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
<bb-m-labs_> build #310 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/310
mumptai has quit [Quit: Verlassend]
<bb-m-labs_> build #1214 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1214 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<_florent_> jesd
<_florent_> sorry...
kuldeep_ has joined #m-labs
kuldeep has quit [Remote host closed the connection]
cr1901_modern has quit [Ping timeout: 245 seconds]
cr1901_modern has joined #m-labs
cr1901_modern has quit [Read error: Connection reset by peer]
cr1901_modern has joined #m-labs
sandeepkr has quit [Ping timeout: 240 seconds]
kuldeep_ has quit [Ping timeout: 258 seconds]
sandeepkr has joined #m-labs
kuldeep has joined #m-labs
mumptai has joined #m-labs
mumptai has quit [Remote host closed the connection]
ohama has quit [Ping timeout: 240 seconds]
ohama has joined #m-labs
<whitequark> sb0: session seems to work, running last tests now
<whitequark> still need to fix a memory leak
<GitHub120> [smoltcp] whitequark closed pull request #1: Fix mislabeled features in README.md (master...patch-1) https://git.io/vMrq6
<GitHub93> smoltcp/master c4c8a44 Taylor Cramer: Fix mislabeled features in README.md
<GitHub93> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vMrpN
<travis-ci> m-labs/smoltcp#32 (master - c4c8a44 : Taylor Cramer): The build passed.
sandeepkr has quit [Read error: Connection reset by peer]
sandeepkr has joined #m-labs
<cr1901_modern> sb0: Who is the person you know doing the RISC-V to lm32 translator layer?
<GitHub130> [artiq] whitequark opened issue #655: Unbreak the !unconditionally_dereferenceable LLVM patch https://git.io/vMoqY
<whitequark> sb0: I would *really* like something like a faster UART, somehow, without constant rebuilding of bitstreams
<whitequark> every time I start tracing something to UART, the entire core device becomes UART-bound
ysionnea1 is now known as ysionneau
<GitHub49> [artiq] whitequark created smoltcp (+2 new commits): https://git.io/vMoZC
<GitHub49> artiq/smoltcp b9bd03d whitequark: firmware: integrate smoltcp instead of lwip.
<GitHub49> artiq/smoltcp d03c546 whitequark: artiq_devtool: add boot+log action; avoid forwarder crashes.
<rjo> whitequark: you can crank up the speed. iirc it worked fine up to a MBit or thereabouts.
<rjo> whitequark: but it seems that you will always be uart bound. why don't you trace into a buffer and read out later?
<whitequark> rjo: not if I am debugging a crash
<whitequark> you could make an argument for never panicking, but it's not always realistic. e.g. out of bounds accesses will *always* crash the entire runtime
<rjo> whitequark: can't you read it through the debugger?
<whitequark> rjo: I can, but the ergonomics of this is atrocious
<whitequark> it'd be a method of last resort
<whitequark> as you may remember we have no JTAG on OR1K so I debug through a gdb stub over network
<whitequark> this works more reliably than what you would expect, but the latency is high, throughput is low, and it sometimes fails for no apparent reason
<whitequark> this also requires a really large buffer...
<whitequark> I actually suspect a better method would be streaming all log messages over UDP, perhaps (but not necessarily) even bypassing the rest of the stack
<whitequark> this would be fairly easy as smoltcp exposes all of its guts, including the packet construction machinery.
<whitequark> rjo: what's the proper baudrate to set?
<whitequark> from the datasheet it looks like ft2232h doesn't care at all as long as you can set the divisor
<whitequark> 923076?
<whitequark> ah, 921600. but let's try 3M first.
<GitHub169> [artiq] whitequark pushed 1 new commit to smoltcp: https://git.io/vMoW6
<GitHub169> artiq/smoltcp e3a4812 whitequark: gateware.targets.kc705_dds: increase UART baudrate to 3Mbps.
<whitequark> bb-m-labs_: force build --help
<bb-m-labs_> Something bad happened (see logs)
<whitequark> bb-m-labs_: force --help
<bb-m-labs_> try 'force build [--branch=BRANCH] [--revision=REVISION] [--props=PROP1=VAL1,PROP2=VAL2...] <WHICH> <REASON>'
<whitequark> bb-m-labs_: force build --branch=smoltcp artiq
<bb-m-labs_> build forced [ETA 46m31s]
<bb-m-labs_> I'll give a shout when the build finishes
<bb-m-labs_> build #1215 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1215
<whitequark> what
<whitequark> bb-m-labs_: force build --branch=smoltcp artiq
<bb-m-labs_> build forced [ETA 46m31s]
<bb-m-labs_> I'll give a shout when the build finishes
<GitHub27> [artiq] whitequark force-pushed smoltcp from e3a4812 to 0578dc7: https://git.io/vMoWb
<GitHub27> artiq/smoltcp 0578dc7 whitequark: gateware.targets.kc705_dds: increase UART baudrate to 3Mbps.
<GitHub27> artiq/smoltcp 61f10bd whitequark: firmware: integrate smoltcp instead of lwip.
sb0 has joined #m-labs
<sb0> whitequark, you don't have to rebuild bitstreams to change the baudrate, it can be modified by CSR
kuldeep has quit [Remote host closed the connection]
kuldeep has joined #m-labs
<whitequark> oh, hm
<whitequark> sb0: uart_phy::tuning_word_write?
<whitequark> which equals to (baudrate/clk_freq)*2**32? ok...
<whitequark> bb-m-labs_: stop build artiq no need
<bb-m-labs_> build 1216 interrupted
<sb0> yes
<whitequark> sb0: that doesn't solve the extremely slow serial upload...
<sb0> upload for what? the bios also supports tftp netboot
<whitequark> so now I will have to spend time scripting that, too
<whitequark> since flterm doesn't work at all with non-TTYs for some stupid reason
<sb0> maybe the board can attempt tftp first? that would also skip some of the flashing on the buildbot
<whitequark> that would work
<sb0> or all, since the BIOS usually keeps working after bitstream changes
<sb0> just load the bitstream in SRAM
<sb0> the downside is, this stops testing artiq_flash
<sb0> also there is no good way to store BIOS config options right now, it's all hardcoded
<bb-m-labs_> build #311 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/311
<bb-m-labs_> build #1216 of artiq is complete: Exception [exception] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1216
<whitequark> sb0: so why can't we just bump the baudrate to 3M, assuming it works reliably?
<whitequark> (or I guess 1M also works, since 3M is not quite standard)
<whitequark> this would fix the issue where the host and the device have to coordinate baudrate switch or lose information
<sb0> I'm fine with that as long as it works ok
<sb0> does windows support it?
<whitequark> and I think I've used it myself before, but that was very long ago
<whitequark> oh, looks like cp2103 only supports up to 2M
sb0 has quit [Quit: Leaving]
<whitequark> sb0: looks like 921600 works OK and the rest has sketchy support in drivers and/or software
<whitequark> that's on linux that is. I guess 921600 is OK...
sandeepkr has quit [Remote host closed the connection]
<GitHub184> [artiq] whitequark force-pushed smoltcp from 0578dc7 to bb0b8c8: https://git.io/vMoWb
<GitHub184> artiq/smoltcp bb0b8c8 whitequark: firmware: increase post-boot UART speed to 921600.
sandeepkr has joined #m-labs
<whitequark> sb0: incidentally, what's the point of UART buffering in libbase?
<whitequark> nevermind, dumb question
<GitHub45> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vMooB
<GitHub45> smoltcp/master 93c7fba whitequark: Always display meaningful endpoint in trace messages.
<travis-ci> m-labs/smoltcp#33 (master - 93c7fba : whitequark): The build passed.
<GitHub170> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vMoXh
<GitHub170> smoltcp/master 778753d whitequark: Do not send RST in response to invalid SEQ or ACK.
mumptai has joined #m-labs
<GitHub117> [artiq] vontell opened issue #656: Default types overridden https://git.io/vMojy
<GitHub129> [artiq] whitequark closed issue #656: Default types overridden https://git.io/vMojy
<GitHub41> [artiq] whitequark commented on issue #656: You can think of the ARTIQ Python compiler to override all built-in types.... https://git.io/vMKJ8
<GitHub12> [artiq] whitequark commented on issue #656: You can think of the ARTIQ Python compiler to override all built-in types.... https://git.io/vMKJ8
<GitHub99> [artiq] sbourdeauducq pushed 1 new commit to master: https://git.io/vMKUR
<GitHub99> artiq/master b4f3be7 Sebastien Bourdeauducq: doc: fix table formatting problem, add 'string constant' as supported compiler type
<GitHub197> [artiq] sbourdeauducq pushed 1 new commit to release-2: https://git.io/vMKUa
<GitHub197> artiq/release-2 4d5ba3f Sebastien Bourdeauducq: doc: fix table formatting problem, add 'string constant' as supported compiler type
sb0 has joined #m-labs
sb0 has quit [Client Quit]
<bb-m-labs_> build #312 of artiq-board is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-board/builds/312
<bb-m-labs_> build #1217 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/1217 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
mumptai has quit [Quit: Verlassend]
<GitHub12> [smoltcp] whitequark pushed 1 new commit to master: https://git.io/vMKg3
<GitHub12> smoltcp/master 1682662 whitequark: Distinguish sockets by debug identifiers (socket set indexes).
<travis-ci> m-labs/smoltcp#35 (master - 1682662 : whitequark): The build passed.
sb0 has joined #m-labs