sb0 changed the topic of #m-labs to: ARTIQ, Migen, MiSoC, Mixxeo & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
jbqubit has quit [Ping timeout: 260 seconds]
<GitHub108> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vQQMP
<GitHub108> smoltcp/master d9567b1 whitequark: Implement loopback interfaces....
<ashafir> is there any example or hint how to use smoltcp without alloc?
<whitequark> ashafir: I'm writing one right now
<GitHub88> [smoltcp] whitequark force-pushed master from d9567b1 to 8ada283: https://git.io/vMLjV
<GitHub88> smoltcp/master 8ada283 whitequark: Implement loopback interfaces....
<whitequark> it'll be ready soon
<GitHub79> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vQQDD
<GitHub79> smoltcp/master 9a4efcd whitequark: LoopbackInterface → Loopback.
<travis-ci> m-labs/smoltcp#142 (master - 8ada283 : whitequark): The build was fixed.
<ashafir> add the cargo.toml example for the features list to the comments
<travis-ci> m-labs/smoltcp#143 (master - 9a4efcd : whitequark): The build was fixed.
<whitequark> ashafir: what do you mean?
<ashafir> Ideally a compilable ARM baremetal+noalloc+dummy_phy project example
<ashafir> but not just a server.rs like now
ashafir has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
ashafir has joined #m-labs
<whitequark> ah, sure
<reportingsjr> whitequark: any idea about how much RAM and flash smoltcp need?
<reportingsjr> needs*
ashafir has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
ashafir has joined #m-labs
<whitequark> reportingsjr: RAM: exactly the spaces you dedicate for its buffers and control structures
<whitequark> the TCP control block is, let's see, about 100 bytes
<whitequark> it could conceivably be squished
<whitequark> quite easily actually
rohitksingh_work has joined #m-labs
<GitHub189> [smoltcp] whitequark opened issue #26: Logged retransmit time is 2x of the actual timeout used https://git.io/vQQ7p
<GitHub46> [smoltcp] whitequark opened issue #27: 3-way handshake must not ever be combined with FIN https://git.io/vQQ5k
<GitHub89> [smoltcp] whitequark opened issue #28: Unspecified local address should be valid when connecting https://git.io/vQQ5L
<GitHub148> [smoltcp] whitequark pushed 2 new commits to master: https://git.io/vQQ5s
<GitHub148> smoltcp/master 042019d whitequark: Remove default impl for Device::limits()....
<GitHub148> smoltcp/master 15ce667 whitequark: Add a bare-metal usage example.
<GitHub60> [smoltcp] whitequark opened issue #29: Infinite challenge ACK retransmit loop https://git.io/vQQ5z
<whitequark> reportingsjr: try building https://github.com/m-labs/smoltcp#bare-metal-usage-examples for your architecture and see?
<travis-ci> m-labs/smoltcp#144 (master - 15ce667 : whitequark): The build passed.
<whitequark> ashafir: regarding phy, here is an example phy: https://docs.rs/smoltcp/0.3.0/smoltcp/phy/index.html
<cr1901_modern> whitequark: https://github.com/m-labs/smoltcp/blob/master/examples/loopback.rs#L44-L47 I'm not sure how allocating statically helps you w/ UB if your stack can potentially crash into your .data/.bss section anyway after allocating static bufs
<whitequark> cr1901_modern: stack probes
<whitequark> they're not in for ARM yet though so I can't do that on stock rustc
<whitequark> well that or an
<whitequark> MPU
<cr1901_modern> stack probe will detect when the stack crashes into the data section (and not just stack corruption; perfectly possible to crash into data section but you "just haven't used that data yet")?
<whitequark> stack probe will *prevent* any corruption
<GitHub136> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vQQdJ
<GitHub136> smoltcp/master c5fc8f7 whitequark: Document the loopback.
froggytoad has joined #m-labs
<travis-ci> m-labs/smoltcp#145 (master - c5fc8f7 : whitequark): The build passed.
<cr1901_modern> Ahhh. Then last q; would allocating those buffers on the stack w/ stack probing have any disadvantages?
sb0 has quit [Quit: Leaving]
froggytoad has quit [Ping timeout: 248 seconds]
<travis-ci> batonius/smoltcp#31 (master - c5fc8f7 : whitequark): The build passed.
<travis-ci> batonius/smoltcp#32 (packet_dispatch - 749aa9f : Egor Karavaev): The build was broken.
<travis-ci> batonius/smoltcp#33 (packet_dispatch - 831c363 : Egor Karavaev): The build was fixed.
<whitequark> cr1901_modern: yes
ashafir has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
ashafir has joined #m-labs
rohitksingh_work has quit [Read error: Connection reset by peer]
jbqubit has joined #m-labs
ashafir has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
ashafir has joined #m-labs
jaeckel has quit [Quit: Goodbye Cruel World]
jaeckel has joined #m-labs
sb0 has joined #m-labs
<GitHub12> [artiq] jbqubit opened issue #782: name space for platform.add_extension() https://github.com/m-labs/artiq/issues/782
<GitHub46> [artiq] jordens commented on issue #782: What do you mean by "namespace", how would the help and what can you do with them that you can't do right now? "i2c" is the only i2c in that kc705 platform. In your platform with more i2c buses, you can call them e.g. "i2c_eem0" and "i2c". https://github.com/m-labs/artiq/issues/782#issuecomment-315362659
<GitHub154> [pdq] jordens pushed 5 new commits to master: https://github.com/m-labs/pdq/compare/bbc9288f165d...ff5b7146add4
<GitHub154> pdq/master 335180f Robert Jordens: README: rtd: pdq2 -> pdq
<GitHub154> pdq/master dc67266 Robert Jordens: README, make: describe channel number and build configurations...
<GitHub154> pdq/master 312a509 Robert Jordens: README: add notes about conda and ISE...
Gurty has quit [Ping timeout: 255 seconds]
<GitHub165> [pdq] jordens pushed 1 new commit to master: https://github.com/m-labs/pdq/commit/08104c9bbb3e3c111d8e3facb6c6d4387762371d
<GitHub165> pdq/master 08104c9 Robert Jordens: README: flashing...
Gurty has joined #m-labs
<GitHub143> [pdq] jordens pushed 2 new commits to master: https://github.com/m-labs/pdq/compare/08104c9bbb3e...f7d75222c5f9
<GitHub143> pdq/master f7d7522 Robert Jordens: examples: add PDQ ARTIQ example...
<GitHub143> pdq/master 8d219f1 Robert Jordens: doc: copyright year
froggytoad has joined #m-labs
ashafir has quit [Quit: http://www.kiwiirc.com/ - A hand crafted IRC client]
<GitHub199> [pdq] dhslichter opened pull request #13: Fix maximum number of frames (master...patch-1) https://github.com/m-labs/pdq/pull/13
<GitHub191> [pdq] jordens pushed 3 new commits to master: https://github.com/m-labs/pdq/compare/f7d75222c5f9...c10b398fb366
<GitHub25> [pdq] jordens closed pull request #13: Fix maximum number of frames (master...patch-1) https://github.com/m-labs/pdq/pull/13
<GitHub191> pdq/master c10b398 Robert Jordens: doc: fix b6f8120...
<GitHub191> pdq/master 4fb917b dhslichter: Merge b6f8120c73a95ed55c76f8f4292bc9f4e7d4638e into f7d75222c5f91f905a2c29dc1a6c3864c746bdb3
<GitHub191> pdq/master b6f8120 dhslichter: Fix maximum number of frames
<GitHub182> [pdq] jordens pushed 2 new commits to master: https://github.com/m-labs/pdq/compare/c10b398fb366...335fc08408ae
<GitHub182> pdq/master 335fc08 Robert Jordens: doc: unused frame bits...
<GitHub182> pdq/master 5bdcdc3 Robert Jordens: doc: also document frame number in register info
<jbqubit> bb-m-labs: force build --props=package=artiq-dev artiq-board
<bb-m-labs> build forced [ETA 13m04s]
<bb-m-labs> I'll give a shout when the build finishes
<bb-m-labs> build #720 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/720
<GitHub47> [artiq] jbqubit opened issue #783: can't build firmware https://github.com/m-labs/artiq/issues/783
<jbqubit> Looks like bb did a trivial build an not a full build of artiq-dev. What's the right way to rebuild artiq-dev? It's stale... June 4.
<GitHub14> [artiq] jordens commented on issue #783: ```... https://github.com/m-labs/artiq/issues/783#issuecomment-315401287
<sb0> jbqubit, what do you mean it's stale? artiq-dev is rebuilt at every commit.
<GitHub168> [artiq] whitequark commented on issue #783: But the bug exists separately. It's related to ARTIQ firmware wishing to know the hash of the git checkout. https://github.com/m-labs/artiq/issues/783#issuecomment-315401497
<sb0> you're probably looking at the release-2 version
<rjo> jbqubit: they are all up to date in conda.
<jbqubit> sb0: when I run $ conda list artiq
<jbqubit> I see artiq-dev 3.0.dev py_1075+gitfdb4e3f8cfad m-labs/label/dev
<jbqubit> which points of a commit June 4
<rjo> jbqubit: that's what you installed. that's your choice.
<jbqubit> Here's what my conda info looks like. https://gist.github.com/jbqubit/7cb9037df0410293bdc18c2c6d57b18a
<rjo> jbqubit: conda search artiq?
<jbqubit> I'd like to be using the latest build!
<jbqubit> (artiq3test7) britton@britton1:~/artiq3test7$ conda update artiq-dev Fetching package metadata ............... Solving package specifications: . # All requested packages already installed. # packages in environment at /home/britton/anaconda3/envs/artiq3test7: # artiq-dev 3.0.dev py_1075+gitfdb4e3f8cfad m-labs/label/dev
<jbqubit> You're right... $ conda search artiq-dev shows that there are more recent build but $ conda update fails to install them
<rjo> conda create -c m-labs/label/dev -n test works fine here
<rjo> jbqubit: this is conda prioritizing main packages over dev in the artiq-dev dependencies.
<GitHub184> [artiq] jordens commented on issue #783: ```conda create -c m-labs/label/dev -n artiq-dev-test artiq-dev``` works fine here. https://github.com/m-labs/artiq/issues/783#issuecomment-315404065
<jbqubit> this is what my .condarc looks like
<jbqubit> Should it make no reference whatsoever to main branch?
<rjo> dunno. does specifying the channel during install work or do we need to research that question?
<sb0> jbqubit, use this as condarc http://paste.debian.net/976530/
<jbqubit> reinstalling anaconda from scratch...
<rjo> jbqubit: that sounds like a knee jerk reaction ;)
<GitHub54> [artiq] jbqubit opened issue #784: conda asyncserial missing https://github.com/m-labs/artiq/issues/784
<rjo> come on...
<jbqubit> ya, after filing the Issue I clone it from github/m-labs and used setup.py to install the missing asyncserial dependencey.
<jbqubit> It's a bug because it's not done correctly by conda.
<GitHub39> [artiq] jbqubit opened issue #785: conda version dependency not in meta.yaml https://github.com/m-labs/artiq/issues/785
froggytoad has quit [Ping timeout: 260 seconds]
<jbqubit> user phaser as starting point... the following compiles but is missing RTIO channels for the differential TTLs.
<jbqubit> Any ideas what I'm doing wrong?
<jbqubit> Also, it would be helpful to have a reference design for an ARTIQ device binary builder for KC705 suitable for toggling TTLs but nothing else.
<GitHub56> [artiq] jbqubit closed issue #783: can't build firmware https://github.com/m-labs/artiq/issues/783
<GitHub54> [artiq] jbqubit commented on issue #783: Conda was stuck on an old version artiq-dev package. That is, conda update didn't work. Nor did conda install=version-no. So, erased ~/anaconda3 and reinstalled anaconda from scratch. And addressed #785. Now I can build. Closing. https://github.com/m-labs/artiq/issues/783#issuecomment-315458687
<GitHub92> [artiq] jbqubit commented on issue #784: Why not include asyncserial in conda package artiq-dev? https://github.com/m-labs/artiq/issues/784#issuecomment-315459792
<GitHub185> [artiq] jbqubit commented on issue #782: I2C wasn't the best example. Consider instead artiq/gateware/nist_clock.py which creates an "spi". This very nearly collides with migen/build/platforms/kc705.py "spiflash" and "mmc_spi". ... https://github.com/m-labs/artiq/issues/782#issuecomment-315467071
<GitHub199> [artiq] whitequark commented on issue #785: We are not necessarily aware of every minor bug in conda, of which there are many. It is always a good idea to keep conda updated. https://github.com/m-labs/artiq/issues/785#issuecomment-315467575
<GitHub178> [artiq] jbqubit commented on issue #785: Adding the conda version constraint will save everybody time as it will flag an outdated version of conda. What's the problem with doing that in meta.yaml? https://github.com/m-labs/artiq/issues/785#issuecomment-315469391
<GitHub148> [artiq] jordens commented on issue #782: Simply because the acronym "i2C" (or "SPI") occurs at all levels and in all repositories! Just because they refer to the same acronym doesn't mean "namespaces" are in order. Anyway, what do you mean by "namespace". And still the same three questions as above. https://github.com/m-labs/artiq/issues/782#issuecomment-315475271
<GitHub67> [artiq] jordens commented on issue #782: What do you mean by "namespace", how would they help and what can you do with them that you can't do right now? "i2c" is the only i2c in that kc705 platform. In your platform with more i2c buses, you can call them e.g. "i2c_eem0" and "i2c". https://github.com/m-labs/artiq/issues/782#issuecomment-315362659
<GitHub35> [artiq] jbqubit commented on issue #780: Can you give me a hand? See attached for the patch of what I've got. I'm modifying the phaser build for kc705 as a starting point. Several observations.... https://github.com/m-labs/artiq/issues/780#issuecomment-315475800
<GitHub182> [artiq] jbqubit commented on issue #782: Thank you for reply. I don't have more to add. https://github.com/m-labs/artiq/issues/782#issuecomment-315477712
<GitHub2> [artiq] jbqubit closed issue #782: name space for platform.add_extension() https://github.com/m-labs/artiq/issues/782
<GitHub120> [artiq] npisenti commented on issue #782: Perhaps the appropriate way forward here is name mangling -- top-level user io mappings could be conventionally prefaced with a meaningful designator which is orthogonal to the underlying artiq/migen/misoc platform extensions. ... https://github.com/m-labs/artiq/issues/782#issuecomment-315480994
froggytoad has joined #m-labs