john_cephalopoda has left ##openfpga ["Trees can see into your soul."]
rohitksingh has joined ##openfpga
rohitksingh has quit [Quit: Leaving.]
kuldeep has quit [Ping timeout: 246 seconds]
clifford has quit [Ping timeout: 272 seconds]
kuldeep has joined ##openfpga
<fseidel>
got a free Cyclone III dev board that the department was throwing out :D
<fseidel>
Terasic DE0
<qu1j0t3>
NICE
<qu1j0t3>
i have a cyclone I terasic
<qu1j0t3>
i still hope to make it blink smething one day
<fseidel>
sadly it requires Quartus 13.1 or earlier and I don't know of any free alternatives
<fseidel>
wow, how old is that Cyclone I?
<rqou>
you can always just use a max v and pretend it's a cyclone 1 :P
<qu1j0t3>
fseidel: Old Quartus is still downloadable.
<qu1j0t3>
fseidel: verified this like 2 weeks ago
<rqou>
(tile structure is identical, interconnect and bit encoding unknown)
<rqou>
QUIP explicitly states that device-level netlists targeting cyc1 work on maxii
<shapr>
Does anyone know of a graph of per-lut cost for various chips?
<shapr>
I want the best per lut price for yosys supported chips.
<shapr>
I could go graph all the chips myself, could be fun
<fseidel>
yeah, I have a VM with 13.1
<fseidel>
and it works fune
<fseidel>
*fine
<fseidel>
I just wish I could use something more recent
<cr1901_modern>
Fun thing I never really thought of till today... ice40xx8k is only _six_ times larger than 1k, not 8 times
<fseidel>
well that's misleading
<cr1901_modern>
must be fun to use rounding to push marketing
<shapr>
but it is almost 8k luts?
<rqou>
cr1901_modern: "40 LE" max vs lol
<cr1901_modern>
yes, 1280 vs 7860 LUTs
<fseidel>
I'll probably build a daughter card with a few YM sound chips on it and do something silly, should be a fun project to stick on my backlog
<fseidel>
the board has VGA out but no audio DAC, sadly
<shapr>
my backlog overfloweth
<fseidel>
right now I need to tackle the SDRAM controller and bus arbiter reference solution for an assignment we're handing out to students in like a month
<fseidel>
the deadline means it'll actually fucking happen :L
rohitksingh has joined ##openfpga
rohitksingh has quit [Ping timeout: 245 seconds]
<shapr>
yeah, that's how I get anything out my backlog
<shapr>
I signed up to teach FPGA dev to a bunch of my coworkers
<awygle>
yeah i have no real dog in cap'nproto vs flatbuffers
<rqou>
it'll really depend on the details imo
<awygle>
both seem better to me than protobufs except maybe for language support (which i have not dug into)
<awygle>
but iirc wq is a flatbuffers fan
<rqou>
in general though I'm increasingly of the opinion that all APIs should be either data-structure-centric or rpc-like (and emphatically not classic Java OO style)
scrts has quit [Ping timeout: 245 seconds]
<awygle>
yup
<rqou>
so... want to work on ParasJTAG? :P
<rqou>
still a crab (kinda), and parasitic because it's yet another rewrite that'll steal code from other projects
<awygle>
lol
<rqou>
been thinking about the API and it'll almost certainly be designed around a rpc-style interface
<awygle>
i would like to have design input
<awygle>
(starting with the name)
<awygle>
if you are interested in my thoughts
<awygle>
i can't promise to write code, i've capped my stack at its current height
<rqou>
don't like projects named after porkymans? :P (ps don't google that)
* awygle
wasn't born yesterday
<rqou>
heh
<rqou>
example of seo not being a good thing? :P
<awygle>
na i just know when something sounds like i don't want to google it
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
azonenberg_work has quit [Ping timeout: 252 seconds]
azonenberg_work has joined ##openfpga
Miyu has joined ##openfpga
<azonenberg_work>
whitequark: propofol is quite the interesting drug
Miyu has quit [Ping timeout: 252 seconds]
<azonenberg_work>
whitequark: i knew it caused anterograde amnesia but it appears to have a short retrograde effect on me as well
<rqou>
just don't use it because you can't sleep :P
<rqou>
too soon?
<azonenberg_work>
rqou: lol
<azonenberg_work>
When they pulled my wisdom tooth this morning they administered (in order) fentanyl, midazolam, and propofol
<azonenberg_work>
(I was a bit surprised as the doc had told me they were going to use ketamine instead of fentanyl during the pre-op visit)
<rqou>
heh, whitequark's favorite intoxicant :P
<azonenberg_work>
According to the nurse i made some comment about the propofol as they were giving it to me, indicating i was conscious enough to talk and understand what was going on (i.e. not knocked out from the midazolam yet)
<azonenberg_work>
But i have no memory of that
<azonenberg_work>
I remember the IV push of midazolam and thinking "I don't feel anything yet" then waking up with the IV and monitors disconnected and being helped into the wheelchair
<rqou>
is there a reason this procedure has to be done under general anesthesia?
<azonenberg_work>
then it's a bit fuzzy until i got into the passenger seat of the car, ally tells me i was talking up a storm about technical aspects of the procedure but i can't recall any of that
<azonenberg_work>
rqou: Quirks of my anatomy and metabolism
<azonenberg_work>
1) I have a nerve very close to the root of the tooth (so difficult to numb with only a peripheral nerve block using lidocaine or similar)
<azonenberg_work>
2) I metabolize lidocaine unusually rapidly
<azonenberg_work>
Every time i've had a peripheral nerve block for tooth drilling, my other 3 wisdom teeth, ingrown toenails when i was younger, etc it wore off unusually soon and I had to ask for more midway through the procedure
<azonenberg_work>
My previous wisdom tooth was an unusually serious offender in this regard, they had to give me 5+ additional injections during the procedure
<azonenberg_work>
This caused some rather nasty effects in the mid term since the anesthetic they used was lidocaine plus epi
<azonenberg_work>
Apparently i do *not* metabolize epi super fast
<azonenberg_work>
So i had a zillion little holes in my jaw muscle plus prolonged vasoconstriction in the area
<azonenberg_work>
from the epi overdose associated with the extra lidocaine :p
<rqou>
hmm, do we have general anesthetics that don't inhibit memory formation?
<azonenberg_work>
The end result was that any time i tried to open my mouth more than about ~20 mm, for over two weeks after the procedure, i felt sudden tension (like i was hitting some kind of obstacle, probably scar tissue)
<azonenberg_work>
and extreme pain
<azonenberg_work>
(with my mouth closed it didnt hurt at all)
<azonenberg_work>
The actual surgical site didnt really bother me after the first ~2 days, i took one 800mg ibuprofen and that was it
<azonenberg_work>
and well, if you're totally unconscious you will have no memory because you didn't really experience the procedure
<azonenberg_work>
Propofol is interesting in that it causes amnesia even when used at low doses for procedural sedation in which you're still awake enough to respond to commands to open your mouth more, etc
<azonenberg_work>
But people normally don't *want* to remember their surgery, so this isnt necessarily a bad thing
<cr1901_modern>
This is a person who couldn't remember who he was for a decade, but could remember how to run a restaurant
<rqou>
hmm, i somehow think that "awake enough to respond but can't form memories" is even more creepy
<azonenberg_work>
rqou: in any case propofol is not an analgesic, all it does is knock you out
<azonenberg_work>
Which is where the fentanyl comes in
<azonenberg_work>
I'm not quite sure what the purpose of the midazolam is... when used in conjunction with ketamine it apparently reduces the psychoactive side effects
<azonenberg_work>
but they didnt give me ketamine
<azonenberg_work>
rqou: another advantage of propofol for short procedures like a tooth extraction is the *extremely* short duration of effect
<rqou>
yeah i know about that one
<azonenberg_work>
<10 minutes, although interestingly enough the elimination half life is much longer
<rqou>
since michael jackson abused that property
<azonenberg_work>
it gets absorbed by other tissue and stops having an anesthetic effect super fast, then is slowly metabolized and excreted over the course of hours
<azonenberg_work>
But this means that you dont have a prolonged drowsy period when waking up in a recovery room, you can be discharged pretty quickly
<rqou>
one property i find "fun" about many anesthetics is that they're fat-soluble
<rqou>
so fat people have slower recoveries
<azonenberg_work>
If something goes wrong and (say) you stop breathing, all they need to do is wait for a couple of minutes of artificial ventilation and it wears off
<azonenberg_work>
no need to blast you with narcan or anything
<azonenberg_work>
(actually i dont even think narcan is an antidote to propofol since it's not an opiate... it would work on the fentanyl though)
<azonenberg_work>
whitequark: you're somewhat familiar with benzodiazepines, right? What's the point of midazolam in this scenario?
<whitequark>
azonenberg_work: sec
<whitequark>
reading backlog
<whitequark>
rqou: if they werent fat soluble
<whitequark>
they wouldnt work
<whitequark>
because they have to cross BBB
<whitequark>
and that involves crossing lipid bilayers
<rqou>
ah, that kinda makes sense
<whitequark>
azonenberg_work: midazolam serves a double purpose
<azonenberg_work>
whitequark: thats only required for general right?
<azonenberg_work>
a local need not be lipid soluble?
<whitequark>
first it sort of knocks you out so you don't freak the hell out, many people fear surgeries
<whitequark>
and second the amnesia effect is desirable
<whitequark>
because sometimes anesthesiologists fuck up
<azonenberg_work>
whitequark: lolol
<whitequark>
and you dont want the patient to remember that
<whitequark>
re local
<whitequark>
depends
<azonenberg_work>
and propofol does that, no?
<azonenberg_work>
(amnesia plus knocking you out)
<whitequark>
lets say they fuck up and propofol wears off but the muscle relaxant doesnt
<whitequark>
now you're stuck inside your body and cant move but see and feel everything
<azonenberg_work>
I understand that, and i understand the purpose of the fentanyl (analgesia)
<azonenberg_work>
Just not sure why add the benzo on top
<whitequark>
those two things basically
<whitequark>
you dont strictly need it, i think its customary in the US to a degree
<whitequark>
like its customary to use n2o in dental practice
<azonenberg_work>
So is it literally just insurance in case the propofol wears off too soon?
<azonenberg_work>
and is not strictly necessary?
<whitequark>
dont forget the anxiolytic effect for easily entering
<whitequark>
say you fear needles or something
<whitequark>
or just generally fear surgeries
<azonenberg_work>
whitequark: and the propofol wouldn't do that?
<azonenberg_work>
All of the drugs were IV push, so one needle and everything plus a bunch of ringer's lactate through that
<whitequark>
different people have different reactions to propofol
<rqou>
ah, not saline? :P
<whitequark>
iirc depending on whether you're a casual drinker you might need dosage adjustment within like a 2x range easily
<whitequark>
casual meaning "you drank any alcohol at all in the last month" basically
<whitequark>
and on some it just works very badly
<rqou>
wait why?
<whitequark>
no one fucking knows
<whitequark>
anesthesia is a dark art
<azonenberg_work>
rqou: i was actually wondering the same thing
<rqou>
do we actually have any idea _how_ any of these drugs work?
<whitequark>
well ok, this case is probably because of liver metabolism
<azonenberg_work>
(ringers vs NAS)
<azonenberg_work>
NS*)
<whitequark>
rqou: not really
<whitequark>
we dont have a working theory of general anesthesia
<azonenberg_work>
But he either never responded
<azonenberg_work>
or i was unconscious by the time he did :p
<whitequark>
we know it works on all biological kingdoms
<azonenberg_work>
i did comment on it
<whitequark>
in similar dosage
<whitequark>
including plants apparently
<rqou>
probably nobody wants to address the shortage :P
<whitequark>
which suggests a very general mechanism
<rqou>
wait what
<rqou>
plants?
<whitequark>
yes
<azonenberg_work>
rqou: lol
<rqou>
like... how?
<whitequark>
like venus flytrap
<rqou>
wtf
<azonenberg_work>
rqou: i assumed ringers and saline would be made in the same factories
<azonenberg_work>
but possible thats not the case
<rqou>
whitequark: so... how did somebody get crazy enough to decide to try that?
<whitequark>
>Mimosa leaves, pea tendrils, Venus flytraps and sundew traps all lost both their autonomous and touch-induced movements after exposure to anaesthetics.
<rqou>
W T F
<whitequark>
the most credible hypothesis i've seen
<rqou>
wait, xenon is an anesthetic?
<whitequark>
oh yeah
<whitequark>
even better
<whitequark>
*nitrogen* is
<whitequark>
we are all permanently under nitrogen narcosis and this is detectable
<whitequark>
sec
<rqou>
i did see the "humans work better without nitrogen" paper you linked
<whitequark>
yeah
<azonenberg_work>
whitequark: so switching to a nitrogen-free atmosphere improves performance?
<azonenberg_work>
innteresting
<whitequark>
yes
<whitequark>
slighly
<whitequark>
by like 5% or something
<rqou>
lifehack :P
<whitequark>
i wanna get an oxygen concentrator someday and try it
<whitequark>
based on PSA
<whitequark>
thing is it's not super good for your lungs i think
<rqou>
wait, so vegas "oxygen bars" might not be 100% bullshit?
<azonenberg_work>
whitequark: do you actually need an o2 conc? (since pure O2 isnt great for you)
<whitequark>
yes
<azonenberg_work>
What about say heliox
<azonenberg_work>
or something like that
<whitequark>
azonenberg_work: i wont waste so much helium
<azonenberg_work>
i.e. 21% O2 mixed with an inert gas other than nitrogen
<whitequark>
but yes heliox works
<azonenberg_work>
I dont literally mean helium, more like "anything but n2"
<whitequark>
it might be possible to use argon since argon isn't a GA at partial pressures of 0.8 atm IIRC
<whitequark>
but it's expensive af
<whitequark>
you could *probably* make another PSA to recover argon
<whitequark>
but im not nearly good enough to make that
m4ssi has joined ##openfpga
<whitequark>
azonenberg_work: neon is definitely not a GA
<whitequark>
but that's even more expensive
<whitequark>
rqou: anyway, so the thing about GAs
<whitequark>
is that all of them are small hydrophobic molecules
<whitequark>
the working theory is that proteins have these hydrophobic pockets that are empty of anything
<whitequark>
as in there's vacuum inside. it's not thermodynamically advantageous to fill them with water and noting else fits
<whitequark>
and they fill those pockets
<whitequark>
and subtly fuck up most of the proteins in your body
<whitequark>
enough to disable consciousness but not enough to actually harm you in any meaningful way
GuzTech has joined ##openfpga
<whitequark>
this is of course not applicable to ketamine which is simply an NMDA antagonist, water soluble too
<rqou>
and how do they just manage to hit this balance without killing you?
<whitequark>
no. one. fucking. knows.
<whitequark>
GA has produced some of the most insane theories in biology
<whitequark>
for example, the soliton theory of nerve conduction
<GuzTech>
I love joining IRC channels in the middle of a conversation, because then I can play a game called "Guess what the topic is!" :D
<whitequark>
which says that electrical potentials along a nerve are actually a *side effect*
<rqou>
GuzTech: it's about fpgas, duh :P
<whitequark>
and the actual information is transmitted using a sort of longitudal pressure waves
<rqou>
what
<whitequark>
this is exactly how the scientific community went
<GuzTech>
rqou: biological fpgas of course :P
<whitequark>
i *think* it's finally been disproven recently, but it had stood mostly untouched for like decades
<whitequark>
mostly because it was too insane to consider
<whitequark>
but not crackpot enough to outright reject
<rqou>
and what happens when there is a demyelination disease?
<whitequark>
oh they had some function for myelin too
<whitequark>
rqou: have i mentioned the other insane theory
<whitequark>
which is that your nose is an IR spectrometer
<whitequark>
and that's one of the leading insane theories.
<whitequark>
see the thing about smell is that there's a structure-smell relationship
<whitequark>
to the degree that perfume chemists can practically match an NMR just by smell
<whitequark>
for some classes of compounds anyway
<whitequark>
and no one can explain that very well.
<whitequark>
like how the fuck do you smell functional groups?
<whitequark>
that doesn't make any conventional sense, this is not how receptors normally work anywhere
<GuzTech>
Years ago my chemistry teacher told us that a nice smelling perfume MUST have a horribly smelling component in it.
<whitequark>
anyway, so the way Luca thinks this works is receptors in your nose have an electron circuit with a tunneling element
<whitequark>
and tunneling depends on the bond oscillation frequency of the functional group stuck in the middle of that circuit
<GuzTech>
The example he gave was something like a soccer team training for a week with the same socks, then putting them in a bag and leaving it under the sun for a month. That kind of horrible smell...
<whitequark>
yeah, this is true
<whitequark>
scatole smells like flowers in low concentrations
<whitequark>
and you can guess how it smells in high concentrations from the name
<whitequark>
no one can coherently explain that either btw
<GuzTech>
Is it known why dogs for example can smell much more sensitively than us?
<whitequark>
that's actually not true
<GuzTech>
Larger area of exposed nerves?
<whitequark>
more careful studies show that humans aren't much worse than dogs at actually distinguishing smells
<whitequark>
we're just not trained to do so, and we use our noses very crudely
<whitequark>
sec
<GuzTech>
hmm
<rqou>
uh, so dumb question: how did we figure out the smells of "exciting" compounds like boranes?
<whitequark>
rqou: it was a simpler time
<whitequark>
chemists would work without PPE and determine if the reaction was successful by a good ol' whiff
<rqou>
uh...
<whitequark>
in fact
<whitequark>
until fairly recently (in global terms) chemists rejected PPE because it wasn't manly enough
<whitequark>
just like doctors rejected washing hands like 200 years ago for same reason
<whitequark>
hashtag men :P
<GuzTech>
Yeah that is always a good argument in scientific circles: "I reject your proposal, because it isn't manly enough."
<whitequark>
GuzTech: back when a doctor suggested that washing hands would greatly reduce rates of puerperal fever
<GuzTech>
You don't want to know how many times I had to turn in a proposal while chewing on rocks for it to be "manly enough" :D
<whitequark>
the rest of scientific community drove him to ruin and he died in an asylum
<whitequark>
"Poor human olfaction is a 19th-century myth"
<GuzTech>
Which is not a phrase I use often
<whitequark>
"However, this idea derives not from empirical studies of human olfaction but from a famous 19th-century anatomist's hypothesis that the evolution of human free will required a reduction in the proportional size of the brain's olfactory bulb."
<GuzTech>
Wow
<whitequark>
"Strangely, the idea that humans have tiny olfactory bulbs and a poor sense of smell is derived in part from the religious politics of nineteenth century France. The Catholic Church in France actively fought secularization, including the denunciation of the Paris Faculty of Medicine for teaching “atheism and materialism.”"
<rqou>
i do remember some people (iirc bunnie and maybe you?) linking the "can we analyze a cpu like we analyze brains?" paper
<whitequark>
azonenberg_work: anyway, how are your teeth etc
<azonenberg_work>
whitequark: the ones i still have? they're fine :p
mumptai_ has joined ##openfpga
mumptai has quit [Ping timeout: 246 seconds]
<whitequark>
azonenberg_work: do you think you could do anything with swd
<whitequark>
i'm gonna get an stm32 something to test
<azonenberg_work>
whitequark: I'm going to sleep shortly but do you have the test environment set up yet?
<azonenberg_work>
Also i just found out the adapter i designed (just a ftdi to swd passive bridge) is en route from oshpark
<azonenberg_work>
when it gets here i have a stm32 swd devkit i can test on
<azonenberg_work>
i need to go into the office soonish anyway to do the monthly fire extinguisher inspection
<azonenberg_work>
so i can assemble it there
<rqou>
lol
<rqou>
the consequences of talking too much :P
<azonenberg_work>
Lol yeah
<azonenberg_work>
When you tell HR the fire extinguishers have blank inspection tags and havent been serviced in 4+ years, one hadn't been hydroed in like 8
<azonenberg_work>
You generally become the guy assigned to do the monthly inspection :p
<whitequark>
azonenberg_work: will have soon, i get the stm32 delivered today
<azonenberg_work>
whitequark: ok keep me posted
<azonenberg_work>
as soon as either i or you have working hardware to test with, i can justify spenidng a lot more time on it
<azonenberg_work>
i just dont want to be coding blind for too long without some way to make sure it works
<rqou>
um... what kind of hardware do you need exactly?
<rqou>
i have ftdi thingies and swd boards
<whitequark>
azonenberg_work: do you think you can just hook up the glasgow code?
<rqou>
just not necessarily hooked up
<azonenberg_work>
whitequark: that involves hooking up the swd backend in general
<azonenberg_work>
i have a proposal i need to throw together for a client tomorrow but should be free after that
<whitequark>
azonenberg_work: alright
<azonenberg_work>
whitequark: btw
<azonenberg_work>
The STM32 board you got
<whitequark>
maple mini
<azonenberg_work>
Is it a PURE swd part? Or a SWD+JTAG capable part
<azonenberg_work>
i.e. SW-DP or SWJ-DP
<whitequark>
azonenberg_work: SWD+JTAG
<azonenberg_work>
i fully support the JTAG-DP, and am working on the SW-DP code, but i don't have anything to do the SWJ-DP switching sequence
<azonenberg_work>
(for compatibility, the SWJ-DP boots up in jtag mode by default)
<whitequark>
azonenberg_work: i implemented it in glasgow already
<azonenberg_work>
ah, ok
<rqou>
wait i thought this was automagic?
<azonenberg_work>
in that case i just need to make that a standard api
<azonenberg_work>
rqou: if you are a swd-only host you need to send the switch sequence to enable swd mode
<azonenberg_work>
except when talking to a sw-dp
<whitequark>
or SWDv2?
<azonenberg_work>
i note that presence of a swj-dp does not necessarily mean that the jtag pins are bonded out to your header
<rqou>
in the black magic firmware i just type swdp_scan and it works
<azonenberg_work>
if the board designer used swd only
<azonenberg_work>
whitequark: Yeah that is only partially supported too
<rqou>
what is SWDv2?
<azonenberg_work>
SWDv2 is hard to find, i havent seen any hardware with it
<azonenberg_work>
rqou: multidrop capable
<azonenberg_work>
mainly
<rqou>
wait why?
<rqou>
i thought we deliberately gave up on that?
<azonenberg_work>
not sure why not just use jtag...
<azonenberg_work>
But i think the idea is to trade fmax off against simpler pcbs
<azonenberg_work>
with less pins than jtag?
<rqou>
also, on an arm mcu do you actually gain anything by using full jtag?
<azonenberg_work>
its in the arm standard but until i see silicon implementing it
<azonenberg_work>
i don't plan to add it to jtaghal
<azonenberg_work>
rqou: yes and no
<azonenberg_work>
Having JTAG somewhere in the system allows you to do boundary scan as well as other JTAG-only stuff like reading the JEDEC IDCODE (although some parts may have this memory mapped at a non-standardized address)
<azonenberg_work>
That being said, if the SW-DP gives you access to a JTAG-AP
<whitequark>
azonenberg_work: doesn't SWD have an IDCODE register?
<azonenberg_work>
You can use that to access the boundary scan chain over SWD
<azonenberg_work>
without ever having 4-wire jtag on the pcb
<rqou>
i've literally never seen boundary scan used
<azonenberg_work>
whitequark: No
<azonenberg_work>
whitequark: SWDv1 has a DP ID register that lets you identify the SWD version in use, etc
<azonenberg_work>
The MEM-AP gives you access to ROM tables that allow enumeration of the ARM IP cores in the coresight system
<azonenberg_work>
however that does not uniquely identify the soc or tell you anything about the memory map of anything outside the world of coresight
<whitequark>
ahh
<azonenberg_work>
SWDv2 has an ID register of sorts, i haven't looked at full details on that yet
<azonenberg_work>
One of the fun challenges for jtaghal will be auto-IDing SWDv1 targets
<rqou>
i mean, idcode doesn't really tell you anything like that either
<azonenberg_work>
My current plan is to walk a list of known SFR addresses and read each one
<azonenberg_work>
rqou: if you have a database of some sort, it does
<azonenberg_work>
but the coresight rom can't distinguish one stm32f1 from another
<azonenberg_work>
it's identical as far as i ca ntell
<azonenberg_work>
vs the idcode tells you exact ram, flash, pin count, etc
<azonenberg_work>
you just need a table lookup
<rqou>
er, no it doesn't? :P
<rqou>
in many cases it just pins you down to a particular die
<azonenberg_work>
with stm32 i think it does everything
<azonenberg_work>
maybe not pin count
<rqou>
ah maybe
<rqou>
does it pin down flash size?
<azonenberg_work>
in any case stm32 also has two jtag taps
<azonenberg_work>
I dont remember but there is a SFR for that as soon as you have parsed enough of the IDCODE to know its a stm32
<azonenberg_work>
(hence the STM32Microcontroller class etc in jtaghal)
<rqou>
afaik it's not necessarily accurate?
<rqou>
china seems to have discovered that some stm32s have more flash than the sfr says
<azonenberg_work>
the sfr has the supported amount of flash
<azonenberg_work>
it is possible some dies are fused down from a larger one
<azonenberg_work>
SRAM i suspect is not fused because that adds static power (unless they have two banks and power-gate one of them with an efuse)
<rqou>
i believe i saw people saying that if you just issue the command to access higher sectors it still succeeds
<azonenberg_work>
I know that 350nm PICs are unique dies for each density
<azonenberg_work>
Have not decapped enough stm32s to know
<azonenberg_work>
in any case my plan for swd ID is to bruteforce probe SFRs that contain vendor IDs etc until i have some idea of what the chip is
<azonenberg_work>
then use that information to narrow down the SFRs i know are out there
<azonenberg_work>
and get a full ID
<azonenberg_work>
Additionally, i will eventually add JTAG-DP support
<azonenberg_work>
So you will be able to get an IDCODE that way if you connect to the SWD half of a SWJ-DP
<azonenberg_work>
i dont know what the scan chain looks like over JTAG-DP yet
<azonenberg_work>
havent played with it
<rqou>
inb4 the probe order becomes critical because some probes hard crash some chips
<azonenberg_work>
over 4-wire JTAG, the stm32 scan chain has two taps
<azonenberg_work>
a fixed ID (ARM standard) JTAG-DP, and a variable ID (chip dependent) boundary scan TAP that i primarily use for ID
<azonenberg_work>
rqou: that is a real possibility however i have yet to encounter any such parts in my research
<azonenberg_work>
The hypothetical risk of this is one of the reasons openocd has limited automagic discovery capability in swd mode
<azonenberg_work>
Given how similar Cortex-M memory maps are, i think the risk is generally low
<azonenberg_work>
this doesn't strike me as a terrible idea at all
<rqou>
inb4 v& for itar violations
<azonenberg_work>
not even that
<gruetzkopf>
st_flash doesn't care
<azonenberg_work>
"we set up this random preconfigured server for you to download all your source code to, you'd better trust that we aren't siphoning off a copy"
<azonenberg_work>
for open source or academic it might be useful but i dont see most big companies biting
<gruetzkopf>
why are people so proud about having reinvented the dumb terminal., but worse?
<azonenberg_work>
gruetzkopf: because their customer base is too young to remember VT-100s
<rqou>
there's a cloud^Wbutt this time
<azonenberg_work>
:p
<azonenberg_work>
also i wonder how the cost of their service compares to the cost of just running an RDP server on an AWS instance yourself
<azonenberg_work>
with a copy of vivado stuck on it
<azonenberg_work>
Clearly higher if they are turning a profit
<awygle>
rqou: semi-serious question - why is this nice?
<rqou>
pwning ME and exploring it just seems like fun
<awygle>
mm yeah ok
<felix_>
from platform security point this is realy bad. but being able to poke that deeply embedded system still sounds like fun though
<gruetzkopf>
sooo anyone here with a sandybridge/ivybridge platform that splits the pcie 3.0 x16 port from the cpu?
<felix_>
sadly no. there aren't that many boards supporting that and i didn't find one for cheap in working condition
<felix_>
iirc on ivb there are also 4 pcie lanes in addition to the pcie x16 and the dmi; i might have a board using those
<gruetzkopf>
the pcie x16 port split is also supported by SNB/IVB mobile
<gruetzkopf>
which is what i'm interested in most
<felix_>
yep. but iirc not on all platforms. iirc the non x16 configurations can only be enabled when using a certain pch; not sure though if this is enforced by hardware or only by firmware
<gruetzkopf>
the schematics for the laptop i'm interested in abusing this with mention the strap directly
<gruetzkopf>
(some fascist dictator in the years around 1940 is the reason for ß being a thing but ẞ not really, his government wanted to do away with ß entirely)
<zkms>
can 阝 be a thing?
<TAL>
lets hope rqou that you didn't disarm gruetzkopfs head for hours now
<TAL>
reiterating everything what he had in school
<rqou>
oh wtf zkms
<rqou>
also we really need the kspacademia bot here
<jn__>
zkms: oooh, i see what you did there
<cpresser>
nerd-sniping with unicode glyphs
<pie__>
wait thats not the same as beta?
<pie__>
fuck
<pie__>
im pretty sure i used that for betas on my old keyboard in my latex stuff
<pie__>
or something
<pie__>
good thing it was only for university labs :p
<pie__>
actually i would have used \beta ....so idk
<qu1j0t3>
no, eszet is a distinct glyph from a different language :)
<pie__>
qu1j0t3, i mean, yeah that makes sense
<pie__>
i wonder how many keyboards actually use a beta? :D
<qu1j0t3>
probably not very many, Germans would quickly notice
<jn__>
pie__: in the printed labels?
<pie__>
jn__, no in the signal to the computer
<jn__>
pie__: well, then you need to first find a family of keyboard that doesn't send scan codes
<pie__>
ah.
<pie__>
yet another tool that never got pulished fml https://youtu.be/ZtSrcq7gscE?t=2325 , alternatively the location it dd get published in is impossible to find. did find some old github
<pie__>
-_-
<gruetzkopf>
i need to try if the dell vendor firmware explodes if i enable PEG bifurcation