05:53 <GitHub177> [misoc] enjoy-digital pushed 1 new commit to master: https://git.io/vwM49

05:53 <GitHub177> misoc/master e136030 Florent Kermarrec: cores/liteeth_mini/mac/sram: fix reception of frames larger than MTU...

05:54 <bb-m-labs> build #101 of misoc is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/misoc/builds/101

05:55 <_florent_> sb0: ^ I think you can retry receiving jumbo frame with that. It should return the correct length but drop the remaining bytes larger than MTU

05:56 <_florent_> sb0: for sram depth, design use slots with depth that are power of of two (to decode slots). If we want to support non power of two depth we have to some change on the design

06:07 <bb-m-labs> build #393 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/393

06:08 <bb-m-labs> build #656 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/656

14:43 <sb0> _florent_, non power of two BRAM != power of two address alignment

14:44 <sb0> you can have non-power-of-two BRAMs are aligned to power-of-two addresses without using any extra FPGA resources

14:46 <sb0> whitequark, can you fix the buildbot?

15:37 <whitequark> sb0: no

15:37 <whitequark> I've no idea what's broken with it

15:37 <whitequark> the correct gateware is flashed

15:42 <GitHub109> [misoc] sbourdeauducq pushed 1 new commit to master: https://git.io/vwMM0

15:42 <GitHub109> misoc/master 98ed0ce Sebastien Bourdeauducq: bios: fix help message

15:43 <bb-m-labs> build #102 of misoc is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/misoc/builds/102

15:55 <bb-m-labs> build #394 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/394

15:57 <bb-m-labs> build #657 of artiq is complete: Failure [failed] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/657

15:59 <GitHub55> [misoc] sbourdeauducq pushed 1 new commit to master: https://git.io/vwMDu

15:59 <GitHub55> misoc/master 0d08038 Sebastien Bourdeauducq: liteeth: use non power-of-two RAM depth

16:00 <bb-m-labs> build #103 of misoc is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/misoc/builds/103

16:12 <bb-m-labs> build #395 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/395

16:18 <sb0> of course, lwip provides exactly zero support for rejecting jumbo frames, and you have to decode frames yourself

16:18 <sb0> sigh

16:18 <bb-m-labs> build #153 of artiq-win64-test is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-win64-test/builds/153

16:19 <bb-m-labs> build #658 of artiq is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/658

16:23 <sb0> setting MTU to 9k in lwip will also cause problems in the other direction, because lwip doesn't support path MTU discovery either

16:23 <sb0> do people actually use lwip?

16:25 <sb0> no jumbo frames = you can't reject them, jumbo frames = you cannot talk to devices that don't support them

16:39 <whitequark> sb0: was that the liteeth power of two issue?

16:42 <sb0> I have no idea why it works again

16:44 <sb0> and last time it definitely flashed an older package. maybe there is some conda non-determinism mixed with vivado non-determinism ...