00:00
<
wolfspraul >
let us know when it's committed somewhere then Xiangfu can try to build images for people to test.
00:00
<
wolfspraul >
or anybody can try, in fact
00:28
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * ra2602e3 / (3 files in 2 dirs): minimac2: RX working in simulation -
http://bit.ly/etJKxT
09:07
<
mwalle >
lekernel: milkymist2 ?
09:07
<
mwalle >
larsc: nice :)
09:08
<
mwalle >
larsc: what happens if you do a 'modprobe <some non-existing module name>'?
09:15
<
mwalle >
lekernel: psync ^= pulse sync?
09:21
<
xiangfu_ >
the 'Keyboard layout' not working. which source file I should look into?
09:30
<
mwalle >
xiangfu_: maybe keymap.c in mtk? dunno
09:30
<
mwalle >
i only see one keymap there
09:30
<
mwalle >
looks like the german one
09:33
<
xiangfu_ >
mwalle: thanks
11:05
<
lekernel >
mwalle: yeah... can't seem to make the rxfifo overflows disappear no matter what
11:06
<
lekernel >
so i'll just store the packets on-chip and do away with that 'streaming dma' idea
11:08
<
CIA-8 >
mtk: Xiangfu Liu master * rb5bd14f / (lib/edit.c lib/entry.c):
11:08
<
CIA-8 >
mtk: fix CtrlA forget one character
11:09
<
lekernel >
(it could probably work with FML, but use more resources, be complex and require L2 flushes...)
12:33
<
xiangfu_ >
hmm... after update all git repo. the flickernoise have no output in VGA. serial console works fine. telnet, ftp works fine.
12:34
<
xiangfu_ >
updated: rtems-milkymist.git, mtk.git liboscparse.git rtems-yaffs2.git flickernoise.git
12:36
<
xiangfu_ >
splash screen, BIOS works fine
13:20
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * r6ba1c0e / (22 files in 9 dirs): minimac2: SoC integration -
http://bit.ly/g6pPX2
13:20
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * rfc972c5 / (4 files in 4 dirs): libnet: support minimac2 -
http://bit.ly/e5p40s
13:28
<
lekernel >
bitgen after 20min synthesis:
13:28
<
lekernel >
ERROR:PhysDesignRules:1151 - Issue with pin connections and/or configuration on
13:28
<
lekernel >
  block:<ethernet/memory/txb>:<RAMB16BWER_RAMB16BWER>.  The block is configured
13:28
<
lekernel >
  to use DIPA0 and it is not connected.
13:38
<
xiangfu_ >
building the branch 'newmac' :)
13:43
<
lekernel >
it doesn' build... moment
13:47
<
xiangfu_ >
just want give you some feedback, see if I can get the same result in my PC :) (it's should be. but since all of you say the 'xst' is very strange, maybe the result is different :)
13:49
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * r553ab22 / coredoc.inc : minimac2: disable doc build -
http://bit.ly/eF6BsA
13:49
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * r1df5140 / cores/minimac2/rtl/minimac2_memory.v : minimac2: ground unused BRAM port to make bitgen happy -
http://bit.ly/dFb5ng
13:50
<
lekernel_ >
no, it won't be different
13:50
<
lekernel_ >
I just fixed it btw
13:51
<
lekernel_ >
but it seems I also fucked up the wb bus interconnect
14:25
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * r5af65df / cores/conbus/rtl/conbus5x6.v : Fix slave address decoding -
http://bit.ly/ghxOwM
14:34
<
mwalle >
lekernel: streaming dma?
14:35
<
lekernel >
yeah, the minimac1 core doesn't store a complete packet on-chip but DMA's it at the same time as it is received
15:00
<
mwalle >
lekernel: are the minimac2 control registers nearly final?
15:00
<
mwalle >
s/final/stable/
15:00
<
lekernel >
should be...
15:01
<
lekernel >
atm tx works on board, rx has only one byte out of two written to RAM for some reason (which doesn't happen in simulation)
15:18
<
mwalle >
lekernel: how big is a rx/tx buffer supposed to be?
15:18
<
lekernel >
2 kilobytes
15:23
<
mwalle >
max mtu will be 2KiB - preamble - fcs ?
15:49
<
lekernel >
nice ISE bugs that completely break the RX FSM and/or the block RAM...
16:08
<
lekernel >
new ethernet works :)
16:09
<
CIA-8 >
milkymist: Sebastien Bourdeauducq newmac * r624a332 / cores/minimac2/rtl/minimac2_memory.v : minimac2: fix block RAM WE -
http://bit.ly/fhMx34
17:00
<
CIA-8 >
rtems-milkymist: Sebastien Bourdeauducq master * r853139d / (2 files in 2 dirs): Ethernet: minimac2 driver -
http://bit.ly/f5Q0Hd
18:38
<
CIA-48 >
milkymist: Sebastien Bourdeauducq newmac * r1c6e205 / boards/milkymist-one/synthesis/common.mak : Workaround for weird Xilinx bug -
http://bit.ly/i3D7C7
18:46
<
lekernel >
i'm putting everything to an organization github... create github accounts if you want to be on it
18:55
<
CIA-48 >
milkymist: Sebastien Bourdeauducq master * r1c6e205 / boards/milkymist-one/synthesis/common.mak : Workaround for weird Xilinx bug (+8 more commits...) -
http://bit.ly/i3D7C7
19:00
<
larsc >
i always wonder how they manage to actually sell stuff with their load of weired bugs
19:01
<
larsc >
the one time i was using altera products i would get different results depending on whether i was using tabs or spaces for indention
19:21
<
lekernel >
now that one corrupts some pixel data... but works for the rescue bitstream and now with a different placer PRNG seed
19:22
<
lekernel >
maybe this has to do with internal FPGA crosstalk, according to several research papers I read it exists and the tools do not deal with it
21:00
<
lekernel >
rejon: oscsend 192.168.0.42 7777 /osd s Hello\ World
21:01
<
lekernel >
with a patch running
21:02
<
rejon >
lekernel cool
22:08
<
CIA-48 >
flickernoise: Sebastien Bourdeauducq master * r07aa3e8 / (7 files): Rename DMX table to DMX desk -
http://bit.ly/gsBKg4
22:29
<
mwalle >
lekernel: pushed minimac2 model, dunno if it works correct the bios fails loading a file via tftp, seems that there is a timeout (the second loop in tftp.c)
22:48
<
lekernel >
tftp works on the board...
22:52
<
lekernel >
rtems too (I just tested ftp with gftp, cyberduck and firefox)