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
<GitHub36> [smoltcp] whitequark pushed 2 new commits to master: https://git.io/vdTXr
<GitHub36> smoltcp/master 440b6f5 whitequark: Reorganize features using namespaces, to match module hierarchy....
<GitHub36> smoltcp/master 17cec87 whitequark: Allow disabling any of: raw, TCP or UDP sockets.
<travis-ci> m-labs/smoltcp#270 (master - 17cec87 : whitequark): The build was broken.
<GitHub160> [smoltcp] whitequark force-pushed master from 17cec87 to a983c62: https://git.io/vMLjV
<GitHub160> smoltcp/master a983c62 whitequark: Allow disabling any of: raw, TCP or UDP sockets.
<travis-ci> m-labs/smoltcp#271 (master - a983c62 : whitequark): The build was fixed.
<GitHub147> [smoltcp] whitequark commented on commit a983c62: cc @briansmith https://git.io/vdT1W
<GitHub144> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vdT1a
<GitHub144> smoltcp/master cedf66a whitequark: Enforce some lints.
<travis-ci> m-labs/smoltcp#272 (master - cedf66a : whitequark): The build was broken.
<GitHub84> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vdT17
<GitHub84> smoltcp/master 16e6e70 whitequark: Fix many warnings.
<travis-ci> m-labs/smoltcp#273 (master - 16e6e70 : whitequark): The build is still failing.
<GitHub135> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vdT1N
<GitHub135> smoltcp/master a7cdcd4 whitequark: Update CI config to not use poorly interacting feature sets.
<travis-ci> m-labs/smoltcp#274 (master - a7cdcd4 : whitequark): The build was fixed.
<GitHub49> [smoltcp] whitequark commented on issue #11: I'm going to closing this PR because it seems like everything except the actual ARP cache iteration is already done, and iteration itself is something I would like to do quite differently. Feel free to open an issue requesting that. https://git.io/vdTMJ
<GitHub3> [smoltcp] whitequark closed pull request #11: implement IntoIterator for SliceCache (master...master) https://git.io/vyRvR
<GitHub144> [smoltcp] whitequark commented on issue #40: Done. https://git.io/vdTMT
<GitHub188> [smoltcp] whitequark closed issue #40: Replace hardcoded constants with associated items https://git.io/v56GP
rohitksingh_work has joined #m-labs
sb0 has quit [Quit: Leaving]
_whitelogger has joined #m-labs
sb0 has joined #m-labs
mumptai has joined #m-labs
<travis-ci> batonius/smoltcp#87 (default_route - d6f204d : Egor Karavaev): The build passed.
<GitHub13> [smoltcp] batonius commented on pull request #44 d6f204d: Apparently not, I've made it explicit. https://git.io/vdkeX
<GitHub18> [smoltcp] batonius commented on pull request #44 d6f204d: Subnets have their own semantics, for example, 192.168.1.0/24 is exactly the same subnet as 192.168.1.255/24, this makes accepting a slice of subnets as interface addresses somewhat confusing. I've changed the documentation to better reflect the fact that CIRD != subnet, but I'd prefer to keep `Cidr` as the name. https://git.io/vdkv3
<GitHub165> [smoltcp] batonius commented on pull request #44 d6f204d: I've rephrased it so the constructors check prefix_len and can fail now. https://git.io/vdkv8
<GitHub115> [smoltcp] batonius commented on pull request #44 d6f204d: I was thinking about adding `contains_subnet` as well. https://git.io/vdkfl
<travis-ci> batonius/smoltcp#88 (master - a7cdcd4 : whitequark): The build passed.
<GitHub161> [smoltcp] whitequark commented on pull request #44 d6f204d: This is a really odd case, because so far every other data structure in `wire` has all fields `pub`. It's even in the module documentation. Thoughts? https://git.io/vdkJd
<travis-ci> batonius/smoltcp#89 (default_route - f02bb34 : Egor Karavaev): The build was broken.
<travis-ci> batonius/smoltcp#90 (default_route - 59d883a : Egor Karavaev): The build passed.
<GitHub51> [smoltcp] batonius commented on pull request #44 59d883a: The idea was to prevent the creation of inconsistent values, instead of using `is_valid` method, since CIDRs have an invariant that needs to be enforced. https://git.io/vdkTq
mumptai has quit [Remote host closed the connection]
<sb0> whitequark, do you have an idea about those errors:
<sb0> Dwarf Error: Cannot find DIE at 0xa271 referenced from DIE at 0xe61a [in module /home/sb/ionpak/firmware/target/thumbv7em-none-eabihf/release/ionpak-firmware]
<sb0> from gdb?
<sb0> those kick in intermittently and are very annoying, it won't flash the board after that, and they happen on some compilations and not with others, even from the same source
<sb0> oh I can just flash with openocd directly, maybe
<sb0> none of this gdb bugware
<sb0> ha yes, works nicely.
<GitHub196> [ionpak] sbourdeauducq pushed 1 new commit to master: https://github.com/m-labs/ionpak/commit/416ac30496ce4d83f0d4a7a1649ad2eb1dc2f2e2
<GitHub196> ionpak/master 416ac30 Sebastien Bourdeauducq: do not use GDB for loading firmware...
<mithro> whitequark: in https://github.com/m-labs/misoc/commit/d625c435910166652553090b5f703b11e005ffc6 why did you reduce the exception stack size down to 128 bytes (from 256 bytes)?
<mithro> It seems changing that value back to 256 seems to fix an issue we are having with the or1k but we don't quite understand why....
<mithro> (the issue being random lock ups)
sb0 has quit [Quit: Leaving]
rohitksingh_work has quit [Read error: Connection reset by peer]
sb0 has joined #m-labs
jbqubit has joined #m-labs
rohitksingh has joined #m-labs
<jbqubit> bb-m-labs: force build --props=package=artiq-kc705-phaser artiq-board
<bb-m-labs> build forced [ETA 13m54s]
<bb-m-labs> I'll give a shout when the build finishes
<bb-m-labs> build #785 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/785
rohitksingh has quit [Quit: Leaving.]
<sb0> jbqubit, installing phaser py_1295+gitc00b3fe8 shouldn't try to install artiq py_1183+gitf4624e08
<sb0> the versions are supposed to match
<jbqubit> That conda doesn't automatically select the latest version is odd. Could this be another conda bug?
<jbqubit> When I specify the latest version of artiq-kc705-phaser I still get a dependency error. https://gist.github.com/jbqubit/752df5cea9e0858ded57975f372e61bb
<sb0> did you add the m-labs main channel?
<sb0> pythonparser is in main but not dev
<sb0> and python-dateutil is in conda-forge, did you set that up as the doc tells you?
trxw has joined #m-labs
trxw has quit [Ping timeout: 260 seconds]
<jbqubit> I had only dev branch installed. I can see from the instructions that dev is additive. "To use the development versions of ARTIQ, /also/ add the dev label." Consider emphasizing that both dev and main
<jbqubit> repositories are required. After adding main repository there are still unsatisfied dependencies. https://gist.github.com/jbqubit/0fad84adec1668ad1c481636585b9d28
<sb0> yeah you have some old artiq-dev installed, remove it
<jbqubit> I wiped and reinstalled Anaconda this morning.... Just now created a new conda environment and see continued trouble.
<_florent_> sb0: I'll retry installing artiq soon and will ask here if I need help
mumptai has joined #m-labs
<_florent_> sb0: i just tried and got it working with the last instructions.
<jbqubit> _florent_: are you using 64-bit Windows?
<whitequark> mithro: hmmm, no idea, it was 2 years ago
<whitequark> feel free to revert
<mithro> whitequark: Do you understand that code at all? -- IE From what I can see only 116 bytes of the stack are used for saving registers later on, so I don't know why it needs 256?
<whitequark> mithro: I can't remember a single thing about it anymore, no
<whitequark> but it looks like my motivation was that
<whitequark> maybe it's related to the red zone?
<mithro> whitequark: I've heard references to this "red zone" what is it?
<whitequark> mithro: so normally a function only accesses stack slots that are below SP
<whitequark> erm
<whitequark> above SP
<whitequark> and it has to decrement SP in the prologue
<whitequark> however, in some ABIs, leaf functions can access a certain number of bytes (usually 128) below SP without adjusting SP
<whitequark> ... yeah
<whitequark> it was probably the red zone.
<whitequark> let me recheck one thing
<travis-ci> m-labs/rust-atomic_ring_buffer#6 (master - 4ac6645 : whitequark): The build was broken.
<whitequark> mithro: are you sing cc?
<whitequark> using gcc*
<mithro> whitequark: Yes
<whitequark> oh ok
<whitequark> can you show me the assembly it generates at -O0 for "int foo() { int bar = 0; return bar; }"
<whitequark> with -fomit-frame-pointers
<mithro> whitequark: Hrm, it doesn't seem to know about -fomit-frame-pointers ?
<mithro> Oh it doesn't have an s
<whitequark> wow that code is garbage
<whitequark> but besides that, no, it doesn't use red zone
<mithro> whitequark: So, any idea how to debug why changing that to 8*32 fixes out issues?
<whitequark> mithro: hmmm
<whitequark> mithro: I would split that macro in two
<whitequark> for e.g. timer use the broken version, for all others use the working one
<whitequark> and s ee why it crashes
<travis-ci> batonius/smoltcp#91 (default_route - aba5b07 : Egor Karavaev): The build passed.
mumptai has quit [Remote host closed the connection]
gric has quit [Quit: leaving]
gric has joined #m-labs
<GitHub161> [artiq] dhslichter opened issue #833: ARTIQ 3 release candidate? https://github.com/m-labs/artiq/issues/833