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
<GitHub> [artiq] whitequark pushed 2 new commits to master: https://github.com/m-labs/artiq/compare/715443e211c4...c2cf60e7f65c
<GitHub> artiq/master c2cf60e whitequark: firmware: use the cslice crate to get rid of unsafe code in ksupport.
<GitHub> artiq/master 9a9a3aa whitequark: firmware: remove unnecessary inline assembly in libboard.
<bb-m-labs> build #1286 of artiq is complete: Failure [failed lit_test] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1286 blamelist: whitequark <whitequark@whitequark.org>
<GitHub> [artiq] whitequark pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/0df8a241350702b73188d10f2f99688b92e3099e
<GitHub> artiq/master 0df8a24 whitequark: Fix c2cf60e.
<bb-m-labs> build #1287 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1287 blamelist: whitequark <whitequark@whitequark.org>
<GitHub> [artiq] whitequark pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/3b54736c02533156a8adfbd7934d848f37fbe120
<GitHub> artiq/master 3b54736 whitequark: firmware: port flash storage management to Rust.
<GitHub> [artiq] whitequark commented on issue #636: > register-pinning of now... https://github.com/m-labs/artiq/issues/636#issuecomment-276871986
<GitHub> [artiq] whitequark commented on issue #544: Moving from #569; this will also need stack probe fixes in LLVM (https://reviews.llvm.org/D9653) to work correctly. https://github.com/m-labs/artiq/issues/544#issuecomment-276873299
<sb0> whitequark, #569 "there will still be -lcompiler-rt+-lalloc needed by Rust and -lbase needed by lwip" << can -lbase go?
<whitequark> sb0: needs the uart junk ported
<whitequark> and, hm, something else, let me check
<whitequark> ah yes. liballoc also depends on -lbase. and there are a few strlen invocations that should die once the compiler finally uses pascal strings.
<bb-m-labs> build #380 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/380
<bb-m-labs> build #1288 of artiq is complete: Failure [failed python_unittest_2] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1288 blamelist: whitequark <whitequark@whitequark.org>
bentley` has quit [Ping timeout: 256 seconds]
kilae has joined #m-labs
<sb0> ooooh
<sb0> the si5324 isn't properly locking
<sb0> that can explain the DRTIO fuckups
<rjo> sb0: got about 10 more codenames for sinara?
<sb0> let me see
<sb0> novogorny, berdenish, bolshaya, sary, urukul, zotino, koster, mirny, selki, kirdy
<whitequark> sb0: ...
<rjo> cheers.
<sb0> whitequark, can you go over the si5324 programming again and see if you can figure out why LOL_INT stays at 1?
<whitequark> sb0: ok. let me finish what I'm doing now and I'll go that
<sb0> ok
<sb0> why are you clearing a reserved bit (D2) in register 0?
<whitequark> typo, probably
<sb0> ah.
<bb-m-labs> build #381 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/381
<bb-m-labs> build #1289 of artiq is complete: Failure [failed python_unittest_2] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1289 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<GitHub> [artiq] whitequark pushed 2 new commits to master: https://github.com/m-labs/artiq/compare/f512ea42dc91...259319e456cf
<GitHub> artiq/master 259319e whitequark: firmware: use compiler_builtins crate instead of our compiler-rt....
<GitHub> artiq/master 44a9a79 whitequark: firmware: port allocator to Rust.
<whitequark> we're now down to only base-nofloat in runtime
<whitequark> so that's just the UART.
<whitequark> oh, and also strlen
<GitHub151> [rust] whitequark pushed 990 new commits to master: https://github.com/m-labs/rust/compare/2539c15b4953...6abe64871e6e
<GitHub151> rust/master 96c52d4 Esteban Küber: Add unittest
<GitHub151> rust/master 521b2ea Ariel Ben-Yehuda: clear discriminant drop flag at the bottom of a ladder...
<GitHub151> rust/master 40053a4 Yamakaky: Impl From<Ipv4Addr, Ipv6Addr> for IpAddr....
<bb-m-labs> build #382 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/382
<bb-m-labs> build #1290 of artiq is complete: Failure [failed python_unittest_2] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1290 blamelist: whitequark <whitequark@whitequark.org>
<whitequark> hmm, why did that start failing
<whitequark> ... hm
<whitequark> the cslice commit?!
<whitequark> this makes utterly zero sense
<sb0> cache problems?
<sb0> see also #485
<whitequark> sb0: hm
<whitequark> can I get performance counters?
<whitequark> instruction counter, l1/l2 miss counter
<whitequark> and I guess I can build a sampling profiler in other ways already
<sb0> I'm not familiar with the mor1kx code
<whitequark> sb0: oh I know what happens.
<whitequark> indeed cache effects.
<whitequark> specifically, the test relies on exceptions, and an exception will invoke the unwinder, which is... large
<whitequark> sb0: hm I can make rtio_output about 2x faster by fixing llvm
<whitequark> right now it generates shit like this
<whitequark> 40803be0: a9 88 00 4c l.ori r12,r8,0x4c
<whitequark> 40803be4: d4 0c 30 00 l.sw 0(r12),r6
<whitequark> should be just l.sw 0x4c(r12),r6 no?
<GitHub> [artiq] whitequark pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/9c93c760fc98aa67d3a9ab6ce61fdae0960830f3
<GitHub> artiq/master 9c93c76 whitequark: firmware: do not use PIC in Rust code....
<whitequark> sb0: actually... why did we switch to PIC in the first place?
<whitequark> there was some reason that I don't remember. but everything works without PIC.
<sb0> you wanted the kernels to be PIC
<rjo> whitequark: i had seen similar code and wondered. must have forgotten to point it out.
<whitequark> sb0: yeah but that seems unrelated to runtime/ksupport being PIC
<whitequark> every function loads the PIC base itself
<whitequark> oh, maybe I thought that isn't the case.
<whitequark> sb0: what's the problematic set of settings for si5324?
<sb0> why is the si5324 acting up...
<sb0> I can't even get it to lock to its crystal
<sb0> whitequark, absolutely nothing works. it never locks.
<whitequark> are you absolutely sure the conversions are correct?
<sb0> at least with the 62.5MHz settings
<whitequark> well, let me check them
<whitequark> what about the old settings we used?
<sb0> those should be the 62.5MHz settings
<whitequark> sb0: uhm
<bb-m-labs> build #383 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/383
<whitequark> sb0: ok, how do I build/run satman?
<whitequark> sb0: btw you haven't added loop bandwidth to parameters
<whitequark> if you're using the new parameters, can you set BWSEL_REG to what it says and try again?
<whitequark> sb0: wohoo, turning off PIC made pulse rate tests pass
<sb0> I've already tried BWSEL_REG=1, no effect
<whitequark> although for some utterly inexplicable reason they're still slower on CI than locally
<sb0> whitequark, good!
<sb0> whitequark, run kc705_drtio_satellite.py
<sb0> add --integrated-rom-size 32768 if that helps
<GitHub> [artiq] sbourdeauducq pushed 2 new commits to master: https://github.com/m-labs/artiq/compare/9c93c760fc98...90ac915dd163
<GitHub> artiq/master 90ac915 Sebastien Bourdeauducq: firmware: do not clear si5324 reserved bit
<GitHub> artiq/master 6bea253 Sebastien Bourdeauducq: firmware: fix misleading si5324 comment
<bb-m-labs> build #415 of artiq-win64-test is complete: Failure [failed python_unittest] Build details are at http://buildbot.m-labs.hk/builders/artiq-win64-test/builds/415 blamelist: whitequark <whitequark@whitequark.org>
<bb-m-labs> build #1291 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1291 blamelist: whitequark <whitequark@whitequark.org>
<whitequark> sb0: oh
<whitequark> did that bit just get inverted inadvertently?
<sb0> I have a bitstream in ~/artiq_drtio if you want
<sb0> that's the one I mentioned above. fixing it did not improve things.
<whitequark> ok
<whitequark> oh, hm
<whitequark> sb0: have you tried rebooting the board?
<whitequark> I vaguely recall some bug related to si5324 that went away after a reboot
<sb0> no, let me try on the other kc705
<sb0> same problem
<sb0> btw, the fpga may generate some garbage on clkin1, may be better to try locking the 5324 to its crystal first
<whitequark> sb0: your values for 62.5mhz are nuts
<whitequark> where did you get them from?
<sb0> from you
<whitequark> sb0: try these https://hastebin.com/fawugoboxa.php
<sb0> the fpga may not generate a valid clkin, and setting n32 to 0 isn't the right way to disable clkin2 afaict
<sb0> also, why are my value nuts? the large integers are expected
<bb-m-labs> build #384 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/384
<whitequark> ohhh
<whitequark> the divisor has large integers because of the crystal
<whitequark> hm
<sb0> yes
<bb-m-labs> build #416 of artiq-win64-test is complete: Failure [failed python_unittest] Build details are at http://buildbot.m-labs.hk/builders/artiq-win64-test/builds/416 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<bb-m-labs> build #1292 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1292 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<whitequark> maybe write(21, 0b11111100)?; ?
<whitequark> though I don't think that should affect anything
<whitequark> also can you read register 128 and see if it does in fact use the correct clock as PLL input?
<sb0> yes, it switches to the correct clock
<sb0> e.g. changing clock priorities affect register 128 normally
<sb0> write(21 didn't change anything
<whitequark> I don't see anything else that may be wrong
<whitequark> ask silabs?
<sb0> hm, the C code worked before
<sb0> right?
<whitequark> but this uses the old divider values.
<whitequark> the ones I gave you above in a hastebin link
<sb0> okay, those fail as well
<whitequark> hm, that seems troublesome
<sb0> yes. typical shit when dealing with any sort of proprietary chip
<sb0> maybe the mapping code does the wrong thing
<sb0> ffs I hate this
<sb0> I wonder how much of this garbage the hmc* will give us on sinara
<whitequark> sb0: afaict it doesn't do the wrong thing
<whitequark> I checked it twice
<larsc> the hmc wont give any garbage ;)
<sb0> okay, disabling "fast lock" makes it work
<sb0> now, is that awful drtio bug still there...
<sb0> panic at src/libcore/slice.rs:519: destination and source slices have different lengths
<sb0> ffs
<sb0> whitequark, that's when I try to artiq_run the kernel that produces the bug
<rjo> sb0, whitequark: we are at 5.6 GB of 3 GB on anaconda. lots of llvm, rustc, whitequark, could you clean up there?
<sb0> and the drtio bug is still there, of course.
<sb0> in a different form
<sb0> hm, so it did something
ohama has quit [Ping timeout: 258 seconds]
ohama has joined #m-labs
kuldeep has quit [Ping timeout: 252 seconds]
<sb0> seems this was due to the network packet buffer changes, but now network is completely broken after recompiling master (with the drtio design)
<sb0> also I cannot get this shitty si5324 to lock again
<GitHub> [artiq] sbourdeauducq pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/aafefee7f52d77a435b9c9b849aaea1c4ae89813
<GitHub> artiq/master aafefee Sebastien Bourdeauducq: targets: make number of ethmac slots consistent
bentley` has joined #m-labs
<bb-m-labs> build #385 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/385
<bb-m-labs> build #1293 of artiq is complete: Failure [failed python_unittest_2] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1293 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<whitequark> sb0: so... how do I repro?
<GitHub> artiq/master 5a30536 Robert Jordens: doc: write section on developing with a conda env
<GitHub> artiq/master edab068 Robert Jordens: conda: pull in openocd in artiq and artiq-dev
<GitHub> [artiq] jordens pushed 2 new commits to master: https://github.com/m-labs/artiq/compare/aafefee7f52d...5a30536d59a8
<bb-m-labs> build #1294 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1294 blamelist: Robert Jordens <rj@m-labs.hk>
<GitHub> [artiq] jordens pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/02917aae6282a7af2387326d66c656fa8a084ed9
<GitHub> artiq/master 02917aa Robert Jordens: doc: newline after label
<bb-m-labs> build #1295 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1295 blamelist: Robert Jordens <rj@m-labs.hk>
<GitHub> [artiq] jordens pushed 1 new commit to master: https://github.com/m-labs/artiq/commit/9c646801e3cf8d2c222f999ffe53e5cea5884131
<GitHub> artiq/master 9c64680 Robert Jordens: doc: fix verbatim sections
<bb-m-labs> build #1296 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1296 blamelist: Robert Jordens <rj@m-labs.hk>
<bb-m-labs> build #1297 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1297 blamelist: Robert Jordens <rj@m-labs.hk>
<rjo> bb-m-labs: build conda-all --package=openocd
<bb-m-labs> build #167 of conda-win32 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/167
<bb-m-labs> build #269 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/269
<bb-m-labs> build #168 of conda-win32 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/168
<GitHub85> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/f975744e26542207176e95780f200fa75396339c
<GitHub85> conda-recipes/master f975744 Robert Jordens: openocd: build 0.10.0 release
<bb-m-labs> build #1298 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1298 blamelist: Robert Jordens <rj@m-labs.hk>
<bb-m-labs> build #170 of conda-win64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/170
<bb-m-labs> build #84 of conda-all is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/conda-all/builds/84
<bb-m-labs> build #169 of conda-win32 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/169
<bb-m-labs> build #270 of conda-lin64 is complete: Failure [failed anaconda_upload] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/270
<bb-m-labs> build #171 of conda-win64 is complete: Exception [exception conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/171
<bb-m-labs> build #85 of conda-all is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/conda-all/builds/85
<bb-m-labs> build #172 of conda-win64 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/172
<GitHub161> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/4e105b22b3e1cb603a5bcb7a241a3fbb35349afe
<GitHub161> conda-recipes/master 4e105b2 Robert Jordens: openocd: rm non-ascii files
<whitequark> rjo: uhm, how do you expect that to work?
<whitequark> в is two utf-8 bytes
<whitequark> it won't be matched by a single shell ?, I don't think
<bb-m-labs> build #271 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/271
cr1901_modern has quit [Ping timeout: 255 seconds]
cr1901_modern has joined #m-labs
<cr1901_modern> Huh, FTDI serial driver caused a BSOD... that's a first for me.
<bb-m-labs> build #1299 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1299 blamelist: Robert Jordens <rj@m-labs.hk>
<rjo> whitequark: whoever wrote the windows build scripts did that.
<rjo> whitequark: and at least in my shell it works fine.
<rjo> whitequark: i.e. ? is in codepoints.
<bb-m-labs> build #170 of conda-win32 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/170
MiW has quit [Ping timeout: 260 seconds]
kilae has quit [Quit: ChatZilla 0.9.93 [Firefox 51.0.1/20170125094131]]
MiW has joined #m-labs
mumptai has joined #m-labs
<GitHub> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/95332e5ac1c93e459bb412bc6353719b9e0f9dd0
<GitHub> conda-recipes/master 95332e5 Robert Jordens: openocd: rm more non-ascii files
<bb-m-labs> build #272 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/272
<bb-m-labs> build #173 of conda-win64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/173
<bb-m-labs> build #171 of conda-win32 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/171
<bb-m-labs> build #86 of conda-all is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-all/builds/86
<bb-m-labs> build #1300 of artiq is complete: Failure [failed conda_build_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1300 blamelist: Robert Jordens <rj@m-labs.hk>
<GitHub> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/7a600c4ca8687ae256e981d94b3192e90edd9c2f
<GitHub> conda-recipes/master 7a600c4 Robert Jordens: openocd: bump
<bb-m-labs> build #273 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/273
<bb-m-labs> build #172 of conda-win32 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win32/builds/172
<bb-m-labs> build #174 of conda-win64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-win64/builds/174
<bb-m-labs> build #87 of conda-all is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-all/builds/87
mumptai has quit [Quit: Verlassend]
<sb0> whitequark, compile kc705_drtio_satellite+satman, load into a kc705
<sb0> I recommend to try locking the 5324 to its crystal first, clkin1 generated by the fpga has a lot more moving parts
<bb-m-labs> build #386 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/386
<bb-m-labs> build #1301 of artiq is complete: Failure [failed artiq_flash] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1301 blamelist: Robert Jordens <rj@m-labs.hk>
kuldeep has joined #m-labs