pbrook>
Our test environment apparently doesn't enforce alignment requirements.
- [14:00:45] <mru>
what do you test it on?
- [14:00:51] <pbrook>
Mostly qemu
- [14:00:57] <mru>
use real hardware
- [14:01:00] <mru>
much more reliable
- [14:01:03] <pbrook>
Real hardware sucks.
- [14:01:16] <pbrook>
And for the most part doesn't really exist yet.
- [14:01:40] <mru>
cortex-a8 hardware is readily available
- [14:01:48] <pbrook>
Barely.
- [14:01:54] <pbrook>
Real hardware tends to be horribly unreliable.
- [14:02:19] <pbrook>
Beagleboard kenrels are certainly fairly hit-and-miss for any sort of serious use.
- [14:02:54] <mru>
more reliable than your compiler at least
- [14:03:57] <pbrook>
In my experience beagleboard USB isn't.
- [14:04:25] <mru>
works for me
- [14:04:33] <mru>
and how does that relate to compilers anyway?
- [14:04:51] <pbrook>
You were claiming that real hardware was more reliable than out compilers.
- [14:04:57] <mru>
it is
- [14:05:05] <mru>
the beagle board works as advertised
- [14:05:08] <mru>
your compiler doesn't
- [14:05:12] <pbrook>
Rubbish.
- [14:05:43] <mru>
are you telling me I'm dreaming, and that it didn't just produce bad code for a trivial copy loop?
- [14:06:01] <hli>
is the new codesourcery gcc version more reliable ?
- [14:06:09] <mru>
that's the one we're talking about
- [14:06:57] <mru>
if it's better than the last one, I dread to think what that was like
- [14:07:07] <hli>
so there are still a lot of bugs :/ ?
- [14:07:24] <pbrook>
Are you telling me I'm dreaming, and it doesn't fall over in a big heap every time I plug in a bluetoot dongle, SVIDEO out plain doesn't work, video mode selection is not hardcoded, and audio craps on on a good proportion of kenrel?
- [14:07:29] <mru>
hli: http://hardwarebug.org/2008/10/11/codesourcery-gcc-2008q3-fail/
- [14:07:54] <mru>
pbrook: why would you need any of those to test a compiler?
- [14:08:01] <pbrook>
USB, yes.
- [14:08:06] <mru>
you messed up, and you might as well admit it
- [14:08:25] <mru>
how hard could it be to load up an sd card with test cases?
- [14:08:28] <pbrook>
Sure, it's a bug. Apparently noone else bothered to test the toolchains either though.
- [14:08:52] <pbrook>
Our test runs take several days on real hardware.
- [14:09:04] <mru>
then run them for several days
- [14:09:26] <hli>
it really sucks :/
- [14:09:38] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [14:09:41] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [14:09:42] <pbrook>
Several wekks, if you have to manually shuffle everything back and forth.
- [14:09:58] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [14:10:12] <mru>
I'd happily wait another week or two if that meant getting something that actually works
- [14:10:32] <pbrook>
mru: So buy a subscription, and do it.
- [14:10:56] <pbrook>
The Lite releases are already delayed by a month.
- [14:10:57] <mru>
the priorites are, in my opinion: 1) do it right. 2) do it fast. 3) do it soon
- [14:11:23] <mru>
in commercial environments, those tend to be reversed
- [14:11:52] <mru>
pay for gcc? are you kidding?
- [14:12:07] <pbrook>
You get what you pay for.
- [14:12:22] <jkridner>
someone has to pay for work.
- [14:12:30] <hli>
gcc is GPL, paying for it is a fraud
- [14:12:46] <pbrook>
You paid nothing, so you get the unsupported twice yearly drops.
- [14:12:58] * no1234 (n=masta@p5493C097.dip.t-dialin.net) has joined #beagle
- [14:13:02] <mru>
unsupported is fine, unusuable is not
- [14:13:04] <jkridner>
I think Stallman's idea was that people get paid for the first release.
- [14:13:22] <no1234>
hi guys!
- [14:13:33] <jkridner>
hi no1234, any luck on that serial port?
- [14:13:39] <pbrook>
mru: You're welcome to go build, test, debug and maintain your own gcc from the FSF sources.
- [14:13:41] <jkridner>
I guess your problems were looking more serious.
- [14:14:04] <mru>
pbrook: why do you even bother with the lite releases?
- [14:14:15] <pbrook>
Because ARM pays us to do them.
- [14:14:26] <mru>
ah, so someone *is* paying
- [14:14:41] <mru>
then why don't *they* get what they paid for?
- [14:14:45] <pbrook>
They do.
- [14:14:56] <no1234>
well actually it turned out that all cables and adapter were correct. I also measured the powersupply wich has 5.18V output.
- [14:15:03] <mru>
did they ask for a broken compiler?
- [14:15:36] <pbrook>
We give them a prerelease a month before it's publicly releases. They test and certify that the release is acceptable.
- [14:15:56] <no1234>
But even without anything but the powersupply connected only the pwr led is on...
- [14:15:57] <jkridner>
certain releases may have issues, but you have to agree that CS is improving the ARM Cortex-A8 support in the compiler significantly.
- [14:16:16] <hli>
i bought a BeagleBoard and finally i saw there is a lot of unsupported things. I'm seriously pissed off because I also bought a pandora and i really feared it would be the same for a long time
- [14:16:19] <jkridner>
no1234: is it possible you provided too much voltage at one point and damaged the board?
- [14:16:31] <jkridner>
no1234: 5.5V is the absolute maximum
- [14:16:39] <mru>
jkridner: I can't tell whether they've improved anything, as long as basic functionality is broken
- [14:16:58] * emeb (n=ericb@ip72-223-84-209.ph.ph.cox.net) has joined #beagle
- [14:17:10] <jkridner>
mru: the patches they get upstream have added significant functionality.
- [14:17:15] <no1234>
i only plugged the powersupply shipped from digi-key... also tried the usb method with the same result
- [14:17:45] <mru>
maybe they do some good things too, I can't tell
- [14:17:48] <jkridner>
I have also seen broken basic functions in the lite releases, but I also find them very helpful. at least enough to not be angry.
- [14:17:55] <mru>
but they certainly have an attitude problem
- [14:18:31] <jkridner>
it would be nice if they would do public bugfix releases.
- [14:18:44] <mru>
now that's an understatement
- [14:19:09] <mru>
or even publish a list of known issues
- [14:19:14] <pbrook>
mru: If you mean we ignore random people who aren't our customers whinging on IRC about something we gave away for free, then yes.
- [14:20:01] <no1234>
do you think there is any chance to get the board working?
- [14:21:16] <jkridner>
no1234: it doesn't look very good.
- [14:21:17] <pbrook>
mru: If you believe there are exception circumstances that warrant a Lite rebuild, you need to convince ARM of that.
- [14:21:35] <mru>
fine, I'll see what ARM has to say
- [14:21:35] <jkridner>
no1234: do you have anything besides the power connector connected?
- [14:22:13] <jkridner>
adding FFmpeg to the test suite might be a nice way to go. a lot of people use that.
- [14:22:57] <Crofton|work>
heh
- [14:23:00] <pbrook>
If you want hardware testing on a particular platform and/or a particular application, you'll probably have to find someone to pay for that.
- [14:23:03] <no1234>
the first time i connected the board there was nothing but the power supply, but only pwr led. Then I checked the rs232 connection without success (nothing being send to my terminal, but the loopback test as described in the doc woks)
- [14:23:19] <Crofton|work>
just use OE to build a bunch of stuff and test the resutl :)
- [14:23:32] <no1234>
after that i connected an dvi monitor and sound with the same result...
- [14:23:39] <jkridner>
no1234: so, USR0/USR1 *never* came on?
- [14:23:45] <no1234>
nop
- [14:23:48] <no1234>
never
- [14:23:59] <jkridner>
sorry to say, but it looks like RMA time then.
- [14:24:03] <jkridner>
where are you located?
- [14:24:08] <no1234>
Germany
- [14:24:53] <jkridner>
in the future, we are going to try to get RMA boards pushed out to the regions, but today it still requires shipping to/from the US.
- [14:25:15] <jkridner>
you can find RMA instructions at http://beagleboard.org/support/rma
- [14:25:32] <jkridner>
well, the form anyway. not much for instructions.
- [14:25:43] <jkridner>
wait for the e-mail reply before sending the board.
- [14:25:54] <no1234>
ok thank you very much
- [14:26:02] <jkridner>
http://beagleboard.org/support/
- [14:26:09] <jkridner>
sorry for the problem.
- [14:26:32] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [14:26:35] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [14:26:42] <no1234>
No problem! Really like this channel and helpfull people!
- [14:26:44] <felipec>
mrc3: read Greg Kroah-Hartman's slides, cs is a major contributor to GCC
- [14:26:55] <felipec>
er, that was for mru
- [14:27:01] * nathanm (n=nathan@adsl-69-106-228-95.dsl.pltn13.pacbell.net) Quit (Read error: 113 (No route to host))
- [14:27:06] <mru>
all the more worrying
- [14:27:50] <felipec>
mru: how so?
- [14:28:05] <mru>
untested code being pushed back into mainline gcc
- [14:28:30] <felipec>
mru: who said it was untested?
- [14:28:34] <mru>
pbrook
- [14:28:46] <felipec>
mru: that was the cs release, not the code merged
- [14:28:52] <mru>
he said they were too cheap to test on real hardware
- [14:29:05] <mru>
why should I believe the merged code is any better?
- [14:30:31] <pbrook>
mru: Most of this code is written before hardware is available.
- [14:30:51] <mru>
cortex-a8 hardware has been around for quite a while
- [14:31:01] <pbrook>
Really?
- [14:31:06] <felipec>
mru: right, you would have to assume there's somebody like you, but that also does something productive like test the code cs pushes upstream
- [14:31:16] <mru>
I've had a beagle board since June
- [14:31:39] <mru>
note to self: never help felipec again
- [14:32:47] <pbrook>
mru: I'm pretty sure it wasn't generally available until a couple of months ago, which is too late for the 2008q3 release
- [14:33:04] <mru>
I'm sure you could have got some early samples had you bothered asking
- [14:33:12] <mru>
worked for me...
- [14:34:36] <felipec>
mru: seriously, I think there's people concerned like you, but are testing upstream gcc, maybe I'm wrong, but the point is that is somebody wants something to get done, you either have to do it or pay somebody else to do it
- [14:34:41] <pbrook>
mru: We talked with someone from ar about that. The conclusion what that the BSP probably wouldn't be done, so there wasn't much point.
- [14:34:53] <felipec>
s/is somebody/if somebody/
- [14:35:16] <pbrook>
Also, the current A8 hardware has the NEON lockup bug, which makes automated testing kinda sketchy.
- [14:35:24] <mru>
bug reports to upstream gcc generally get ignored
- [14:35:49] <mru>
or denied
- [14:36:17] <jkridner>
pbrook: seems that the NEON lockup bug could be somehow configuration related. I'm not sure how that can be, but we are internally said to have a Linux release that prevents the issue, even on ES2.1.
- [14:36:33] <jkridner>
sorry for the sketchy information, but I'm just learning about this.
- [14:36:57] <mru>
disabling L1 cache would prevent it from showing
- [14:37:14] <jkridner>
we'll see if it is that ugly of a hack or not.
- [14:37:27] <pbrook>
This is a prime example of why we don't like hardware :-)
- [14:37:41] * Olipro (i=Olipro@uncyclopedia/Olipro) Quit (Read error: 113 (No route to host))
- [14:37:49] <pbrook>
Especially early/pre-production hardware.
- [14:37:50] <jkridner>
well, ultimately things must work on hardware.
- [14:38:13] <jkridner>
better to work-around real issues than to target vapor.
- [14:38:40] <mru>
pbrook: if you tested on hardware earlier, some hardware bugs would be found sooner too
- [14:39:00] <felipec>
jkridner: so what do you propose? start to work on compilers after the hardware is stable?
- [14:39:07] <pbrook>
We've done that before. And wasted lots of time.
- [14:39:07] <jkridner>
not at all.
- [14:39:19] <mru>
start testing on hardware as soon as any hardware at all is available
- [14:39:20] <jkridner>
just that we should supply hardware as early as possible...
- [14:39:37] <jkridner>
but I understand if it cannot be 100% integrated into the production flow.
- [14:40:26] <felipec>
yes, but the emulator development takes a lot of effort, and then you want testing on hardware, which also takes a lot of effort, my guess is that more resources are needed to target both
- [14:40:32] <pbrook>
It's certainly doable, but not something we currently have the resources to do on a large scale.
- [14:41:36] <jkridner>
does all of this come down to someone funding a release for Beagle or some other hardware platform?
- [14:41:38] <pbrook>
The emulator is part of our product anyway, so we know how to do that. We don't have any hardware engineers and don't have any real kernel engineers or distro builders, so are reliant on someone else to all that.
- [14:42:21] <Crofton|work>
koen added 2008q3 to OE before he took off
- [14:42:32] <Crofton|work>
it is possible to build Angstrom for the beagle with it
- [14:42:51] <jkridner>
pbrook: but it is always a problem when you own both sides (spec and implementation) that some things don't get communicated.
- [14:43:12] <pbrook>
jkridner: That's certainly a possibility. Buying a subscription, doing the testing and filing bugs against the prelease would probably get you a lot of the way there.
- [14:44:07] <jkridner>
k, TI owns some subscriptions, but hasn't been filing many bugs against it, as far as I know. I only know of one bug filed.
- [14:45:14] <felipec>
jkridner: maybe TI should be taking a look at the issues the community finds?
- [14:45:17] <jkridner>
no reason to point all the issues back to CS if the hardware folks aren't keeping up to our part of the bargain.
- [14:45:30] <jkridner>
felipec: sounds like it.
- [14:45:57] * igor321 (n=igor@160-6.dsl.iskon.hr) has joined #beagle
- [14:46:23] * jkridner pulls from his pool of infinite resources. :)
- [14:46:36] <felipec>
probably we at Nokia should be doing that too... but we are not so close to the community as we should be =/
- [14:47:11] <jkridner>
in case the sarcasm didn't come through, it was there, but I think I can entice someone to take that on.
- [14:47:24] <jkridner>
depending on the size/number of bugs we are talking about.
- [14:48:05] <jkridner>
from what I've seen, they've been minor, but fundamental. clearing that layer, however, could turn up new issues.
- [14:48:45] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [14:49:01] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [14:49:23] <jkridner>
perhaps we could start by accepting bugs on code.google.com/p/beagleboard?
- [14:49:51] <Crofton|work>
accepting bugs is one thing
- [14:49:59] <Crofton|work>
you need a way to fix bugs :)
- [14:50:11] <mru>
admitting them would be a start...
- [14:50:14] <jkridner>
some issues have lived there a long time...
- [14:50:24] <Crofton|work>
yeah
- [14:50:27] <Crofton|work>
that gets depressing
- [14:50:39] <jkridner>
but, I think the attention beagleboard is getting might help get someone looking at them.
- [14:53:15] * felipec didn't notice the bb "bugzilla" until now
- [14:53:44] <Crofton|work>
jkridner, I suspect you'd have to pay Hogwarts to have someone read the bb bugs
- [14:54:57] <jkridner>
we have paid some, but probably not enough. :)
- [14:56:06] <jkridner>
just having someone track the issues could have its own rewards.
- [14:57:29] <jkridner>
the OE stuff has been going so well (along with using CSL 2007q3) that I have not been pushing hard on fixing the other releases.
- [14:58:09] <Crofton|work>
someone could try building Angstrom with 2008q3
- [14:58:17] <Crofton|work>
but I am busy with other things
- [14:58:27] <Crofton|work>
jkridner, you need more interns
- [14:58:45] <jkridner>
indeed!
- [14:59:07] * lukketto (n=luca@host86-192-dynamic.7-87-r.retail.telecomitalia.it) Quit (Read error: 113 (No route to host))
- [14:59:30] <jkridner>
no one comes by my office without something to do. :)
- [15:00:13] <jkridner>
of course, it would be nice to actually get into the office. won't be in until Friday as I'm back on the road again.
- [15:00:26] <Crofton|work>
where now?
- [15:00:33] <jkridner>
Seattle, WA.
- [15:00:41] <Crofton|work>
I may have to do some work in Phoenix
- [15:00:48] <Crofton|work>
hopefully not until Novemeber though
- [15:00:51] <Crofton|work>
heh
- [15:00:56] <Crofton|work>
home of the evil empire
- [15:00:56] <jkridner>
then I have a few days to finish my ESC-Boston training before spending a week there.
- [15:00:58] <mru>
november is closer than you thing...
- [15:01:00] <mru>
think
- [15:01:03] <Crofton|work>
yeah
- [15:01:16] <Crofton|work>
I have to rpesent some Beagle SDR stuff the last week on Oct
- [15:01:21] <jkridner>
need to insert about 4 new weeks into October somehow. :)
- [15:03:31] <Crofton|work>
I need to make no commitments for December
- [15:38:10] * garren (n=garren@iburst-41-213-57-157.iburst.co.za) has joined #beagle
- [15:50:13] <garren>
hi all
- [16:04:30] * garren (n=garren@iburst-41-213-57-157.iburst.co.za) Quit ("Ex-Chat")
- [16:15:41] * no1234 (n=masta@p5493C097.dip.t-dialin.net) Quit (Read error: 60 (Operation timed out))
- [16:24:29] * Olipro_ is now known as Olipro
- [16:25:52] * BThompson (n=BThompso@cpe-76-185-93-11.tx.res.rr.com) has joined #beagle
- [16:39:18] * igor321 (n=igor@160-6.dsl.iskon.hr) Quit ("leavin'")
- [16:50:35] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Remote closed the connection)
- [16:59:51] * valhalla (n=valhalla@81-174-25-222.dynamic.ngi.it) Quit ("Leaving")
- [17:00:43] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [17:01:37] * emeb (n=ericb@ip72-223-84-209.ph.ph.cox.net) has left #beagle
- [17:03:15] * mpoullet (n=mpoullet@p4FC18611.dip0.t-ipconnect.de) has joined #beagle
- [17:16:42] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [17:16:42] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [17:33:25] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [17:33:39] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [18:02:28] * stud (n=stud@pool-71-96-223-186.dfw.dsl-w.verizon.net) has joined #beagle
- [18:03:35] * stud (n=stud@pool-71-96-223-186.dfw.dsl-w.verizon.net) Quit (Remote closed the connection)
- [18:15:49] * Fl1pFl0p (n=Fl1pFl0p@ip68-8-230-62.sd.sd.cox.net) Quit ("Leaving.")
- [18:15:57] * nathanm (n=nathan@adsl-69-106-228-95.dsl.pltn13.pacbell.net) has joined #beagle
- [18:21:38] * Beagle2 (n=Beagle2@nat/microsoft/x-25bf65668adb16bb) has joined #beagle
- [18:22:42] * Beagle2 (n=Beagle2@nat/microsoft/x-25bf65668adb16bb) Quit (Client Quit)
- [18:23:35] <kulve>
testing official builds solely with an emulator instead of real hardware sounds a bit risky to me. And testing with more complex real world examples instead of only hand crafted test cases usually reveals new issues. You can't never test enough..
- [18:24:20] <pbrook>
You've got to draw the line somewhere.
- [18:25:17] <pbrook>
We do some testing on real hardware.
- [18:25:25] <kulve>
yeah and usually it's drawn too low because of schedules etc. business issues..
- [18:25:34] <mru>
pbrook: what? hello world?
- [18:25:54] <pbrook>
That's one of them yes.
- [18:26:23] <Crofton|work>
:)
- [18:26:24] <kulve>
pbrook: maybe you should publish your testcases? Others could use them then too and maybe add more missing test cases.. :)
- [18:26:37] <pbrook>
kulve: Mostly it's the gcc testsuite.
- [18:26:38] <Crofton|work>
that would be really embaressing to have not work
- [18:26:53] <pbrook>
Crofton: You'd be surprised...
- [18:26:57] <mru>
well, they did break memcpy
- [18:27:16] <mru>
hard to say which is more embarassing
- [18:27:26] <pbrook>
There's also several dozen different targets.
- [18:27:34] <mru>
excuses, excuses...
- [18:27:53] <Crofton|work>
I need to figure out what generic_interupt does in the kernel .....
- [18:27:54] <pbrook>
Getting "complete" test coverage is something that would probably be a full time job for a team of people.
- [18:28:24] <mru>
I'm not asking for testing complete test coverage
- [18:28:34] <mru>
I'm asking for basic stuff working with default settings
- [18:28:36] <pbrook>
I don't remember if any of our current ARM hardware is stable enough to do real test runs on.
- [18:29:26] <Crofton|work>
http://rafb.net/p/gfQW5J36.html
- [18:29:30] <mru>
someone at ARM told me they *did* report bugs on the prerelease
- [18:29:34] <mru>
didn't go into specifics
- [18:29:44] <pbrook>
And did those bugs get fixed?
- [18:30:30] <mru>
he expressed surprise at this being released now
- [18:31:07] <pbrook>
There should be no surprise about the release date. It coincides with ARM DevCon.
- [18:31:25] <mru>
oh, so that's the explanation
- [18:31:33] <mru>
release on specific date, rather than when functional
- [18:31:39] <pbrook>
No.
- [18:31:43] <mru>
idiotic practise
- [18:31:55] <pbrook>
At the time of release, we weren't aware of any critical bugs.
- [18:32:12] <mru>
that's your word against ARM's
- [18:32:22] <mru>
I suggest you contact them and sort things out
- [18:32:22] <pbrook>
We have a bug tracker to prove it.
- [18:32:36] <pbrook>
If they didn't file a bug, that's their own fault.
- [18:32:55] <mru>
they say they did...
- [18:32:59] <Crofton|work>
can you two agree to disagree?
- [18:33:19] * mru goes off to stab someone on the ps3
- [18:33:30] <Crofton|work>
what game?
- [18:33:47] <Crofton|work>
I've been playing Warhawls lately
- [18:33:56] <mru>
assassin's creed
- [18:34:23] <Crofton|work>
I'll have to look it up
- [18:34:33] <mru>
beatiful graphics
- [18:34:37] <mru>
and you get to stab people
- [18:39:15] <Crofton|work>
um stabbering people
- [19:06:36] * gcohler (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) has joined #beagle
- [19:16:15] * bazbell (n=a0192809@nat/ti/x-c14cd1acfd8540da) Quit ("Leaving.")
- [19:38:51] <jkridner>
hi gcohler
- [19:41:03] <Crofton|work>
http://rafb.net/p/gfQW5J36.html
- [19:41:16] <Crofton|work>
need to convert the fir stuff to real NEON :)
- [20:00:32] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [20:02:55] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Nick collision from services.)
- [20:02:55] * Olipro_ is now known as Olipro
- [20:03:06] * john280z (n=johnm@user-0c2h5av.cable.mindspring.com) has joined #beagle
- [20:21:57] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [20:22:03] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Nick collision from services.)
- [20:22:05] * Olipro_ is now known as Olipro
- [20:31:03] <nathanm>
Crofton: Does it run at usable speed right now?
- [20:32:17] <mpoullet>
Hi, regarding csl 2008q3, I've tried it with u-boot 1.3.3 and the resulting image was broken: a lot of i2c errors/timeout; with 2007q3 no problem. But maybe it's a bad idea to test it with u-boot?
- [20:32:19] <Crofton|work>
not yet
- [20:32:34] * gcohler (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) Quit (Read error: 113 (No route to host))
- [20:37:02] * gerrynjr (n=gerrynjr@gentoo/user/gerrynjr) has joined #beagle
- [20:48:01] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [20:48:01] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) Quit (Read error: 104 (Connection reset by peer))
- [21:02:02] * rsalveti (n=salveti@189.70.235.170) Quit (Read error: 110 (Connection timed out))
- [21:03:41] * rsalveti (n=salveti@189.70.34.34) has joined #beagle
- [21:04:55] * Crofton|work (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [21:05:33] * Crofton (n=balister@66-207-66-26.black.dmt.ntelos.net) has joined #beagle
- [21:18:30] <jkridner>
mpoullet: u-boot seems like a relatively simple target for testing the compiler.
- [21:24:18] * dcordes_ (n=dcordes@unaffiliated/dcordes) has joined #beagle
- [21:26:31] <mpoullet>
jkridner: ok, I'll then test the latest Steve's u-boot too. Or has anybody already tried to do it?
- [21:28:47] <kulve>
2008q3 doesn't handle neon, so 2007q3 is still the recommended toolchain for beagle
- [21:30:16] * gduncan (n=spec@h-67-101-179-14.lsanca54.covad.net) has joined #beagle
- [21:35:59] * dcordes (n=dcordes@unaffiliated/dcordes) Quit (Read error: 110 (Connection timed out))
- [21:38:08] * gduncan (n=spec@h-67-101-179-14.lsanca54.covad.net) Quit ()
- [21:59:20] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [22:02:53] <mru>
u-boot is simple code, true, but if something does go wrong, it's very hard to debug
- [22:06:30] * Xenion (n=robert@p579FC994.dip.t-dialin.net) Quit ("Verlassend")
- [22:13:34] * dcordes_ is now known as dcordes
- [22:16:40] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Read error: 110 (Connection timed out))
- [22:19:59] * Olipro_ is now known as Olipro
- [22:37:29] * mpoullet (n=mpoullet@p4FC18611.dip0.t-ipconnect.de) has left #beagle
- [22:38:09] * gcohler (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) has joined #beagle
- [22:54:11] * dcordes (n=dcordes@unaffiliated/dcordes) Quit (Remote closed the connection)
- [23:04:37] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [23:08:34] * Olipro_ (n=Olipro@uncyclopedia/Olipro) Quit (Read error: 104 (Connection reset by peer))
- [23:08:43] * Olipro__ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [23:11:03] * Olipro (n=Olipro@uncyclopedia/Olipro) Quit (Read error: 60 (Operation timed out))
- [23:16:27] * hli (i=chaton@vig91-2-82-232-97-149.fbx.proxad.net) Quit ("ChatZilla 0.9.83 [Firefox 3.0.3/2008092417]")
- [23:18:25] * Olipro_ (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [23:18:47] * zedstar (n=john@fsf/member/zedstar) Quit (Remote closed the connection)
- [23:18:58] * woski (n=dleme@65.182.51.67) Quit (Read error: 104 (Connection reset by peer))
- [23:19:24] * woski (n=dleme@65.182.51.67) has joined #beagle
- [23:23:26] * guillaum1 (n=Guillaum@AMontsouris-153-1-38-79.w90-2.abo.wanadoo.fr) Quit ("Leaving.")
- [23:24:25] * guillaum1 (n=gl@AMontsouris-153-1-38-79.w90-2.abo.wanadoo.fr) has joined #beagle
- [23:25:13] * dcordes (n=dcordes@unaffiliated/dcordes) has joined #beagle
- [23:34:35] * gcohler (n=gcohler@pool-72-93-217-203.bstnma.fios.verizon.net) Quit (Read error: 113 (No route to host))
- [23:35:28] * Olipro__ (n=Olipro@uncyclopedia/Olipro) Quit (Success)
- [23:40:41] * Dionysus (n=dan@adsl-75-45-108-57.dsl.scrm01.sbcglobal.net) Quit (Read error: 54 (Connection reset by peer))
- [23:44:38] * JimDonovan (n=jimd@c-66-30-223-191.hsd1.ma.comcast.net) has joined #beagle
- [23:45:34] * Olipro (n=Olipro@uncyclopedia/Olipro) has joined #beagle
- [23:49:45] * JimDonovan (n=jimd@c-66-30-223-191.hsd1.ma.comcast.net) Quit ("Leaving.")
- [23:50:47] * JimDonovan (n=jimd@c-66-30-223-191.hsd1.ma.comcast.net) has joined #beagle
- [23:56:52] * john280z (n=johnm@user-0c2h5av.cable.mindspring.com) has left #beagle
- [23:57:58] * felipec (n=felipec@a88-113-6-167.elisa-laajakaista.fi) Quit ("Leaving")
- [23:59:46] * Olipro_ (n=Olipro@uncyclopedia/Olipro) Quit (Connection timed out)
These logs were automatically created by BeagleLogBot on
irc.freenode.net
using the Java IRC LogBot.
See http://microformats.org/wiki/mflogbot for more information.