<GitHub71>
[artiq] whitequark commented on issue #637: The reproduction steps we have:... https://git.io/v1lF0
<GitHub8>
[artiq] klickverbot commented on issue #637: From working with embedded lwIP on the Zynq before, I'd be inclined to triple-check that the assumptions made by its API are followed (e.g. w.r.t. send buffer lifetimes and the copying flags). The lifetime mechanics are sometimes poorly documented to the point where they could be called buggy under a less charitable interpretation.... https://git.io/v1lFA
rohitksingh_work has joined #m-labs
<GitHub109>
[artiq] whitequark commented on issue #637: > I'd be inclined to triple-check that the assumptions made by its API are followed (e.g. w.r.t. send buffer lifetimes and the copying flags).... https://git.io/v1lbk
stekern has joined #m-labs
<GitHub146>
[artiq] whitequark pushed 1 new commit to master: https://git.io/v1lNI
<GitHub146>
artiq/master 4c94873 whitequark: runtime: upgrade lwip to release 2.0.0.
<GitHub152>
[artiq] sbourdeauducq commented on commit 4c94873: What is that for? https://git.io/v1lNY
<GitHub58>
[artiq] whitequark commented on commit 4c94873: You have `-Wmissing-prototypes` enabled somewhere, I'm not sure for what reason. The warnings are useful when upgrading lwip because they remove APIs aggressively, and this missing prototype warning reduces SNR. https://git.io/v1lNC
<whitequark>
rjo: I salvaged the approach I used earlier for casts.
<GitHub117>
[artiq] whitequark pushed 1 new commit to master: https://git.io/v1lNl
<GitHub117>
artiq/master 4305903 whitequark: compiler: rein in overzealous cast monomorphization....
<GitHub58>
[artiq] sbourdeauducq commented on commit 4c94873: Then shouldn't we have those functions ``static``, or their prototypes moved to a header file? https://git.io/v1lNR
<GitHub199>
[artiq] whitequark commented on commit 4c94873: These functions are only used from Rust. It doesn't make any real sense to declare them in a header file, except to make changing their prototype more noisy. https://git.io/v1lNo
<GitHub87>
[artiq] sbourdeauducq commented on commit 4c94873: OK so this is a Rust problem. In this case we should just drop ``-Wmissing-prototypes``. https://git.io/v1lNy
<whitequark>
hm, wait, that wasnt legal
<GitHub79>
[artiq] whitequark pushed 2 new commits to master: https://git.io/v1lAW
<GitHub79>
artiq/master 218720c whitequark: Revert "compiler: rein in overzealous cast monomorphization."...
<GitHub79>
artiq/master f68e4ae whitequark: compiler: rein in overzealous cast monomorphization.
<sb0>
whitequark, yes, had the same issue when i tried upgrading to 2.0
<sb0>
did you get all the branches?
<whitequark>
I pushed the tag
<whitequark>
and it actually checked out the submodules on the artiq builder
<whitequark>
bb-m-labs: force build artiq
<bb-m-labs>
build #1146 forced
<bb-m-labs>
I'll give a shout when the build finishes
<GitHub181>
[artiq] whitequark pushed 1 new commit to master: https://git.io/v1lAx
<GitHub181>
artiq/master 668928a whitequark: runtime: fix a lifetime issue in lwip::Pbuf....
<whitequark>
hrm, test_pulse_rate_dds passes after 668928a
<whitequark>
so that could have been the rust binding issue after all...
<whitequark>
let's see if that fixes the other obscure crashes too.
<GitHub33>
[artiq] whitequark commented on issue #637: I've fixed a lifetime issue (interestingly, on the Rust side exclusively) in 668928a, let's see if that helps. https://git.io/v1lAj
rohitksingh_work has quit [Ping timeout: 250 seconds]
rohitksingh_wor1 has joined #m-labs
kuldeep has quit [Ping timeout: 268 seconds]
kuldeep has joined #m-labs
kuldeep has quit [Ping timeout: 260 seconds]
FabM has joined #m-labs
kuldeep has joined #m-labs
kuldeep has quit [Ping timeout: 250 seconds]
kuldeep has joined #m-labs
ohama has quit [Ping timeout: 244 seconds]
ohama has joined #m-labs
<GitHub38>
[artiq] jordens commented on issue #570: No. We just have to dig up all the contributors and have them signal consent. https://git.io/v18Iz
<cr1901_modern>
sb0: I fixed *a* bug in the SPI core. I am fairly confident this will close #615, but because I couldn't duplicate jmizrahi's results exactly, I'm holding off on saying "#615 can be closed for sure". Also, I broke one of the core's unit tests (I know why it broke), and I want to discuss before making final changes.
<cr1901_modern>
Ehhh, I fixed the unit test, but at the cost of making data valid one system clock cycle later than before my changes. If this is a dealbreaker, I have a few options. We can talk later.
sandeepkr has joined #m-labs
<GitHub118>
[artiq] sbourdeauducq pushed 7 new commits to master: https://git.io/v18Y4
<GitHub38>
[artiq] cr1901 commented on issue #615: @jmizrahi Okay, the following patches to MiSoC and ARTIQ should (hopefully) fix your issue:... https://git.io/v18co
<cr1901_modern>
I made a patch as a gist.
Felix__ is now known as felix_
rohitksingh_wor1 has quit [Read error: Connection reset by peer]
rohitksingh has joined #m-labs
<whitequark>
sb0: ok
<GitHub169>
[artiq] whitequark commented on issue #637: 668928a has not fixed anything (more or less as I expected; the code in sched.rs was mildly suspicious but ultimately it did the same thing). https://git.io/v185H
<sb0>
whitequark, there's gradator on #electrolab who seems knowledgeable about lwip.
<whitequark>
ok...
sandeepkr has quit [Ping timeout: 260 seconds]
<whitequark>
sb0: looks like the TMPU doesn't work properly
<whitequark>
it does seem to trap on write transactions but not read transactions
<sb0>
it ignores the write bit, so that would be a mor1kx problem
<whitequark>
sigh
<whitequark>
that makes it rather useless
<whitequark>
wonder if mor1kx can be fixed... but not now at least
<GitHub172>
[artiq] whitequark commented on issue #544: mor1kx ignores wb.err on read cycles, so looks like this will not work so easily. https://git.io/v18h4
<GitHub119>
[artiq] whitequark commented on issue #637: > Integrate the trivial MPU and become really sure that Rust code is not clobbering lwIP by accident with its stacks (as the method I used to check before now is unreliable).... https://git.io/v18jf
<GitHub175>
[artiq] whitequark commented on issue #637: I've tried always copying input data in the call to tcp_write. This did not improve anything. https://git.io/v14vL
<GitHub13>
[artiq] sbourdeauducq pushed 1 new commit to master: https://git.io/v14JL