[OpenWrt-Devel] I've come round to your way of thinking
Sorry sent from the wrong email address, not sure where it it actually got posted and where not. Hi Oswald, I'm sorry I suggested you were an unrealistic idealogue, and for questioning your credentials; while I haven't verified them I'm sure you do have more experience than I gave you credit for, despite how overly strong you come into this discussion. With the creation of the tool I created for myself that gets in-between me and posting email (requiring me to confirm I really want to send it after cooling off period), I think that the kinds of discussion of David Lang (and I agreed) were to volatile to have publicly ought to be able to kept to reasonable level, even with hotheads like myself, John, and Imre, who need tools not just an admonition to think before (or even knowing one has an issue it not entering our thoughts when we are posting). In fact I'd argue with such a mechanism, for folks like me and them a public email discussion is *preferable* to an IRC discussion (which I've seen frequently turn toxic in their and other cases), especially a private one. Regards, daniel ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
[OpenWrt-Devel] Transparency and discussion of merge openwrt and lede
Hi, Might I humbly submit that given the different timezone and the fact that LEDE claims to be wanting to be transparent, and that remaining OpenWrt claims to be willing to accept such policies, that Jow's suggestion of doing the discussion openly on the openwrt-devel and lede-dev mailings lists is what makes the most sense? If you're worried that things could get nasty perhaps, at least for this discussion, both sides could agree to adopt something like the tool I created for myself to help me control my own tendency to hasty, hotheaded, and generally unhelpful or noisy emails, so that if one has a tendency to drop the gloves that one has a chance to rethink what they've said after a cooling off period? If a .rpm would help, I'll make time to build version of the package for .rpm distros too. Regards, Daniel ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] OpenWrt / LEDE
On Wed, 2016-05-25 at 01:13 -0700, mbm wrote: > > [snip] > > Let's see if any of the remaining OpenWrt devs at least publicly support > > adopting them or some variation of them. As I've said before my > > impression is that LEDE-style rules are not all that welcomed (and > > that's based on the interactions I saw on the private openwrt channels > > when I was a developer, not just a pure outsider view; it's possible my > > impression is wrong, but the toxicity described previously was in large > > part negative reaction by folkds in the LEDE team to toxic comments from > > at least one of the remaining OpenWrt devs; certainly it damaged my > > opinion of him (although the toxicity also damaged my opinion of a > > couple of LEDE folks too)). > Sigh... > > It's not as if LEDE offered the changes to OpenWrt and we voted > against them causing a split -- there was literally no discussion and > no warning prior to the public announcement. LEDE just suddenly > existed and the story quickly became that the reason for their > existence was because OpenWrt somehow prevented them from making > changes. At no point did OpenWrt veto changes or even have the option > to; the truth is that we agree changes need to be made. By not I guess part of the problem is that the LEDE team didn't believe that changes would actually be possible, because the toxic way in certain members interacted made reasonable discussion impossible (this is not a one-sided thing either IMO). Certain if transparency and greater community participation, and more opportunities for new blood to join, were adopted by OpenWrt, and a reasonable set of rules (i.e. not necessarily LEDE's rules) regarding who makes the decisions (e.g. I'm not sure I entirely buy LEDE's only committers should vote, if the goal is truly a greater community voice, although if committers are the only ones voting then I agree that it should be active committers, not just anyone who was at one time active; I think a notion of activity should include measure other that commits, however (I don't buy that the only thing that matters in an open source project is code commits)) Also a number of the other rules make sense, although I don't see that merger means that LEDE rules should necessarily be adopted as-is with no discussion. > including OpenWrt in the discussions LEDE ran afoul of their own > transparency leading to the false impression that OpenWrt was somehow > against the changes, causing a split in the community in terms of LEDE > vs OpenWrt with various amounts of hostility on the mailing lists. > None of this is healthy or constructive. That is largely my fault for expression my impressions that I had because of the toxic interactions between developers when I was on the private channel. I apologize for that, for all the good it will do (I don't know as there is anything I can do now to fix that). I would submit, however, that a solution to the toxicity is essential to any merger. Part of the reason I stepped down is that I wasn't helping matters because of personal issues (and have done it again more publicly now). I can't fix that, but I *can* point out that the the environment that creates this situation needs to be fixed. > > Let me be very clear: nobody on the OpenWrt side is against the > changes LEDE is trying to make. It is our position that this whole > thing is a misunderstanding and that the projects should attempt to > merge again. I have been trying to arrange talks between the two sides > but between work schedules, timezone conflicts and FUD regarding the > split it's been very difficult. > I would very much like to see openwrt and lede merge, but I know I'm not the calm voice that can make that happen. You and Hauke (for example) may be the calm heads that could make this happen, although it takes the more headstrong ones being willing to listen to you. Regards, Daniel ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
[OpenWrt-Devel] How to debug/config qos-scripts to work with OpenWRT AA?
Hello, I am trying to use qos-scripts in OpenWRT AA. I have an issue that the qos-scripts can limit uplink speed but not downlink speed. For example: I set 128kbit uplink and 1024kbit downlink, however, the downlink is limitless This is the speedtest I captured http://speedof.me/show.php?img=160526022701-7038.png - uplink can go up to 104kbps - downlink can go up to 7861kbps (which is higher than the limitation I set) --- I also tried with wshaper and the got same result. Here is my setup: - eth1 is the WAN port - eth0 is connect to my PC - OpenWRT AA - Linux kernel 3.3.8 * cmd: cat /etc/config/qos # QoS configuration for OpenWrt # INTERFACES: config interface wan option classgroup "Default" option enabled 1 option upload 128 option download 1024 # RULES: config classify option target "Priority" option ports"22,53" option comment "ssh, dns" config classify option target "Normal" option proto"tcp" option ports"20,21,25,80,110,443,993,995" option comment "ftp, smtp, http(s), imap" config classify option target "Express" option ports"5190" option comment "AOL, iChat, ICQ" config default option target "Express" option proto"udp" option pktsize "-500" config reclassify option target "Priority" option proto"icmp" config default option target "Bulk" option portrange"1024-65535" # Don't change the stuff below unless you # really know what it means :) config classgroup "Default" option classes "Priority Express Normal Bulk" option default "Normal" config class "Priority" option packetsize 400 option avgrate 10 option priority20 config class "Priority_down" option packetsize 1000 option avgrate 10 config class "Express" option packetsize 1000 option avgrate 50 option priority10 config class "Normal" option packetsize 1500 option packetdelay 100 option avgrate 10 option priority5 config class "Normal_down" option avgrate 20 config class "Bulk" option avgrate 1 option packetdelay 200 * cmd: /usr/lib/qos/generate.sh all | insmod cls_u32 >&- 2>&- | insmod em_u32 >&- 2>&- | insmod act_connmark >&- 2>&- | insmod act_mirred >&- 2>&- | insmod sch_ingress >&- 2>&- | insmod cls_fw >&- 2>&- | insmod sch_hfsc >&- 2>&- | insmod sch_fq_codel >&- 2>&- | ifconfig eth1 up txqueuelen 5 >&- 2>&- | tc qdisc del dev eth1 root >&- 2>&- | tc qdisc add dev eth1 root handle 1: hfsc default 30 | tc class add dev eth1 parent 1: classid 1:1 hfsc sc rate 128kbit ul rate 128kbit | tc class add dev eth1 parent 1:1 classid 1:10 hfsc rt m1 74kbit d 6103us m2 12kbit ls m1 74kbit d 6103us m2 71kbit ul rate 128kbit | tc class add dev eth1 parent 1:1 classid 1:20 hfsc rt m1 68kbit d 15258us m2 64kbit ls m1 68kbit d 15258us m2 35kbit ul rate 128kbit | tc class add dev eth1 parent 1:1 classid 1:30 hfsc ls m1 0kbit d 10us m2 17kbit ul rate 128kbit | tc class add dev eth1 parent 1:1 classid 1:40 hfsc ls m1 0kbit d 20us m2 3kbit ul rate 128kbit | tc qdisc add dev eth1 parent 1:10 handle 100: fq_codel | tc qdisc add dev eth1 parent 1:20 handle 200: fq_codel | tc qdisc add dev eth1 parent 1:30 handle 300: fq_codel | tc qdisc add dev eth1 parent 1:40 handle 400: fq_codel | tc filter add dev eth1 parent 1: prio 1 protocol ip handle 1/0xff fw flowid 1:10 | tc filter add dev eth1 parent 1: prio 2 protocol ip handle 2/0xff fw flowid 1:20 | tc filter add dev eth1 parent 1: prio 3 protocol ip handle 3/0xff fw flowid 1:30 | tc filter add dev eth1 parent 1: prio 4 protocol ip handle 4/0xff fw flowid 1:40 | ifconfig ifb0 up txqueuelen 5 >&- 2>&- | tc qdisc del dev ifb0 root >&- 2>&- | tc qdisc add dev ifb0 root handle 1: hfsc default 30 | tc class add dev ifb0 parent 1: classid 1:1 hfsc sc rate 1024kbit ul rate 1024kbit | tc qdisc del dev eth1 ingress >&- 2>&- | tc qdisc add dev eth1 ingress | tc filter add dev eth1 parent : protocol ip prio 1 u32 match u32 0 0 flowid 1:1 action connmark action mirred egress redirect dev ifb0 | tc class add dev ifb0 parent 1:1 classid 1:10 hfsc rt m1 232kbit d 1907us m2 102kbit ls m1 232kbit d 1907us m2 568kbit ul rate 1024kbit | tc class add dev ifb0 parent 1:1 classid 1:20 hfsc rt m1 533kbit d 1907us m2 512kbit ls m1 533kbit d 1907us m2 284kbit ul rate 1024kbit | tc class add dev ifb0 parent 1:1 classid 1:30 hfsc ls m1 0kbit d 10
Re: [OpenWrt-Devel] [OpenWrt-Users] [PROPOSAL] move OpenWrt codebase to Git and GitHub
On Wed, May 25, 2016 at 7:06 PM, Benjamin Henrion wrote: > On Tue, May 24, 2016 at 10:35 PM, Jo-Philipp Wich wrote: >> Hi, >> >> here's a few numbers we gathered with our buildbot setup: >> >> We currently need roughly 35GB per target when building OpenWrt plus the >> entire package world and currently there are roughly ~70 >> target/subtarget combinations in the OpenWrt tree. >> >> If fast build tests are desired then the only way to do so is by >> implementing incremental building which only works if there's enough >> space to retain all build trees at once which means there need to be >> about 2.5TB of storage available. > > A BTRFS volume with deduplication would help here? I wouldn't trust BTRFS with photo album of my cats... I had it running and couldn't compile OpenWrt on BTRFS volume because it ran out of space, it was a knows bug that small files used up more space than df and other tools saw... But I as an advanced OpenWrt user and beginner openwrt developer would love to see move to github, it would make things much, much easier, please go for it. ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] [OpenWrt-Users] [PROPOSAL] move OpenWrt codebase to Git and GitHub
Thanks for the numbers Jo. The current hello-world setup with drone.io was done on cheap SSD based VPS. That said, with some "optimizations" (or hacks if you want) I think it should be possible to have less powerful servers but more of them to do what is needed. For example, if one makes pull request for package A. Then for every target only the core system with package A and it's dependencies should be built. That said, if pull request is valid it will result with a successful build. We should avoid situations where somebody makes a patch for package A and if fails to build because package Z unrelated to package A is broken. Luka On Tue, May 24, 2016 at 10:35:42PM +0200, Jo-Philipp Wich wrote: > Hi, > > here's a few numbers we gathered with our buildbot setup: > > We currently need roughly 35GB per target when building OpenWrt plus the > entire package world and currently there are roughly ~70 > target/subtarget combinations in the OpenWrt tree. > > If fast build tests are desired then the only way to do so is by > implementing incremental building which only works if there's enough > space to retain all build trees at once which means there need to be > about 2.5TB of storage available. > > For only building all base systems without package feeds the entire > required space is around 800GB. > > A base system build currently requires 1 hour and 15 minutes on a > machine having a Xeon E3-1246 v3 4 core / 8 thread CPU with prepopulated > dl/, ccache and make -j8. > > A build of all packages from all feeds takes around 70 minutes on a Xeon > E5-2630 v3 8 core / 16 thread machine with 12GB ram and make -j16. > > HTH, > Jo > ___ > openwrt-devel mailing list > openwrt-devel@lists.openwrt.org > https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] [PROPOSAL] move OpenWrt codebase to Git and GitHub
On Tue, May 24, 2016 at 10:29:30AM -0700, David Lang wrote: > OpenWRT has already moved to using Git instead of SVN, No, it has not. To users is exposed the Git frontend while the commits are made to the SVN repo. > so why do they need to move from hosting the git repository themselves to > having it hosted on github? See the reasons below. > There can be a mirror of the repo on github (remember that git is a > Decentralized VCS) Also, we have discussed of having a mirror on our server and this is something that we can do. If everything happens on GitHub then I don't see a point in having clone on GitHub instead of a having the main repo on GitHub and having clone elsewhere. > > * GitHub and similar services will allow us to integrate more easily > > with other projects > > > > Here specifically I mean integration with modern CI. Here is an example > > of integration with drone.io [3][4]. At the moment this is only in the > > POC stage but what I'd like to do down the line is to: > > > > - build OpenWrt images for all architectures for every pull request > > - build OpenWrt package binary for every package pull request for all > > architectures and make it available for download > > > > - build and host OpenWrt qemu and/or Docker image for every pull request > > the build farm isn't large enough to do this Current one is not. > It's also not neccessary to move to github to be able to do this, it just > needs more systems in the build farm to be able to build things fast enough. With GitHub it will be able to see compile status of each pull request. If it is not GitHub or simmilar service then this would need to be developed and I think we have better things to do then that :) > > This will allow easy review of the work since flags will be shown in the > > pull request if the build was sucessful or not. Also, this will allow > > people to test changes without building the image and thus lowering the > > time that needs to be spent on maintenance work. > > > > If this proposal gets accepted I'll be sending out an email to get > > access to more build servers so this new build infrastructure can > > properly support the project in a timely fashion. > > why should providing more build servers be contingent on moving to a > commercial hosting provider vs running things themselves? IMO move to GitHub will allow us to manage contributions more easily and handle them in timely fashion. This, combined with other perks explained in my previous email is possible today without need to develop the features that others provide today. Luka ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] [PROPOSAL] move OpenWrt codebase to Git and GitHub
On Tue, May 24, 2016 at 10:51:46AM -0500, Eric Schultz wrote: > My free-software side worries about using something non-free like drone.io > for CI but this is a huge task certainly and I'm not sure a free tool would > meet everyone's needs (plus there's the huge added burden of maintenance). The drone.io is actually Apache 2.0 [1] and the example build was configured on a private machine. Luka [1] https://github.com/drone/drone ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] OpenWrt / LEDE
On 25 May 2016 at 10:09, mbm wrote: > The hackers email address represents the primary point of contact for > OpenWrt, particularly in regards to donations. Following the surprise LEDE > announcement, forwarding rules for @openwrt.org email addresses were > disabled. This was done to mitigate further damage to OpenWrt due to > misrepresentation, intentional or otherwise. Hackers e-mail address (mailing list) was also used for internal discussions. You not only disabled forwarding rules for @openwrt.org personal e-mails but also kicked out private e-mails from the hackers mailing list. I never really cared about hardware donations offered to hackers, but knowing what's going on (like release plans) is important for contributing. -- Rafał ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] OpenWrt / LEDE
mbm writes: > The hackers email address represents the primary point of contact for > OpenWrt, particularly in regards to donations. Following the surprise > LEDE announcement, forwarding rules for @openwrt.org email addresses > were disabled. This was done to mitigate further damage to OpenWrt due > to misrepresentation, intentional or otherwise. Failing to see the damage your action has caused is your biggest problem right now. Even if we accept the rather far fetched possibilty of misrepresentation, there is no way that can outweight the effect on the maintainership status OpenWrt. Right now, 95 of the 145 (PKG_)MAINTAINER entries for OpenWrt packages points to an openwrt.org email address belonging to a LEDE committer: bjorn@canardo:/usr/local/src/openwrt$ git grep 'MAINTAINER:=.*<\(lynxis\|noltari\|dangole\|nbd\|hauke\|jow\|blogic\|neoraider\|rmilecki\|cyrus\|stintel\|thess\)@openwrt.org>' origin/master -- package/|wc -l 95 bjorn@canardo:/usr/local/src/openwrt$ git grep 'MAINTAINER' origin/master -- package/|wc -l 145 I don't know if all these were disabled, but the package I tried to submit to after the split was one of these. You don't seem to understand the devastating effect it has on OpenWrt if occasional contributors gets an email bounce from the published maintainer address. There is no way you can blame those maintainers for this situation. The problem is solely the responsibility of whoever decided to disable those addresses. Bjørn ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] OpenWrt / LEDE
Date: Tue, 24 May 2016 17:56:38 -0400 From: Daniel Dickinson To: OpenWrt Development List Subject: Re: [OpenWrt-Devel] OpenWrt / LEDE Message-ID: <1464126998.1239.137.camel@homehost> Content-Type: text/plain; charset="UTF-8" [snip] Let's see if any of the remaining OpenWrt devs at least publicly support adopting them or some variation of them. As I've said before my impression is that LEDE-style rules are not all that welcomed (and that's based on the interactions I saw on the private openwrt channels when I was a developer, not just a pure outsider view; it's possible my impression is wrong, but the toxicity described previously was in large part negative reaction by folkds in the LEDE team to toxic comments from at least one of the remaining OpenWrt devs; certainly it damaged my opinion of him (although the toxicity also damaged my opinion of a couple of LEDE folks too)). Sigh... It's not as if LEDE offered the changes to OpenWrt and we voted against them causing a split -- there was literally no discussion and no warning prior to the public announcement. LEDE just suddenly existed and the story quickly became that the reason for their existence was because OpenWrt somehow prevented them from making changes. At no point did OpenWrt veto changes or even have the option to; the truth is that we agree changes need to be made. By not including OpenWrt in the discussions LEDE ran afoul of their own transparency leading to the false impression that OpenWrt was somehow against the changes, causing a split in the community in terms of LEDE vs OpenWrt with various amounts of hostility on the mailing lists. None of this is healthy or constructive. Let me be very clear: nobody on the OpenWrt side is against the changes LEDE is trying to make. It is our position that this whole thing is a misunderstanding and that the projects should attempt to merge again. I have been trying to arrange talks between the two sides but between work schedules, timezone conflicts and FUD regarding the split it's been very difficult. - mbm ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] OpenWrt / LEDE
Date: Tue, 24 May 2016 23:26:37 +0200 From: Rafał Miłecki To: Hauke Mehrtens Cc: OpenWrt Development List, LEDE Development List Subject: Re: [OpenWrt-Devel] OpenWrt / LEDE Message-ID: Content-Type: text/plain; charset=UTF-8 [snip] *However* I'd like to maintain 15.05 OpenWrt branch for some time (few months?). Unfortunately I feel unsure about my access to OpenWrt repo in the future. First some @openwrt.org e-mails were deleted/disabled. That made me ask about my commiting permissions: [2016-05-05] [14:41:32] [mbm]: Kaloz: can we still commit to OpenWrt? [2016-05-05] [14:45:28] as far as I know you can [2016-05-05] [16:21:09] rmilecki: yes it looked fine, but few days later I was kicked out of openwrt-hackers@ mailing list silently. There is no relation between email addresses and commit access. At no point was commit access revoked for any LEDE members nor have any email messages been deleted. You are encouraged to continue contributing. The hackers email address represents the primary point of contact for OpenWrt, particularly in regards to donations. Following the surprise LEDE announcement, forwarding rules for @openwrt.org email addresses were disabled. This was done to mitigate further damage to OpenWrt due to misrepresentation, intentional or otherwise. It is hoped that the projects may yet merge and the email access will be restored. - mbm ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] [LEDE-DEV] OpenWrt / LEDE
Zoltan HERPAI writes: >> On 05/24/2016 10:31 PM, Hauke Mehrtens wrote: >> >> This is my personal opinion and this was not somehow internally planned >> with other LEDE people. > > If I start a discussion about my employer-related topics along a beer > with a couple friends, that's a private discussion with personal > opinions. If I do it on any public channel, I can be felt to represent > my employer on that topic. You seem to be representing LEDE. No. This is basic netiquette which you are assumed to know *before* posting to a puvblic mailing list. You can find it "everywhere" on the Internet. One good source is for example this informational RFC: https://tools.ietf.org/html/rfc1855 Quoting from "3.1.1 General Guidelines for mailing lists and NetNews" : - Assume that individuals speak for themselves, and what they say does not represent their organization (unless stated explicitly). Bjørn ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
Re: [OpenWrt-Devel] [PATCH] bcm53xx: Move SF mmap patch
Marek Vasut - 17:12 24.05.16 wrote: > On 05/24/2016 04:57 PM, Felix Fietkau wrote: > > On 2016-05-24 16:48, Marek Vasut wrote: > >> The patch adding SPI flash mmap read capability does not compile due to > >> missing > >> m25p80_rx_nbits() function. Move it to bcm53xx patch directory, where the > >> patch > >> adding this m25p80_rx_nbits() function resides. > > This doesn't make any sense to me. The function is already in the driver > > in 4.4, it is not added by any patch... > > > > - Felix > > > Well is there any way to obtain kernel tree with the > target/linux/generic/patches-4.4 applied, so I can base socfpga patches > on top > of it ? > > I tried git am on those patches, but that's obviously not possible, > since a lot of these patches were not generated with git-format-patch > or the relevant header was removed. Quilt work fine: https://wiki.openwrt.org/doc/devel/patches#adding_or_editing_kernel_patches ___ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel