ali-as has quit [Remote host closed the connection]
ali-as has joined #glasgow
egg|anbo|egg_ has joined #glasgow
etrig has joined #glasgow
modwizcode has quit [Quit: Later]
m42uko has quit [Ping timeout: 264 seconds]
egg|anbo|egg_ has quit [Remote host closed the connection]
egg|anbo|egg_ has joined #glasgow
Griwes has joined #glasgow
m42uko has joined #glasgow
jstein has quit [Quit: quit]
egg|anbo|egg_ has quit [Ping timeout: 246 seconds]
electronic_eel has quit [Ping timeout: 246 seconds]
electronic_eel has joined #glasgow
PyroPeter_ has joined #glasgow
PyroPeter has quit [Ping timeout: 240 seconds]
PyroPeter_ is now known as PyroPeter
meklort has joined #glasgow
lxdr has quit [Ping timeout: 256 seconds]
lxdr has joined #glasgow
<_whitenotifier>
[glasgow] hansfbaier opened issue #264: VGM protocol: - https://git.io/JtYZK
<d1b2>
<david.lenfesty> yeah you can completely shut off GoodCRC with the FUSB302
<d1b2>
<david.lenfesty> just clear AUTO_CRC. It would probably make a pretty good sniffer I think you can just shut off TX and ignore that side of things
<d1b2>
<david.lenfesty> context: I've been writing a C++ driver for it for STARSHIPRAIDER because we're abusing USB-C for some power delivery
<d1b2>
<david.lenfesty> also just bit the bullet and got a glasgow so I'm excited for it to get here 🙂
<d1b2>
<rwhitby> @david.lenfesty is that driver available anywhere to look at?
<d1b2>
<david.lenfesty> I plan on lifting it out into a seperate repo as a standalone library once it's more complete and I've actually tested it (don't even have the hw yet)
<d1b2>
<rwhitby> I like how you send the power over the high speed lanes 🙂
<rwhitby>
except it doesn't break out SBU1/SBU2 which are the key signals that marcan needs for accessing debug interfaces on the M1
<rwhitby>
but yeah, just adding some more probe points to an existing design would be a great way to go
<kc8apf>
Jason also spent time breaking out the Chromium EC driver into something that could be used elsewhere.
<kc8apf>
I have one in a drawer somewhere from when I was interested in PD and before I realized how much of a pain it is
<d1b2>
<david.lenfesty> That's why I said we were abusing USB-C 😉
<kc8apf>
You say abuse. I say alternate mode.
<d1b2>
<david.lenfesty> +-7VDC on the superspeed lines is in the spec right?
GNUmoon2 has joined #glasgow
<kc8apf>
USB C spec alternate mode requirements only impose a voltage limit to reduce risk of damage to superspeed receivers. I guess the risk is if you negotiated the alternate mode and a cable was rapidly moved to a regular port, you could damage the receiver before you noticed the unplug/replug events.
Stephie has quit [Quit: Fuck this shit, I'm out!]
Stephie has joined #glasgow
<d1b2>
<Hardkrash> For what it's worth, all interfaces in a type-c port should be protected from shorts the a 20V VBus, this was not considered in earlier hardware and now products exist for this.
<d1b2>
<Hardkrash> As for a PD-Phy I was considering using the ST USB1602 for its BMC Phy over SPI; but there is a bit of unwanted logic built in. It would take an investigation to find out if the USB1602 could be put into a passive mode where only the BMC was active, but I suspect that that is not likely. So for Glasgow as a sniffer it is likely preferable to use some protection from VBus shorts, a comparator for the digital PD levels, and an ADC for the analog
<d1b2>
levels. For completeness adding resistors or a current output DAC to emulate the various resistors would allow for additional testing with out of spec CC pulls.
<agg>
probably costs them more in engineer-minutes to write down a four layer stackup than to just do a prototype run on whatever mad HDI process they were already using
<d1b2>
<Attie> oh nice, thanks for the link... been keeping my eyes open for this
<d1b2>
<Attie> it might have been nice to use a button alongside the illegal cable for access to the bootloader? (or am I mis-reading this)
<modwizcode>
I do not think this means what you think it does
<whitequark>
ha
<modwizcode>
Anyway I'm pretty sure that was the actually interesting chip you told me about. I think the one I found that only did storage might have been via oh wait I think it was realtek or something
<modwizcode>
realtek is another one of those blackhold of products/information companies
<modwizcode>
Maybe it's just these phy/communication chip companies in particular?
<whitequark>
they're only interested in b2b
<agg>
are any chip companies interested in b2c?
<whitequark>
atmel?
<whitequark>
well, it depends on your definition of "c"
<agg>
I guess you could call rpi a chip company now
<whitequark>
i guess it would be more correct to say "they're only interested in very large orders"
<modwizcode>
I mean yeah but like how is your business supposed to do anything unless you can get a sample easily :p
<modwizcode>
Also the definition of large order is many orders of magnitude over what I even thought
<modwizcode>
rpi seems fairly successful and they're a tiny customer apparently?
<agg>
right, there's b2b and b2B, where the B is Big Business :p
<whitequark>
a few companies are not interested in your business period. if they want you to use their products, they'll call you
<whitequark>
broadcom is the most notorious example
<modwizcode>
that's such a weird model for a business lol
<modwizcode>
imo dumb
<whitequark>
from what i've heard, you don't necessarily get treated well by broadcom *even if you are a large corp*
<modwizcode>
It'd be nice if chip companies weren't kinda hard to actually compete with given the fact that spinning up a chip is like... next to impossible to do to enter the market
<agg>
the funny thing is that since broadcom got bought by avago you can actually buy broadcom-brand stuff at your local distributor
<modwizcode>
SiFive is the only example I know of that has done that
<whitequark>
it's not that hard to make *a* chip
<agg>
like plastic optical fibre and LEDs
<modwizcode>
Wait broadcom got bought by a company I never heard of
<whitequark>
broadcom bought avago, not the other way around, no?
<d1b2>
<daveshah> Ah the company with a brand called "OpenFive" exclusively selling closed source IP
<gruetzkopf>
sigh..
<hell__>
so much for the freedom
rcombs has quit [*.net *.split]
Twix has quit [*.net *.split]
Twix has joined #glasgow
<gruetzkopf>
also, glasgow jtag-svf does not like my fresh-out-of diamond-deployment-tool svf (stumbles over LOOP ) ( relevant part of file, output): https://paste.debian.net/plain/1182125
<whitequark>
yes, ddt creates extended svf
<whitequark>
LOOP would be a reasonable extension to implement
<gruetzkopf>
okay, off to jtag-openocd and openocds svf player
<whitequark>
yep that should work
<modwizcode>
damn...
rcombs has joined #glasgow
GNUmoon has quit [Quit: Leaving]
GNUmoon has joined #glasgow
<Ultrasauce>
ah i see the datasheet download links on their site are contact form gated
<gruetzkopf>
there's also a revB _somewhere_ in my room
<whitequark>
revB's usefulness is so limited
<gruetzkopf>
worked well enough for what i used it for
<whitequark>
i suppose. there is a reason i still support it
<gruetzkopf>
also, since i got the revC, i haven't once gotten a jtag pin order other than glasgows default out of jtag-pinout :D
<whitequark>
ha!
<tnt>
revb with the voltage shifter removed/bypassed is plenty useful.
<tnt>
95% of stuff I do is 3v3 ...
<whitequark>
yes, true
<whitequark>
i do a lot of both 5V and 1V8, so i'm biased
<gruetzkopf>
quite a lot of stuff i do is high-fanout LS-TTL
GNUmoon has quit [Ping timeout: 268 seconds]
<gruetzkopf>
only thing i need to do is shout at their dock not working with any FX2 device i've tried
<gruetzkopf>
*HP
<whitequark>
what happens?
<gruetzkopf>
[203832.933395] usb 3-1.3.2: Device not responding to setup address.
<whitequark>
does it try to drop to FS and that breaks? i'm not sure if glasgow can do FS (quite possibly not)
<gruetzkopf>
(pure xhci, external ports are 2 hubs down)
<whitequark>
hm
<whitequark>
if it's FS related i can fix that
<gruetzkopf>
i don't believe it is, works every few 10s of tries
<gruetzkopf>
well, enumeration does
<gruetzkopf>
glasgow run selftest
<gruetzkopf>
I: g.device.hardware: building bitstream ID 408ddb4725ad0e22592b85560d402a6c
<gruetzkopf>
E: g.cli: transfer error: LIBUSB_TRANSFER_ERROR
<whitequark>
weird
<whitequark>
needs a vizsla
<gruetzkopf>
all native ports work, so i'm not too worried
<gruetzkopf>
much easier to blame the microchip U
<gruetzkopf>
*USB5734 hub
<gruetzkopf>
"icrochip’s Smart hubs also enable a downstream device to take control of the host system by swapping roles and becoming the host port. The Smart hub can also switch between two different hosts if required. This role changing technology is called FlexConnect and can be initiated through hardware or software commands. " AAAAAAH
<whitequark>
wh... what
<whitequark>
is this like... PCIe NTB failover setups
<gruetzkopf>
most likely for superspeed capable KVMs
<whitequark>
oh!
<gruetzkopf>
okay, a *blank* fx2 (on one of those blue test boards) enumerates repeatably
<gruetzkopf>
*someone* in my social circle should have a usb analyser
<whitequark>
yep
<whitequark>
if it fails before setup address... you can just use a rigol
<gruetzkopf>
my patched-to-1104Z worked well on my last "fun" usb bug (*old* cypress usb FS/LS host controller with bad PLL which essentially overclocked the usb bus (i've seen SOF intervals of 720µsec - and some devices even worked with that (mostly clockless usb sticks))
<whitequark>
oh i recall that
egg|anbo|egg has joined #glasgow
egg|anbo|egg_ has quit [Ping timeout: 260 seconds]
<d1b2>
<fishmonger> When I get the Glasgow from Crowd Supply, one of the first things to do will be replacing Arduino-hacks for dumping EEPROMs.
<d1b2>
<fishmonger> Do I remember correctly that you built something to automatically vary the voltage when dumping whitequark? My solution for many years has been to do it manually, so writeups on better methods I can steal sure would be neat.
<whitequark>
fishmonger: yes, I repeatedly read and check whether the images match
<whitequark>
there is also some code to generate 0/1 bit content histograms
<whitequark>
this can be developed further but the existing code works quite well
<d1b2>
<fishmonger> How do you change the voltage? Programmable supply on the side?
<gruetzkopf>
integrated!
<whitequark>
^ yep. not very high current but it can source a hundred milliamps without issues
<d1b2>
<fishmonger> In Glasgow. Oh ... that's cheating! 🙂
<whitequark>
cheating is how we get stuff done
<d1b2>
<fishmonger> I like it! My knowledge of exactly what Glasgow can do is very limited. Should read some docs, but that will make me frustrated when the device isn't here to play with.
<whitequark>
hehe
<gruetzkopf>
(i remember why I usually avoid openocd at all cost)
<modwizcode>
I was having fun looking at the software pll to recover floppy data, that lead to a fun bit of time trying to write a software PLL of my own to lock on a simple square wave software generated signal. For some reason mine never worked though it couldn't be dialed in and it would always lock at at a slightly wrong frequency. I should figure out why that didn't work.
<gruetzkopf>
next step: horribly hack loop support into glasgows svf interpreter
<gruetzkopf>
shoudn't be too bad (no nesting)
<gruetzkopf>
oh, made it spit out a non-LOOPy svf
<whitequark>
what's the terminating condition on the loop?
<gruetzkopf>
loop entry is "LOOP count ;" exit is "ENDLOOP ;"
<d1b2>
<fishmonger> @whitequark Is you dump code available? Might as well start learning nMigen by looking at interesting code.
<tnt>
whitequark: btw, with ROM errors reduced with lower voltage, did you get an explanation ? I was wondering if maybe those were using high Vt implant to turn off transistors and those degrated and the Vt was no longer high enough to be "off" at normal Vcc.
<whitequark>
tnt: possibly
<whitequark>
I don't know a reason for certain
<tnt>
were the errors all in the same "direction" ?
<tnt>
(i.e. all 1->0 or all 1->0)
<gruetzkopf>
it's old wisdom to read degraded memory at low voltage
<whitequark>
tnt: yes, but the direction differed for different vendors
<whitequark>
iirc
<tnt>
yeah, that I would imagine depends if the read circuit has an odd or even number of inverters in the path.
<modwizcode>
I think whitequark said it best "do not think about google thing. that only leads to suffering"
<gruetzkopf>
" I knew there was a reason why I prefer Altera. " best comment
<whitequark>
someone should save these videos
<modwizcode>
for sure
<d1b2>
<david.lenfesty> I bookmarked them
<gruetzkopf>
currently dl'ing
<d1b2>
<fishmonger> Oh, the community Element14 tried to make a thing by buying Ben Hecks channel.
<d1b2>
<david.lenfesty> Yeah this is going in my "Important Videos" folder
<modwizcode>
Waittttt
<modwizcode>
they did?
<modwizcode>
that's pretty uh... yikes
<d1b2>
<fishmonger> Well, it provided employment for Mr Heck for years, so I'm sure he's ok with it. But when he quit he no longer had a channel and created a new one. The original is pretty dead now I think.
<modwizcode>
Is there a picture of the glasgow board inside the case? I feel like I haven't seen that.
<electronic_eel>
modwizcode: there is the rendering on the crowdsupply page
<electronic_eel>
also a photo of the case prototype
<electronic_eel>
i think esden has just ordered a more recent version of a case prototype
<_whitenotifier>
[archive] electroniceel closed pull request #6: added few technical notes and reference manual - https://git.io/Jtvxh
<_whitenotifier>
[GlasgowEmbedded/archive] electroniceel pushed 1 commit to master [+8/-0/±0] https://git.io/JtO8U
<_whitenotifier>
[GlasgowEmbedded/archive] Alenlu 278a67f - added few technical notes and reference manual (#6)
<_whitenotifier>
[archive] electroniceel commented on pull request #6: added few technical notes and reference manual - https://git.io/JtO8T
<whitequark>
tnt: "What's the background music" good question, i wonder if you can like... search by riff
<whitequark>
shazam definitely doesn't know what it is
<modwizcode>
google's music identification doesn't seem to work either
<whitequark>
it's very catchy.
<modwizcode>
that's one way to put it
<electronic_eel>
hmm, maybe the people making the video also made it themselves? otherwise they'd have to license it from somewhere
<modwizcode>
It's gotta be based on something
<whitequark>
i think it would qualify as "parody"
<modwizcode>
yeah
<electronic_eel>
not sure you need to base it on something. if you play in a band as hobby and compose your own songs this doesn't sound out of reach to me
<d1b2>
<daveshah> Parody is a bit questionable here given it does have some degree of marketing purpose
<whitequark>
electronic_eel: the other video was based on "slap chop"
<whitequark>
so i feel like this is based on some popular work from around 2013
<whitequark>
maybe a bit earlier
<modwizcode>
Parody can still be used for a commercial purpose. It just shouldn't interfere with the existing purpose
<modwizcode>
for some reason 2011 seems like the right year to be looking around
<whitequark>
it could be 2011, slap chop is from 2009
<whitequark>
electronic_eel: in particular the video composition seems clearly based on *something*
<electronic_eel>
it reminds me a bit of beasty boys, but i can't exactly place it
bvernoux has quit [Read error: Connection reset by peer]
aquijoule_ has quit [Remote host closed the connection]