<awygle>
azonenberg: do you put fast-discharge circuits on your FPGA power supplies? To take the voltage down quickly when power is removed?
m_w has quit [Quit: Leaving]
<qu1j0t3>
is anyone here doing decapping? I have a linear chip, dead, LM248N, I can send to a decapper.
<qu1j0t3>
awygle: bitcoin _would_ seem to be a good example. Can you argue it's not?
<cr1901_modern>
Ask digshadow
<azonenberg>
awygle: I do not do fast discharge, no
<cr1901_modern>
qu1j0t3: ^^
<azonenberg>
most of my chips dont have hard requirements for power-down sequencing
<qu1j0t3>
cr1901_modern: thanks. If digshadow sees this pls ping me.
<azonenberg>
awygle: if i did have power-down sequencing i'd throw a greenpak on there and do soft power-up/down
<azonenberg>
But i still wouldn't hard discharge
<azonenberg>
i'd simply sequence the rails off and cut each rail one at a time as the other rails came down
<awygle>
azonenberg: mk, jw. I'm just irritated that I have to hold this rail disabled for like 50 ms so it will discharge all the way.
<azonenberg>
awygle: So you actually have a power-down sequencing rule?
<azonenberg>
For xilinx parts there's a *recommended* sequence to reduce power consumption during boot but generally it's not harmful to the chip to sequence any other way
<azonenberg>
Just not optimal
<rqou>
qu1j0t3: i'm also available to do decap, but imaging needs more work
<rqou>
but i'm not going to be available for like another month
<azonenberg>
I will be available for decap + imaging after i move and renovate the lab :p but that will be closer to a year before my lab is fully online, i think
<pie__>
i was gonna say dont worry it takes a month for mail form upside-down-land but he isnt there anyomre
<pie__>
:P
<awygle>
azonenberg: no, but I have a minimum time below X mV and I need to do a full power cycle
m_w has joined ##openfpga
<awygle>
They don't let me do hardware here anyway so I couldn't add fast discharge if I wanted to
m_w has quit [Client Quit]
m_w has joined ##openfpga
<azonenberg>
awygle: I see
m_w has quit [Client Quit]
m_w has joined ##openfpga
<digshadow>
qu1j0t3: did you ask me on twitter?
<digshadow>
"do you only do digital decapping/microphotography, or would you be interested in a dead LM248N quad op amp? (I killed it accidentally but I want to donate its body to science)"
<digshadow>
sorry, but response is the same...
<digshadow>
can you convince me you will do something fun with it? Otherwise I think I need to focus on existing projects
<digshadow>
also, if you are able to get someone to decap it for me
<digshadow>
like rqou, I can image it
<Zorix>
digshadow, i have a mach210 cpld that is dead.. some kind of internal short.. think it would be possible to recover the data from it by decapping? probably not but i figure id ask
<qu1j0t3>
digshadow: not under this nick. I asked @babbageboole
<rqou>
data recovery is much much harder
<qu1j0t3>
digshadow: that's fine!
<Zorix>
yea.. its too bad, its a chip to one of my amiga 2000 blizzard 68060 cards..
<qu1j0t3>
digshadow: i just thought i'd ping this channel in case anyone else could do it.
<rqou>
ping me again after projects/finals weeks
<qu1j0t3>
me?
<digshadow>
qu1j0t3: this was from @Symbo1ics
<digshadow>
Zorix: there are several ways to do this ,but its a bit of work no matter what. is the data very valuable or are you just curious
<digshadow>
for example, worst case you can use a SEM to measure the charge, but its difficult to prepare the sample correctly for that
<Zorix>
its kinda valuable... could be used to repair some fairly rare accelerator cards for the amiga 2000
<digshadow>
can you get another one working and just clone that?
<Zorix>
been looking for another one for 5+ years
<Zorix>
a few people that i have tracked down that had it basically said no way to taking the chip off the card for cloning
<Zorix>
and i think it might have protection too
<Zorix>
actually, its been 8 years
<cr1901_modern>
digshadow: qu1j0t3 == Symbo1ics. Look very closely at the edge of their avatar
<qu1j0t3>
LOL
<qu1j0t3>
yeah i post there
<qu1j0t3>
digshadow: didn't intend a redundant request. just trying to see if this thing i killed can give its body to science...
<cr1901_modern>
Oh wait, Twitter's lovely image cropping removes the "qu1jot3" part of the avatar
<qu1j0t3>
cr1901_modern: clearly you checked way back when
<azonenberg>
gaah, on phone with comcast asking about service options at the new place
<azonenberg>
the sales guy has never heard of ipv6
<azonenberg>
the technical guy he connected me to at least knows what it is, but has no idea what their offerings are
<azonenberg>
On hold waiting for supervisor now
<Zorix>
they offer ipv6
<Zorix>
i use comcast and i have it
<azonenberg>
Yes but i want a static ipv6 allocation
<Zorix>
ah
<azonenberg>
a /60 or larger
<azonenberg>
so i can run multiple SLAAC subnets
<azonenberg>
I have a business pipe now with one IPv4 address
<Zorix>
they give out 64 bit prefixes
<Zorix>
heh
<azonenberg>
Yeah i need more than that
<azonenberg>
I know you can buy more addresses but i need rate quotes and specifics
<Zorix>
you do know its twice as many prefix bits as the entire ipv4 address space.. heh
<azonenberg>
also the default /64s are probably dynamic
<azonenberg>
Yes but i cant run SLAAC on a /66
<azonenberg>
And some of my gear doesnt even *do* DHCPv6
<Zorix>
hmm
<azonenberg>
it requires a /64 or larger subnet to work at all
<azonenberg>
b/c slaac
<Zorix>
they tend to keep the same network prefix based on.. DUID i think
<rqou>
again, comcrap resi gives you /60 if you request it
<rqou>
but it's dynamic
<azonenberg>
Yeah i need static so i can host services etc
<azonenberg>
Welp
<azonenberg>
Within 4 hours i should get a call from a level 2 tech
<azonenberg>
who can answer this one simple question
<digshadow>
qu1j0t3: well anyway, i think you have a path forward
<digshadow>
if you can get it to rqou and he decaps it, I will image it
<qu1j0t3>
via rqou ? sure
<qu1j0t3>
digshadow: that sounds great.
<digshadow>
I'll probably see him next in two weeks fwiw
<qu1j0t3>
it's not so much that i need to see inside an LM248N _myself_ but more that i wanted to contribute to the overall decapping effort if it's valuable at all.
<qu1j0t3>
rather than tossing it.
<digshadow>
qu1j0t3: he well I have several hundred pounds of chips...
<digshadow>
but I'm happy to hoard chips if you send them my way :)
<digshadow>
qu1j0t3: if you are looking for random ways to contribute, one thing that would be useful would be to update some of the indexes on sipr0n :)
<digshadow>
for example, to make it easier to see if someone has decapped a LM248N
<cr1901_modern>
Okay I'll send him a tweet asking for some advice
<azonenberg>
cr1901_modern: as in something with adjustable DPI, nice resolution, and ideally a few extra buttons
<cr1901_modern>
azonenberg: Not sure that's gonna help with the wrist problems :(
<awygle>
^ extra buttons are key for pcb, no need to hit keyboard for vias etc
<cr1901_modern>
So there's 3 types of trackball I want to test: Full blown trackball w/ hand, mouse with index finger trackball, and mouse with thumb trackball.
<cr1901_modern>
index finger trackball doesn't seem like it would help much, mimicking the motions right here/now in meatspace
<awygle>
Adjustable dpi helps *some* with wrist issues. Not a ton ime
<cr1901_modern>
thumb seems better, if I can learn to control how fast I move the cursor
<cr1901_modern>
full blown is probably best, but then I have to make extra movements to do mouse click :/
<awygle>
If rsi is your issue I *strongly* recommend you at least try a pen tablet
<cr1901_modern>
I mean worth a shot (as long as "pen down/pen up" corresponds to mouse down/mouse up)
<cr1901_modern>
I have to use the keyboard anyway when vectorizing w/ left hand (which of course means I have to relearn the damn inkscape keybindings)
<azonenberg>
awygle: although right now kicad cant use the extra mouse buttons well
<cr1901_modern>
Lord_Nightmare: Still have any of those old crappy trackballs you could send?
<azonenberg>
i dont think anyone ever wrote a tool to do mappings :p
<digshadow>
azonenberg: couldn't you do some generic OS level mouse to keyboard mappings
<digshadow>
ie "this mouse button sends ctrl-z"
<azonenberg>
yes i could
<azonenberg>
but i dont want that to be done globally
<azonenberg>
as i have other apps that support proper mappings and use the extra buttons directly
<digshadow>
fair enough
<awygle>
cr1901_modern: I spent some time with autohotkey and now my tablet buttons are mapped to track, via, etc contextually for CircuitStudio. Bet it'd work for kicad too
<awygle>
Err I guess that was azonenberg
<cr1901_modern>
autoremapping skeeves me
<awygle>
Yeah it's kinda sketchy but I only use the tablet for like four things
<awygle>
So nbd
<Lord_Nightmare>
cr1901_modern: the itac trackballs need repair, the marble mouse is still around though
<pointfree>
OT: Are there any channels/forums about reversible computing? Is anyone else interested in this? I'm considering opening an irc channel if one doesn't already exist.
<Lord_Nightmare>
cr1901_modern: if you want one to fix, i could give you one of the PS/2 ones, they need little tiny screws to hold the pcb down to the case; the original used plastic pegs which broke
<awygle>
pointfree: I'm interested if uninformed
<pie__>
i just realized its actually pretty amazing that things as big as an os dont crash every 5 seconds
<pie__>
let along year long uptimes
teepee has quit [Ping timeout: 248 seconds]
<qu1j0t3>
haha
<qu1j0t3>
multi-year*
<pointfree>
awygle: ##reversiblecomputing the start of a new channel huzzah!
<pointfree>
One more person and it will be a group by definition..
<qu1j0t3>
pie__: i wouldn't say getting to that point is easy. It's taken 70+ years of software engineering practice and we're still not very good at it. :D
teepee has joined ##openfpga
<qu1j0t3>
pie__: but hopefully we don't go backwards _too often_
<qu1j0t3>
despite reversible computing
<pie__>
pointfree, its come up a few times in ##dependent (as in dependent types) but those guys are pretty hardcore x_x
<pointfree>
pie__: good to know, thanks. I'll mention it over there some time.
<pie__>
someone htere has been to a workshop on it but i dont remember who specifically
<digshadow>
azonenberg: didn't carothers have an interest in reversible computing
<azonenberg>
Hmmm
<azonenberg>
i know he did computer arch
<azonenberg>
but i dont remember if he did reversible stuff
<rqou>
azonenberg: we really need some secure element hax :P
<digshadow>
rqou: what happened to it
<rqou>
infineon ROCA vulnerability
<rqou>
something is messed up with the prime number selection algorithm, and generated primes are very not random
<rqou>
but i haven't had time to sit down and understand the details
<rqou>
there was a whole bunch of noise a while back because the academics that discovered it didn't want to talk about it until they had presented at an academic conference
<rqou>
this apparently pissed people off enough and djb et al went and reverse-engineered what the issue was
<digshadow>
hmm interesting
<rqou>
also people pointing out that CC EALn is worthless
<digshadow>
not sure if I would go so far as worthless, but yes, most security certifications are questionable
<digshadow>
IMHO the threat of being audited has some value
<rqou>
not inherently
<rqou>
see: WoSign/StartCom
<lain>
I wonder if they'll send me a replacement yubikey
<lain>
pretty sure mine is vuln to ROCA
<lain>
though afaik that only matters if I generate privkeys on the yubikey, which I wouldn't
digshadow has quit [Ping timeout: 255 seconds]
<rqou>
i mean, neither of my keys are even provisioned
<rqou>
i just wanted another yubikey :P
Bike has quit [Quit: Lost terminal]
<rqou>
the key i actually use is a Neo, which isn't even an infineon part
<rqou>
also, i actually use NFC
nicdev has quit [Read error: Connection reset by peer]
<lain>
I use a yubikey 4 nano
nicdev has joined ##openfpga
<lain>
but the real question is
<lain>
would they let me get a replacement as the usb-c version
<lain>
and save me the expense of having to buy one :P
<rqou>
they should
<lain>
(I don't have the usb-c version, not sure if they care about you getting exactly the same model)
<rqou>
ooh
<rqou>
i'm not sure
<rqou>
hmm probably not
<lain>
ah well :3
* lain
checks yubikey version to be sure
<rqou>
also O_o the YubiHSM gen 2 is actually good
<rqou>
too bad it's $650
<lain>
elaborate?
<rqou>
iirc the YubiHSM gen 1 was more-or-less only designed to verify TOTP-type tokens
<rqou>
the gen2 is a "normal" hsm to do signing operations
<lain>
ah nice
<rqou>
it also supports secp256**k**1 and curve25519
<rqou>
and 256 objects rather than something silly like 4
<rqou>
and it's still cheaper than "real" HSMs
<rqou>
but i'm not running a CA or anything, so i don't really need anything like that
<rqou>
too bad all the tooling surrounding HSMs is absolute garbage (e.g. PKCS#11)
<awygle>
Wow that's pretty cool
* awygle
repeats "I don't need it" to himself until he believes it
<rqou>
i really wish secure elements were more open
<rqou>
azonenberg: time to pull a C<redacted> T<redacted> and break into some? :P
<azonenberg>
if i can break in, i wont use it :p
<azonenberg>
lain: iirc, the bug is only related to the RNG
<azonenberg>
if you generate keys off the device then load it on
<rqou>
what if breaking in requires a FIB?
<azonenberg>
you're good
<azonenberg>
rqou: my general rule is, i avoid security tech i can break
<azonenberg>
if i can pick a lock, i wont use it
<azonenberg>
etc
<rqou>
but CT broke into a TPM
<cr1901_modern>
azonenberg: You have access to a FIB
<azonenberg>
Yes, and? :p
<rqou>
so you don't trust that TPM anymore?
<azonenberg>
cr1901_modern: i've had access to one since like 2013 or so
<azonenberg>
rqou: i dont really like TPMs in general, that whole boot process is kind of a pain
<rqou>
completely offtopic: does anybody know what the difference is between the MPU9255 and the MPU9250?
<lain>
azonenberg: correct @ RNG, I generated my keys on an airgapped machine with no NV storage (but after using a livecd for a bit on a network, so it built up some entropy, because I /also/ don't trust a freshly booted non-networked machine to have much entropy)
<rqou>
wait lain you don't like wireless routers that managed to **reuse primes** across different devices? :P
<lain>
lolol
<pie__>
rqou, dude what xD
<rqou>
yes, this has happened
<pie__>
i mean i get it but...
<pie__>
holy crap
<lain>
anyone else remember the similar key attack some years ago, where they found that tons of servers had similar (in an attackable sense) keys because they generated ssh keys on first boot, and the hardware was similar enough that they all basically had the same entropy
<rqou>
yes, it's the same idea
<lain>
iirc the researchers could actually classify a machine's hardware by key similarity :P
<pie__>
i havent actually looked at the link but yeah
<pie__>
something something collision avoidance failure at collision avoidance demo
<azonenberg>
lain: lol
<azonenberg>
pie__: yeah some guys decided for lulz to collect a few million SSH public keys by scanning the internet
<azonenberg>
and finding the GCD of them
<azonenberg>
A shocking number of them had common factors
<azonenberg>
I signed up for a VPS a while ago that had the same public key as all of their other customers (the SSH private key must have been cloned with the image)
<lain>
OMG
<lain>
I was pleased to see digitalocean's freebsd image generates the keys on first boot
<pie__>
azonenberg, that was different though?
<pie__>
probably thinking of phuctor
<azonenberg>
pie__: There were two different issues
<azonenberg>
Bug #1: embedded devices have poor entropy and often reuse primes
<azonenberg>
Bug 2: VPSes sometimes clone keys with the image
<azonenberg>
Totally unrelated
<azonenberg>
incidentally this is one of the things holding me back from adding SSH/TLS support to my embeded projects
<azonenberg>
I want to make sure i have a solid RNG first
<azonenberg>
And by this i mean an actual hardware noise source
<rqou>
meh, externally loaded keys would be good enough for me
<azonenberg>
session keys? :p
<rqou>
oh, not session keys
<rqou>
i believe the "djb cipher suite" doesn't require a good RNG?
<pie__>
uh
<pie__>
how can you not require a good rng
<azonenberg>
You need one for the ephemeral keys
<rqou>
iirc it uses hash(data)
<pie__>
i mean djb probably expects you to use urandom or something
<azonenberg>
however a predictable key will only compromise that session
<azonenberg>
and not retroactively compromise everything
<pie__>
rqou, oh. idm
<azonenberg>
as long as your long lived key is solid
<pie__>
*idk
<rqou>
ooh right you still need an RNG for the actual symmetric crypto part
<azonenberg>
Yes
<azonenberg>
Exactly
<rqou>
but messing up won't result in a Sony
<rqou>
because the elliptic curve part uses hash(data) instead of an RNG
<azonenberg>
Correct
<azonenberg>
So you lose the session instead of everything
<rqou>
still better
* rqou
is not looking forward to a potential "oops, apparently djb wasn't familiar with <field of mathematics>, everything everywhere is broken"
<pie__>
ow can you use has(data) to make an ephemeral key...
<pie__>
nvm, DH?
<rqou>
yeah, this is just for ECDH/ECDSA
<qu1j0t3>
pie__: Go to bed
<azonenberg>
Signs you've spent too much time working with FPGAs
<azonenberg>
you see a random hex constant that ends in 0x093
<azonenberg>
and think "oh, Xilinx device ID!"
<lain>
see also: always hash your entropy sources
<lain>
like, never output raw entropy
<lain>
RNG vuln -> pop everything forever
day__ has joined ##openfpga
daey has quit [Ping timeout: 276 seconds]
day__ is now known as daey
<awygle>
i've noticed recently it's basically impossible for me to get effective programming work done after heavy exercise
<rqou>
what is this "exercise" you speak of? :P
<lain>
awygle: maybe depleting all your sugars?
<awygle>
lain: could be... but i ate a cheeseburger and fries afterward, at least in this particular case
<awygle>
shoulda got some juice maybe
<awygle>
(fruit, not steroidal)
<lain>
awygle: do you drink water while your exercise? (or drink anything for that matter?)
<lain>
yeah
<lain>
was going to suggest, if you drink lots of water, add some electrolyte powder maybe, or have something like that after exercise
<awygle>
i drink water usually, sometimes IT'S GOT WHAT PLANTS CRAVE instead
<lain>
hahaha
<rqou>
wtf why is ##openfpga full of gym rats?
<lain>
I've noticed plain water can have a negative effect on me sometimes after an intense workout, I think it's diluting whatever's left in my body :P
<awygle>
makes sense
<lain>
rqou: I was fat and now I'm not!
<awygle>
i had an unexpected and _very_ intense hike once and when i got back i wanted to eat handfuls of raw salt
<awygle>
rqou: i'm trying not to become fat
<lain>
at one point I could eat everything and stay like 140 lbs. then my metabolism died :<
<lain>
lol @ handfuls of raw salt
<awygle>
i was 200 pounds from like 16 to 25 but this year i hit 220 and said eff that
<rqou>
i've been pretty lucky with the metabolism so far
<lain>
recently got interested in cooking thanks to sgstair, and now I'm even healthier :P
<lain>
used to be all fast food
<awygle>
i am into cooking because i am a cheap bastard
<lain>
then I went mostly-Soylent for a while during maximum fat burn, just so I could like, meter the intake
<rqou>
wait, do you live with sgstair?
<awygle>
i mostly cook things like "two pork chops and two cups of the Loaded Baked Mashed Potatoes instant mix"
<lain>
rqou: ish :P
<rqou>
why is everybody in this channel in the seattle area?
<lain>
awygle: delicious
<rqou>
come to the bay and be broke! :P
<awygle>
oh lain are you up here too? cool
<lain>
rqou: but you'll fall into the ocean!
<awygle>
rqou: i can put a lower bound on how much money you'd need to pay me to live in the bay and it's pretty high
<awygle>
seattle's great but i need to buy property or something before things _really_ explode...
<rqou>
but but but the bay is bear territory! :P :P
<awygle>
don't let the The Color Red's hear you say that
<rqou>
i believe there are no trees in this channel :P
* lain
has no idea what is going on
<rqou>
Cal/Stanford sportsball rivalry
<awygle>
lain: rqou and i both went to berkeley ("bear territory"). stanford is our rival, and their two mascots are a tree and the color red.
<awygle>
there might also be a bird? i'm not sure
<lain>
oic
<awygle>
not only don't i understa nd python's modules, when they break i have no expectation that i can fix it without help from On High
<azonenberg>
awygle: Hyponatremia is not fun
<azonenberg>
I've seen it up close, $WIFE got it during a long hike on a hotter-than-expected day
<azonenberg>
rqou: My weight management strategy is "bike to work every day, go rescue people in the woods on the weekends, and eat whatever the heck i want" :p
<azonenberg>
awygle: To accept a job in NY or SF you'd need to offer upwards of $500K for me to give the courtesy of reading the offer to the end
<azonenberg>
And i'd still rip it up
<azonenberg>
:p
<awygle>
lmao
<awygle>
i would probably take 500k
<awygle>
... probably
<awygle>
and NY would be way less just because i've never been there
<awygle>
re hyponatremia, what's the actual treatment? i assume it's not "eat handfuls of raw salt"?
<pie__>
treatment is probably osmething lke rehydration solution
<pie__>
any ideas for good keywords for looking for papers related to image processing for extractign schematics from silicon imaging? (or...well...you know what i mean?) , slightly related, looks like http://siliconzoo.org/tutorial.html got some more polish since i last looked at it?
user10032 has joined ##openfpga
<azonenberg>
awygle: Eating salty foods if minor, if more severe then oral rehydration salt solution
<azonenberg>
if it's to the point you can't keep food down, hypertonic saline IV
<azonenberg>
other than that, minimize additional fluid intake
<azonenberg>
reduce activity to avoid sweating out more salt
<azonenberg>
The best treatment is prevention, recognizing heavy sweating and increasing salt intake to compensate
<pie__>
oh right we have our local survival expert :P
<azonenberg>
pie__: lol
<azonenberg>
Wilderness First Responder covers a lot of prevention as well as treatment
<pie__>
azonenberg, id fathom a guess that its hard to a problematic point by yourself and just eating some stuff will generally suffice
<pie__>
*hard to get to a problematic point
<awygle>
pie__: ever been to death valley?
<azonenberg>
pie__: Generally hyponatremia is seen in one of two scenarios
<azonenberg>
First is heavy sweating due to exercise, accompanied by fluid intake with insufficient electrolytes
<azonenberg>
Second is severe, extended diarrhea or vomiting
<azonenberg>
Digestive juices include a lot of salt, which is normally re-absorbed
<azonenberg>
But if you spray it out either end before that happens, you lose that sodium
<pie__>
i see
<azonenberg>
With diseases like cholera etc, the hyponatremia often will kill you before the dehydration does
<pie__>
i never really thought about that
<awygle>
Doesn't low salt levels make you thirsty?
<azonenberg>
it more just makes you feel like crap :p
<azonenberg>
Which is where the WHO oral rehydration salt formula comes in
<azonenberg>
it's got lots of water to replenish that
<pie__>
though maybe i kinda figured something like you all diluting up yo shit would just wash the nutrients out of you
<azonenberg>
but also includes sodium and potassium salts to restore electrolytes
<pie__>
pun not intended...
<azonenberg>
Plus a bit of glucose, because apparently having sugar in the solution helps increase permeability of your bowel so you soak it up faster
<awygle>
I tend to over hydrate, I need to work some electrolytes into that routine on gym days
<azonenberg>
Which is important if 90% of it goes right through you
<azonenberg>
Sports drinks are nice for lighter workouts but for heavy stuff they often arent enoguh to replace everything you sweat out
<pie__>
huh.
<awygle>
They're also an enormous amount of sugar
<pie__>
(better drink my own sweat :P)
<azonenberg>
Yes, way too much for real electrolyte balancing
<azonenberg>
I keep a packet or two of ORS powder in my SAR med bag since dehydration is like the #1 complaint our patients have
<awygle>
I need to find something a little better in that front
<azonenberg>
if they're not sweating too heavily you can give them pure water or a dilute solution
<awygle>
I usually drink juices when not water
<azonenberg>
if they've been sweating a lot or have GI issues, make it full strength
<pie__>
idk, is there some kind of homemad rehydration solution thats the best thing ever(tm) or is it easy to get the proportions off
<pie__>
*homemade after-workout
<azonenberg>
if you google "oral rehydration salt" the WHO formula is pretty easy to find
<pie__>
i glanced at the wiki page a while back
<azonenberg>
there's various homebrew formulas that conme close
<azonenberg>
really you just want sodium and potassium salts plus a bit of sugar
<pie__>
azonenberg, but i figured thats generally a medical emergencyish thing or something, idk about long term use suitability
<azonenberg>
The concentration has to be adjusted based on the intensity of the exercise etc
<azonenberg>
Yeah its also not the most palatable
<awygle>
There's a pretty wide band of acceptability for salt
<azonenberg>
the ideal prevention is just to drink plenty of fluids and keep eating
<azonenberg>
Don't just drink
<azonenberg>
I've never had problems with it myself, but my typical diet during heavy exercise is rather salty
<awygle>
Yeah I'd guess food would be better for low level stuff
<awygle>
Liquid more for acute cases
<azonenberg>
Looking at my spreadsheet, the first 24 hours of my food supply for the SAR training exercise this weekend
<azonenberg>
is 3174 calories and 4274 mg of sodium
<azonenberg>
Along with 5 liters of water
<awygle>
"400% of your recommended daily value!"
<azonenberg>
i probably wont burn through all of that in 24 hours but i budget some safety margin
<azonenberg>
2600-2800 cal is my typical daily consumption during heavy activity
<awygle>
(I know rdv are mostly useless)
<pie__>
awygle, re: satellite, ugh bad day indeed
<azonenberg>
but i've gone closer to 3k in extreme situations
<pie__>
re: hypothetical multiple satellite failures, inb4 screw came loose or someone left a wrench in "Glavkosmos, the company that arranged the secondary payloads for the July mission as well as this launch, said there were no issues with the vehicle that would have caused satellites to malfunction, despite reports that several of the more than 70 spacecraft launched on that mission subsequently failed."
<pie__>
though afaict thats actually referring to some *earlier* launch
<azonenberg>
awygle: The hard numbers are 138g fat, 397g carbs (178 g sugar), 86 g protein
<awygle>
pie__: have you seen planetes?
<pie__>
awygle, i read a little bit of the manga and saw the first episode iirc
<azonenberg>
awygle: the other thing i do is, i don't normally have 3 large meals
<awygle>
"screw came loose" always makes me think of the opening
<pie__>
ahhh lol
<pie__>
i vaguely remember something about a tether? might have been something else
<azonenberg>
awygle: I tend to bring on the order of ten 300 calorie portions and eat them pretty evenly throughout the day, with one of them being a hot meal for dinnertime
<awygle>
I enjoyed it. Also space brothers is way better than something called apace brothers should be
<awygle>
azonenberg: I've been thinking of trying something like that in daily life. Going to five meals to even out glucose through the day
<azonenberg>
awygle: I actually eat like that routinely
<azonenberg>
I have small meals and lots of snacks in between
<awygle>
Currently I might as well not be at work from about 11 to noon. Some days I literally stalk the office looking for snacks.
<azonenberg>
not for any particular dietary reason, just how my urges to eat work
<azonenberg>
i'll have a clif bar or something out of my pocket en route to work since i often leave without time for breakfast
<pie__>
awygle, oh shit dude
<pie__>
i was reading space brothers the other day
<azonenberg>
grab a snack from the break room
<pie__>
for some reason there sonly 3 volumes of like 24 translated :(
<awygle>
azonenberg: does ioa need a smart engineer easily motivated by available snacks?
<azonenberg>
an hour or two later i either eat half the sandwich i brought, or go to subway and buy a footlong and eat half of it
<pie__>
did they make an anime of that or something?
<azonenberg>
Then i have another snack or two through the afternoon
<awygle>
pie__: very yes
<azonenberg>
Nuke the other half of the sandwich before riding home
<awygle>
There's like 200 episodes
<azonenberg>
then have another small meal in the evening at like 21:00
<pie__>
awygle, how have i not heard of this
<pie__>
question is, does it measure up
<awygle>
Okay as much as I love anime and life hacking my #1 life hack is getting enough sleep. Goodnight all
<azonenberg>
awygle: Lol, we actually are looking for senior people at this point (hardware RE experience, ideally some experience in security consulting etc)
<azonenberg>
But not engineers so much as hackers
<pie__>
awygle, wait no this is something else...
<azonenberg>
i.e. not much forward engineering going on
<awygle>
azonenberg: I is babby. Maybe I should apply anyway tho
<cr1901_modern>
azonenberg: Does REing a 30 year old chip no one cares about count?
<azonenberg>
cr1901_modern: Silicon isnt a big enough portion of our work for that to be a hiring priority
<azonenberg>
its cool, but you have to be able to do firmware
<cr1901_modern>
You mean glitching and stuff? Yea never done that
<pie__>
awygle, ok what i was thinking of was Moonlight Mile
<azonenberg>
well firmware code review etc
<azonenberg>
not just side channel
<cr1901_modern>
You mean "opening IDA and hating yourself every second"?
<cr1901_modern>
And how can you do code review if you can't get to the firmware :P?
<pie__>
azonenberg, just out of curiosity you guys got people doing formal methods?
<azonenberg>
pie__: No, it's a research interest of mine but i havent had to do it at $DAYJOB
<azonenberg>
cr1901_modern: The customer gives you the code :p
<awygle>
cr1901_modern: I suspect *most* people hand them the code and say hey tell us why we're idiots
<azonenberg>
i do maybe 30% black box
<cr1901_modern>
Oh, that
<awygle>
Though idk
<cr1901_modern>
Yea, not much experience w/ that
<cr1901_modern>
I started microcorruption a few months back (don't spoil it for me)
<azonenberg>
grr
<cr1901_modern>
I should finish it
<azonenberg>
"undefined reference to ioctl"
<pie__>
awygle, "What did someone put the accelerometer in upside-down again?" wtf i swear i keep seing mentions of this everywhere xD
<pie__>
"Actually, more worrying than the actual failure is the fact that in 2017 they don’t seem to have near-instantaneous knowledge of the situation with their vehicle and spacecrafts. Second stage et al video alone would have let them know what was happening." this is also what i find strange
indy has joined ##openfpga
massi has joined ##openfpga
<azonenberg>
rqou: btw, i fixed the baud rate issue
<azonenberg>
Can you test?
<rqou>
er, i didn't even test last time i complained
<rqou>
uh, you're always using termios2?
<azonenberg>
Yes, i did that a couple hours ago
<rqou>
looks fine on linux, don't have time to test right now
<rqou>
not gonna work on macos/bsd though
<azonenberg>
File another ticket for that? There's no windows support already so that has to get fixed at some point
<azonenberg>
(actually, file a ticket for that too?)
<azonenberg>
against xptools
<azonenberg>
Not very cross platform now :p
<rqou>
nothing but 8N1?
<azonenberg>
As of now, yes
<azonenberg>
i havent found any hardware that does anything different
<azonenberg>
my fpga uart core is 8n1 only too
<rqou>
i've personally used a device that was 8N2
<rqou>
it was weird
<rqou>
iirc smartcards also need something like that?
user10032 has quit [Remote host closed the connection]
<rqou>
so, my housemate and i just talked about the new firefox and how i have tons of addons that aren't compatible
<rqou>
after i told him what addons i had, he mentioned that this addon set sounded quite old
<rqou>
and then i mentioned that this profile is at least as old as the old v1 firefox sync
<rqou>
that came out in 2011
<rqou>
and then my housemate had the best comment: "you do realize that in the time between you creating this browser profile and today, microsoft developed, launched, and killed an entire OS (windows phone)"
massi has quit [Remote host closed the connection]
waylon531 has quit [Ping timeout: 248 seconds]
Dolu2 has joined ##openfpga
Dolu has quit [Read error: Connection reset by peer]
<digshadow>
rqou: I feel like I've had performance issues with the new fireforx
<digshadow>
and have been thinking of switching to all chrome
<balrog>
digshadow: new firefox being which version?
<digshadow>
balrog: hmm 57.0 I guess?
<digshadow>
the logo changed
<digshadow>
ha
<digshadow>
evidently its supposed to be faster :P
<balrog>
do you have a lot of plugins/extensions?
<balrog>
I've noticed a performance increase
<balrog>
(or lots of history)
<digshadow>
balrog: I do have adblock, which I have noticed can lead to some bad performance
<balrog>
try ublock origin
<balrog>
adblock and abp load a massive CSS file into every page (iirc)
<balrog>
Lord_Nightmare knows a bit more about the internals of that
<Lord_Nightmare>
i havent kept up with adblock plus
<Lord_Nightmare>
i just dealt with an issue yesterday on my dad's office pc remotely where he couldn't click on any google link in search results, caused by some sort of profile corruption interacting with noscript
<Lord_Nightmare>
firefox 'refresh' to nuke the profile fixed it
<Lord_Nightmare>
technically probably a noscript bug when the whitelist has more than 1000 entries in it
<Lord_Nightmare>
or so
<Lord_Nightmare>
and don't ask why it had so many
Dolu2 has quit [Ping timeout: 276 seconds]
<Lord_Nightmare>
if your firefox has a pair of "capability.policy.maonoscript.*" entries visible in about:config, delete them (right click on the entry and hit 'reset')
<Lord_Nightmare>
that's ancient old "CAPS" entries which noscript hasn't created since pre-firefox-31 but if it sees them it will use the data in them!
digshadow has quit [Ping timeout: 268 seconds]
<balrog>
Lord_Nightmare: thoughts on uMatrix?
<Lord_Nightmare>
havent tried it
<balrog>
uMatrix also lets you filter non-script network requests
<Lord_Nightmare>
as does noscript
<Lord_Nightmare>
click custom and you can set whether it allows script, object, media, frame, font, webgl, fetch or 'other' requests or not
<Lord_Nightmare>
untrusted blocks everything
<Lord_Nightmare>
'default' blocks scripts objects and fetch
<Lord_Nightmare>
trusted allows everything
Dolu has joined ##openfpga
Dolu has quit [Read error: No route to host]
Dolu has joined ##openfpga
digshadow has joined ##openfpga
Hootch has quit [Remote host closed the connection]
Hootch has joined ##openfpga
<awygle>
How complicated do people's Yosys synthesis scripts usually get?
<awygle>
Seems like there are a lot of passes but for most operations there's a pretty streamlined path..
<azonenberg>
Mine are generally almost trivial
<azonenberg>
a bunch of read_verilog, then either write_smt2 or synth_greenpak4
<azonenberg>
the exception is some of my bitstream RE work
<azonenberg>
That comment may be old, i think i fixed the bug? but the project kinda got tabled after that con while i worked on billable stuff
<azonenberg>
in any case me and rqou wrote that pass
<azonenberg>
i dont remember if it got merged to mainline yet or not
<azonenberg>
My fork has a lot of extra RE-specific code that isnt merged b/c clifford wanted tweaks to it or, in some cases, wanted to completely redo the structure
<azonenberg>
I do remember extract_reduce sometimes dying on certain netlists and not having time before the con to debug
<awygle>
azonenberg: I am zero surprised that bitstream RE is a special case lol
<awygle>
Glad to hear the usual matches my experience
<awygle>
Do you write long complicated synthesis scripts anywhere else? Vivado maybe?
<balrog>
I suggested to someone to use the open ice40 toolchain for a project, and they ran into that :D
<awygle>
I have done a little hacking on arachne, the code is weird but not tooooo weird
<azonenberg>
i've never written any vivado tcl
<azonenberg>
i'm going to be creating a command line non-project flow soon for Splash to use
<azonenberg>
All of my command line builds now are ISE
user10032 has joined ##openfpga
<awygle>
lol azonenberg calling me out on Twitter :-P
<awygle>
Related to that tweet, anybody know how parallel synthesis is?
<awygle>
It's not the bottleneck but it's also not 10x less, ime
<azonenberg>
I want to explore paralleism in yosys as a separate project
<azonenberg>
parallelism*
<rqou>
i honestly don't care about this
<rqou>
yosys is already super fast
<azonenberg>
Yes, PAR is much more important
<azonenberg>
But if we can make synth faster we should do that as well
<rqou>
i believe most of the effort needs to go into abc
<balrog>
stupid question that I should know the answer to: what is the equivalent to "linking" in HDL context?
<balrog>
or is everything basically #include'd?
<rqou>
um.. "elaboration"?
<azonenberg>
i would actually say flattening the netlist
<azonenberg>
Up until then you have a bunch of separate hierarchial blocks with pointers to each other
<balrog>
and is there an equivalent to link-time optimization?
<azonenberg>
Yes
<Dolu>
I'm realy not sure that linking is somwere equivalent to elaboration (it depend what you mean by elaboration) For me elaboration mean instanciating the hardware from its description
<azonenberg>
You can optimize before or after flattening
<azonenberg>
Optimizing before flattening doesnt cross hierarchial boundaries
<azonenberg>
So for example if i pass !x into a port
<azonenberg>
and the module then inverts that port
<rqou>
i thought optimizing always happened after?
<azonenberg>
it won't optimize out the pair of inverters
<azonenberg>
After you flatten, a second optimization pass will
<Dolu>
Defining a module is like defining a class, then doing it's elaboration is like instanciating your design
<azonenberg>
balrog: basically, all hardware ends up being statically linked
<azonenberg>
or inlined actually would be better
<azonenberg>
You replicate each gate/module as many times as you intantiated it
<azonenberg>
So global optimization would be done after you inline it
<azonenberg>
Initial optimization within each module cant cross those boundaries
<azonenberg>
Typically you'd optimize at both points, because you dont want to repeat that initial effort for every copy of a module
<rqou>
er, you skipped over parameters/generics
<azonenberg>
Yes
<azonenberg>
That happens during elaboration
<azonenberg>
the "hierarchy" pass in yosys
<azonenberg>
First you parse the HDL into an AST
<azonenberg>
Then you run the elaboration to replace parameterized modules with fixed sizes for regs etc
<azonenberg>
then you flatten out processes into cells
<azonenberg>
do various optimizations and transformations on the cells
<whitequark>
I wonder if you could easily have parts of your design techmapped so you don't need to constantly recompile them
<azonenberg>
whitequark: yes, you can read_json / read_ilang
<whitequark>
like I'm hacking on this tiny FSM
<azonenberg>
to pull in a synthesized module
<azonenberg>
at any level of synthesis
<whitequark>
don't want to rebuild the entire SoC
<whitequark>
mm
<azonenberg>
So you could just pull that in after techmapping the rest
<azonenberg>
then flatten and optionally do final global optimization
<azonenberg>
Yosys is fast enough that for most stuff i dont see the point right now
<cr1901_modern>
^
<azonenberg>
more important would be supporting hard macros for PAR
<whitequark>
yes
<azonenberg>
where you can re-PAR one module at a time and keep the rest as is
<azonenberg>
That would require better stability in yosys as far as post-synth instance/net names
<azonenberg>
So you can trace what's what
<cr1901_modern>
pnr is overwhelmingly the longest part of synthesizing misoc for me
<azonenberg>
This is one of thethings i want to fix with ABC
<azonenberg>
Look at how ISE does net naming
<azonenberg>
it may not be great but it's relatively consistent and you can usually tell what's what
soylentyellow has joined ##openfpga
<awygle>
Yeah abc is clearly a huge problem
<rqou>
O_o i just got an ad for edmund optics
<rqou>
i didn't even know they ran ad campaigns
<rqou>
and i don't even actually do optics research
day__ has joined ##openfpga
daey has quit [Ping timeout: 248 seconds]
day__ is now known as daey
user10032 has quit [Quit: Leaving]
<pie__>
awygle, apparently an UPSat guy is holding a ccc talk