Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
1. Patches (https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
2. Merging the changes, assuming they're ok. Infra people don't have commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
Thanks, Paul
Linaro.org | Open source software for ARM SoCs Follow Linaro: http://www.facebook.com/pages/Linaro http://twitter.com/#%21/linaroorg - http://www.linaro.org/linaro-blog
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost all kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.orgwrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot*
cooler.
Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too.
Actually,
it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes.
I
am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of
people (or
Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
As discussed off-line with Renato, it seems there is no wifi on Panda (non-ES), and I didn't notice any useless daemons.
Let's wait for the gcc build to complete with the thermal driver enabled.
Christophe.
On 22 January 2013 14:31, Renato Golin Linaro renato.golin@linaro.org wrote:
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost all kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.org wrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
I just had another lockup this night, given enough load, it will fail. :(
cheers, --renato
On 22 January 2013 16:18, Christophe Lyon christophe.lyon@linaro.orgwrote:
As discussed off-line with Renato, it seems there is no wifi on Panda (non-ES), and I didn't notice any useless daemons.
Let's wait for the gcc build to complete with the thermal driver enabled.
Christophe.
On 22 January 2013 14:31, Renato Golin Linaro renato.golin@linaro.org wrote:
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost
all
kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.org wrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote: > > Hello, > > I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks > from > Infra should be present on Tue hangout). > > We've been doing integration testing since end of last week, and > generally it looks good. We didn't have complete end-to-end GCC
build
> due to PandaES + USB drive availability issues, but I tested gcc > build > on Panda instead (this has some OOMs during "make check" in gcc),
and
> smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
> So, everything looks good for deployment on Thurs, just 2 following > issues are on critical path: > > 1. Patches > (https://code.launchpad.net/~linaro-infrastructure/+activereviews) > review by TCWG. Matt, I guess you were busy with toolchain release > last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
> 2. Merging the changes, assuming they're ok. Infra people don't
have
> commit access to CBuild repos, so we depend on TCWG here too. > Actually, > it may be expected that after initial launch, we'll need to do more > changes and tweaks, so it may be good idea to give Infra
(temporary)
> commit access to streamline process. It would be nice to discuss
this
> during Tue hangout.
I thought you had access, and was expecting you to merge all
changes.
I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
Hi all, I could run the gcc testsuite last night on tcpanda01: it took about 21h (!), but I still don't know how hot the board became :-(
Christophe.
On 24 January 2013 10:48, Renato Golin Linaro renato.golin@linaro.org wrote:
I just had another lockup this night, given enough load, it will fail. :(
cheers, --renato
On 22 January 2013 16:18, Christophe Lyon christophe.lyon@linaro.org wrote:
As discussed off-line with Renato, it seems there is no wifi on Panda (non-ES), and I didn't notice any useless daemons.
Let's wait for the gcc build to complete with the thermal driver enabled.
Christophe.
On 22 January 2013 14:31, Renato Golin Linaro renato.golin@linaro.org wrote:
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost all kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.org wrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
> On 21/01/13 19:38, Paul Sokolovsky wrote: >> >> Hello, >> >> I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks >> from >> Infra should be present on Tue hangout). >> >> We've been doing integration testing since end of last week, and >> generally it looks good. We didn't have complete end-to-end GCC >> build >> due to PandaES + USB drive availability issues, but I tested gcc >> build >> on Panda instead (this has some OOMs during "make check" in gcc), >> and >> smaller builds like cortex-strings on PandaES. > > > Some tests deliberately try to exhaust all memory - if they're not > already we should be explicitly ulimiting the make check process to > not > cause random processes to be killed. > >> So, everything looks good for deployment on Thurs, just 2 >> following >> issues are on critical path: >> >> 1. Patches >> (https://code.launchpad.net/~linaro-infrastructure/+activereviews) >> review by TCWG. Matt, I guess you were busy with toolchain release >> last week, but we'd appreciate your review now. > > > I'll try and do that in the next 48 hours. > >> 2. Merging the changes, assuming they're ok. Infra people don't >> have >> commit access to CBuild repos, so we depend on TCWG here too. >> Actually, >> it may be expected that after initial launch, we'll need to do >> more >> changes and tweaks, so it may be good idea to give Infra >> (temporary) >> commit access to streamline process. It would be nice to discuss >> this >> during Tue hangout. > > > I thought you had access, and was expecting you to merge all > changes. > I > am happy for anyone to merge their own stuff when the merge has > been > approved. Can someone send me (off-list) the appropriate list of > people (or > Launchpad group) to give commit access to? > > Thanks, > > Matt > > > -- > Matthew Gretton-Dann > Toolchain Working Group, Linaro > > _______________________________________________ > linaro-validation mailing list > linaro-validation@lists.linaro.org > http://lists.linaro.org/mailman/listinfo/linaro-validation
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
I would say getting it to completion is a good start. Can we look at deploying this on the other PandaBoards (not the ESes) in CBuild and then reintroducing them to the queues?
Also have you written up what you've done on a Wiki page somewhere?
Thanks,
Matt
On 29/01/13 15:24, Christophe Lyon wrote:
Hi all, I could run the gcc testsuite last night on tcpanda01: it took about 21h (!), but I still don't know how hot the board became :-(
Christophe.
On 24 January 2013 10:48, Renato Golin Linaro renato.golin@linaro.org wrote:
I just had another lockup this night, given enough load, it will fail. :(
cheers, --renato
On 22 January 2013 16:18, Christophe Lyon christophe.lyon@linaro.org wrote:
As discussed off-line with Renato, it seems there is no wifi on Panda (non-ES), and I didn't notice any useless daemons.
Let's wait for the gcc build to complete with the thermal driver enabled.
Christophe.
On 22 January 2013 14:31, Renato Golin Linaro renato.golin@linaro.org wrote:
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost all kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.org wrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
Interesting. Christophe do you want to take a look?
Matt
On 22/01/13 10:24, Dave Pigott wrote: > > Hi Matt, > > Interesting discovery by Renato re overheating on Pandas - he had a > similar problem and just disabled wifi, and the board ran a *lot* > cooler. > Could this be your problem in the toolchain pandas? > > Thanks > > Dave > > On 21 Jan 2013, at 19:53, Matthew Gretton-Dann > matthew.gretton-dann@linaro.org wrote: > >> On 21/01/13 19:38, Paul Sokolovsky wrote: >>> >>> Hello, >>> >>> I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks >>> from >>> Infra should be present on Tue hangout). >>> >>> We've been doing integration testing since end of last week, and >>> generally it looks good. We didn't have complete end-to-end GCC >>> build >>> due to PandaES + USB drive availability issues, but I tested gcc >>> build >>> on Panda instead (this has some OOMs during "make check" in gcc), >>> and >>> smaller builds like cortex-strings on PandaES. >> >> >> Some tests deliberately try to exhaust all memory - if they're not >> already we should be explicitly ulimiting the make check process to >> not >> cause random processes to be killed. >> >>> So, everything looks good for deployment on Thurs, just 2 >>> following >>> issues are on critical path: >>> >>> 1. Patches >>> (https://code.launchpad.net/~linaro-infrastructure/+activereviews) >>> review by TCWG. Matt, I guess you were busy with toolchain release >>> last week, but we'd appreciate your review now. >> >> >> I'll try and do that in the next 48 hours. >> >>> 2. Merging the changes, assuming they're ok. Infra people don't >>> have >>> commit access to CBuild repos, so we depend on TCWG here too. >>> Actually, >>> it may be expected that after initial launch, we'll need to do >>> more >>> changes and tweaks, so it may be good idea to give Infra >>> (temporary) >>> commit access to streamline process. It would be nice to discuss >>> this >>> during Tue hangout. >> >> >> I thought you had access, and was expecting you to merge all >> changes. >> I >> am happy for anyone to merge their own stuff when the merge has >> been >> approved. Can someone send me (off-list) the appropriate list of >> people (or >> Launchpad group) to give commit access to? >> >> Thanks, >> >> Matt >> >> >> -- >> Matthew Gretton-Dann >> Toolchain Working Group, Linaro >> >> _______________________________________________ >> linaro-validation mailing list >> linaro-validation@lists.linaro.org >> http://lists.linaro.org/mailman/listinfo/linaro-validation > >
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
Yes, I can do the same for the other boards. There are 2 more of them, right ? (the ones that are marked as 'reserved' in the scheduler)
the corresponding telnet serial2 70XX does not answer currently, neither does it for tcpanda01 on which I already have a connexion open: is it restricted to one connection at a time? If so, maybe someone is currently connected to tcpanda0[23], so maybe I shouldn't reboot them now?
I have written this https://wiki.linaro.org/ChristopheLyon/Sandbox/UpdateKernelPanda
As you'll notice, I don't know how to perform the next upgrade, since /dev/mmcblk0p1 no longer exists?
Thanks, Christophe
On 29 January 2013 21:13, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
I would say getting it to completion is a good start. Can we look at deploying this on the other PandaBoards (not the ESes) in CBuild and then reintroducing them to the queues?
Also have you written up what you've done on a Wiki page somewhere?
Thanks,
Matt
On 29/01/13 15:24, Christophe Lyon wrote:
Hi all, I could run the gcc testsuite last night on tcpanda01: it took about 21h (!), but I still don't know how hot the board became :-(
Christophe.
On 24 January 2013 10:48, Renato Golin Linaro renato.golin@linaro.org wrote:
I just had another lockup this night, given enough load, it will fail. :(
cheers, --renato
On 22 January 2013 16:18, Christophe Lyon christophe.lyon@linaro.org wrote:
As discussed off-line with Renato, it seems there is no wifi on Panda (non-ES), and I didn't notice any useless daemons.
Let's wait for the gcc build to complete with the thermal driver enabled.
Christophe.
On 22 January 2013 14:31, Renato Golin Linaro renato.golin@linaro.org wrote:
Hi Christophe,
It could be the wireless, or something else. Below is my blacklist on modprobe.d:
# Rack mount no extra rubbish blacklist rfcomm blacklist bnep blacklist bluetooth blacklist ppdev blacklist lp blacklist parport blacklist wl12xx_sdio blacklist wl12xx blacklist mac80211 blacklist twl6040_vibra blacklist ff_memless blacklist cfg80211
This involves wireless, bluetooth, parallel ports etc. It is also almost all kernel modules active (with the exception of the led driver).
I also removed some silly packages and disabled most daemons, but I guess your image is much cleaner than mine.
cheers, --renato
On 22 January 2013 13:26, Christophe Lyon christophe.lyon@linaro.org wrote:
It makes sense indeed (reminds me of my guru-plug getting very hot with wifi enabled).
But currently ifconfig only shows eth0 and lo; it looks like wifi is not activated already :-(
Christophe.
On 22 January 2013 11:43, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote: > > Interesting. Christophe do you want to take a look? > > Matt > > On 22/01/13 10:24, Dave Pigott wrote: >> >> >> Hi Matt, >> >> Interesting discovery by Renato re overheating on Pandas - he had a >> similar problem and just disabled wifi, and the board ran a *lot* >> cooler. >> Could this be your problem in the toolchain pandas? >> >> Thanks >> >> Dave >> >> On 21 Jan 2013, at 19:53, Matthew Gretton-Dann >> matthew.gretton-dann@linaro.org wrote: >> >>> On 21/01/13 19:38, Paul Sokolovsky wrote: >>>> >>>> >>>> Hello, >>>> >>>> I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks >>>> from >>>> Infra should be present on Tue hangout). >>>> >>>> We've been doing integration testing since end of last week, and >>>> generally it looks good. We didn't have complete end-to-end GCC >>>> build >>>> due to PandaES + USB drive availability issues, but I tested gcc >>>> build >>>> on Panda instead (this has some OOMs during "make check" in gcc), >>>> and >>>> smaller builds like cortex-strings on PandaES. >>> >>> >>> >>> Some tests deliberately try to exhaust all memory - if they're not >>> already we should be explicitly ulimiting the make check process to >>> not >>> cause random processes to be killed. >>> >>>> So, everything looks good for deployment on Thurs, just 2 >>>> following >>>> issues are on critical path: >>>> >>>> 1. Patches >>>> (https://code.launchpad.net/~linaro-infrastructure/+activereviews) >>>> review by TCWG. Matt, I guess you were busy with toolchain release >>>> last week, but we'd appreciate your review now. >>> >>> >>> >>> I'll try and do that in the next 48 hours. >>> >>>> 2. Merging the changes, assuming they're ok. Infra people don't >>>> have >>>> commit access to CBuild repos, so we depend on TCWG here too. >>>> Actually, >>>> it may be expected that after initial launch, we'll need to do >>>> more >>>> changes and tweaks, so it may be good idea to give Infra >>>> (temporary) >>>> commit access to streamline process. It would be nice to discuss >>>> this >>>> during Tue hangout. >>> >>> >>> >>> I thought you had access, and was expecting you to merge all >>> changes. >>> I >>> am happy for anyone to merge their own stuff when the merge has >>> been >>> approved. Can someone send me (off-list) the appropriate list of >>> people (or >>> Launchpad group) to give commit access to? >>> >>> Thanks, >>> >>> Matt >>> >>> >>> -- >>> Matthew Gretton-Dann >>> Toolchain Working Group, Linaro >>> >>> _______________________________________________ >>> linaro-validation mailing list >>> linaro-validation@lists.linaro.org >>> http://lists.linaro.org/mailman/listinfo/linaro-validation >> >> >> > > > -- > Matthew Gretton-Dann > Toolchain Working Group, Linaro
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
Oh, I went for a more copper based solution:
http://www.swiftech.com/mc14bgamemoryramsinks.aspx
I expect they are overkill, but they fit nicely and have adhesive on them so fitting is trivial. They also look nice :-)
Jameds
On 22 January 2013 10:24, Dave Pigott dave.pigott@linaro.org wrote:
Hi Matt,
Interesting discovery by Renato re overheating on Pandas - he had a similar problem and just disabled wifi, and the board ran a *lot* cooler. Could this be your problem in the toolchain pandas?
Thanks
Dave
On 21 Jan 2013, at 19:53, Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:38, Paul Sokolovsky wrote:
Hello,
I'm offline Tue/Wed, so here's update on CBuild/LAVA (other folks from Infra should be present on Tue hangout).
We've been doing integration testing since end of last week, and generally it looks good. We didn't have complete end-to-end GCC build due to PandaES + USB drive availability issues, but I tested gcc build on Panda instead (this has some OOMs during "make check" in gcc), and smaller builds like cortex-strings on PandaES.
Some tests deliberately try to exhaust all memory - if they're not already we should be explicitly ulimiting the make check process to not cause random processes to be killed.
So, everything looks good for deployment on Thurs, just 2 following issues are on critical path:
- Patches
(https://code.launchpad.net/~linaro-infrastructure/+activereviews) review by TCWG. Matt, I guess you were busy with toolchain release last week, but we'd appreciate your review now.
I'll try and do that in the next 48 hours.
- Merging the changes, assuming they're ok. Infra people don't have
commit access to CBuild repos, so we depend on TCWG here too. Actually, it may be expected that after initial launch, we'll need to do more changes and tweaks, so it may be good idea to give Infra (temporary) commit access to streamline process. It would be nice to discuss this during Tue hangout.
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
Thanks,
Matt
-- Matthew Gretton-Dann Toolchain Working Group, Linaro
linaro-validation mailing list linaro-validation@lists.linaro.org http://lists.linaro.org/mailman/listinfo/linaro-validation
On 21/01/13 19:53, Matthew Gretton-Dann wrote:
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
All merge requests have, I believe, now been approved. linaro-infrastructure members should also have commit rights. So feel free to commit your changes :-).
Thanks,
Matt
Hello,
On Tue, 22 Jan 2013 11:07:41 +0000 Matthew Gretton-Dann matthew.gretton-dann@linaro.org wrote:
On 21/01/13 19:53, Matthew Gretton-Dann wrote:
I thought you had access, and was expecting you to merge all changes. I am happy for anyone to merge their own stuff when the merge has been approved. Can someone send me (off-list) the appropriate list of people (or Launchpad group) to give commit access to?
All merge requests have, I believe, now been approved. linaro-infrastructure members should also have commit rights. So feel free to commit your changes :-).
Thanks Matt, starting with that today. I don't expect big downtimes, will announce on IRC any 15min or so switchovers.
Thanks,
Matt
On Thu, 24 Jan 2013 12:32:42 +0200 Paul Sokolovsky Paul.Sokolovsky@linaro.org wrote:
[]
Thanks Matt, starting with that today. I don't expect big downtimes, will announce on IRC any 15min or so switchovers.
Update: basic deployment is done. Previous versions of components are available in *.before-lava dirs, post-merge components are in *.lava dirs, with a symlink pointing from a basename dir. So, it would take to just re-make symlinks to go back to previous version in case of emergency.
The system seem to work well, I watched few CBuild native builds to finish and be recorded properly, as well as launched one myself (all of these were EC2-based ones, as there're long queue of long-running native ARM builds). LAVA-based builds works too, though there're still tweaks to do for them to fit better with CBuild process (will send more info later).
You can have a look at it via http://cbuild.validation.linaro.org/helpers/scheduler and other usual places ("Build log", "Recent" pages). Ah, and I added myself and few other Infra folks to "admin" group, so we can launch jobs.
So, I keep watching it closely, and Matt, let me know if there's any unexpected behavior.
Thanks, Paul
Linaro.org | Open source software for ARM SoCs Follow Linaro: http://www.facebook.com/pages/Linaro http://twitter.com/#%21/linaroorg - http://www.linaro.org/linaro-blog
linaro-validation@lists.linaro.org
-
Christophe Lyon -
Dave Pigott -
James Tunnicliffe -
Matthew Gretton-Dann -
Paul Sokolovsky -
Renato Golin Linaro