THIS TRANSCRIPT IS AUTOMATICALLY GENERATED BY COMPUTER VOICE
RECOGNITION.  It is good enough to be worth posting, but not perfectly
accurate.  Do not cite it or take it too seriously; if people do, then
I will take it offline.

i
welcome to my scheduled stream for monday labor day
i am in this stream going to work on finalizing the leapfrog phi cf 
kits that i've been
that i've actually been kind of putting off for awhile because i've had 
other things come up that i've had to deal with
and then i'm probably also going to do some work on the deeper than the 
fat thirty two code for the
the usb host module that i've been working on that is another thing 
that is kind of behind schedule although we'll talk about some of the 
stuff that wedding went into that now i'm trying
inasmuch as today is likely to be a slow day with it being the holiday 
and so forth and so on i am trying some different settings and a 
different set up here with the the streaming arrangements in particular 
i've got another laptop computer involved here
and its running devil one linux which is not a
distribution i'm very familiar with but i i've been wanting to try dev 
one so why install that and i can't say i'm terribly thrilled with it 
in particular i had a hell of a time trying to get the touchpad to work 
and that's never been an issue for me with
laptops and other linux so
apparently by default it has the tap to click feature turned off and 
its not at all obvious how to turn that on and it still doesn't work 
perfectly but it works well enough that i'm able to run the
the twitch stream
what do they call it stream commander or whatever it is
ah and that way i can have my other laptop here free for use with the
the development tools for microcontrollers which i am
i'm going to have an easier time debugging i think if i don't also have 
to do stream control from that laptop i've commented on this before but 
the stream control with twitch is a hugely expensive in terms of cpu 
power and i don't understand why because you'd think they would 
understand the importance making that by
wait if you're going to also be running gaming and whatever else on 
your computer that you stream from
anyway are in other streams mostly and a little bit off stream i have 
put together bags of goodies that go into way leapfrog kit
and i also have an eye on tax some of these on on stream a little while 
back too i have sets of boards
now the new boards from india
is there a darker blue
and i debated you know i don't have exactly identical numbers of the 
older lighter blue ones so i was thinking well that's ok i'll use up my 
older ones first and then i'll just putting the darker blue these are 
almost perfectly as
and as as needed but then that would mean shipping some kits with 
boards that don't match you know like you get to board one here that 
slider blue
board three that's darker perfectly blue
and i thought that that would probably not be a good thing i think 
people would if their kid builders they'd probably like to have the 
boards all match so i'm thinking that i'll save the last few boards 
that won't match and use them in my hand assembled modules possibly 
especially for the accustomed panel
assembled modules were i don't think it'll show as much and you know if 
people are getting accustomed panel anyway they expect it to look a 
little funky and that should be fine
so that's the state of those boards and i have here the
i call it a happy user letter for complicated historical reasons but 
then this is the sheet of paper that goes into the kit and newly 
updated version with a qr code which will actually take you to the
a page on my website that has the assembly instructions although heaven 
help you if you're trying to read the pdf on and off
on a smartphone but you know people like to see qr codes now you've 
heard me complain at length about
the used
in bags that piling up i went through and i found all the ones that i 
thought were the right size to contain a kit and i've got five of them 
here which i figure i'll reuse for these now you know why i want to be 
environmental and so on and use a re reused bag
tags wherever possible and i hope that my customers will appreciate 
that however these bags are not quite as nice as the ones that i bought 
that were brand new
they often have little holes punched in them i think that's to prevent 
like pressure build up or something and they're out there a little 
wrinkly and they're thinner plastic but i decided i'm going to try 
using them for this batch of kits and see how people feel if i get 
complaints about the the reused bags then i may have to
only use
only use virgin fresh bags but i'd like to reuse bags wherever possible 
so and especially given that i have just a few that are really the 
right size for kit the ones that are the right size i think i 
definitely want to try to reuse
hello to a beaut and haas one three twelve underscore
i
i don't know how big a turnout we're going to get today it being
labor day when i was when i was in denmark the may first holiday came 
around
and some people asked me do you have may first in canada
and you know i was tempted to say well no we just go directly from last 
of april to may second or
what of course what they meant was is may first celebrated as a holiday 
in canada indeed is not
we have labor day in early september
and i think the reason is because of the associations of may first 
specifically with communism and the american influence here we can't 
have anything that looks like a show recognition of communism being in 
any way acceptable we even have a thanks to the conservative government 
we have a monument
to the victims of communism in ottawa
so celebrating the
the value of
of work and have people who do work
that has to be done on some other day and labor day holiday at least
in this era is not very political at all like on may first in denmark 
people want to go listen to speeches and here its just another day 
that is the marker for when in the school year will start
don't get me started on the fact that we're starting another year of 
school in the middle of a
a steadily growing wave of pandemic illness
and we still are not authorizing children to receive the vaccine so 
we're sending sending them to
to gather in groups indoors
without any protection
this whole adventure has really opened my eyes to i i always thought oh 
but think of the children was like the the
thought ending
line for any
any issue in north america but evidently sending children to school has 
to take priority over
over their safety
and so it looks like these are these banks as i thought were the right 
size and
the paper does fit nicely
still looks kind of wrinkly on there though i hope that's going to be 
okay i have really mixed feelings about especially given how much i 
charge people for these products about selling a product its in a bag 
that looks like this but
what can you do right
the the environmental concern of disposing of these bags is real
and its not going to be that many because i don't have many bags that 
i'm in a position to reuse that are this size can really be used this 
way
so i'm still going to be using a brand new bags most of the time
what we do what we can
this one does not fit perfectly
yeah
i think i'm going to make it work anyway though i'm going to cut a 
little chunk off the
in this paper like two millimeters will be enough
so that i feel it will fit

this is probably closer to one millimeter but that may still be enough


think i need another millimeter or two
maybe i'll take it off this side too i hand
some adventures where i
upgraded my
slackware installation on my desktop machine
and from that day on
i
my printer didn't want to print with correct margins
and after a lot of work i eventually determined that what was going on 
was that
it was attempting to rescale everything to fit on an a for page now in 
a four page
he used in places where they have the first of may is
its narrower and taller than the letter-sized standard page that i use
and so
the result was that it was actually shifting vertically on the the real 
page that i was printing on
i
in screwing up the margins that way
and i went through
now this is this is ocular the katie e
i
pdf reader
and customer support for
katie he is a joke
especially when it comes to printing
so there was not really much hope there
i ended up actually going through the source code and figuring out what 
was going on to some extent i never found out why he thought that my 
printer was configured for a four paper but i was able to find that it 
was running the lpr program passing it every time unconditionally no 
matter what options were
actually selected it was passing it a command line argument that told 
it okay rescale this page to a four
and so what i ended up doing was i actually replaced the lpr program 
with a perl script that i wrote myself
and
and it detects that option and removes it before pep calling the real 
lpr program
so that so that works you know i and now every time i i print something 
i have a little thrill of hey its actually printing correctly
i'm just up here looking for panels
gonna have to go outside the range of my lie share to what
get them let's see i need
here's three
and here's two more
yeah
so its kind of a thrill now everytime i print something to see that it 
does really print the size it should be
alright so panels check this off on the checklist
and i need your rep power cables do i have those already know
i

our cables have two kinds
well actually three
there's the sixteen pin kind and these there's only one of those right 
that there's not like a left-hand and right-hand version
but for the ten pin kind
which are used by this module
there's the left-hand and right-hand version
i
because you will basically always have the stripe down and the two ends 
are not the same so the ten pin end fits into the module on the sixteen 
pin end fits into the
bus board so i call this one a left hand cable because when you plug it 
into the module the cable goes off to the left
and i call this one the right-hand cable
because when you plug it into the module author never went into the 
module it goes off to the right
for the leapfrog in particular
it doesn't matter which one you use they they'll both fit on the board
but i
there are some other modules where it has to be one or the other so i 
have to keeper
a stock of both
and i think i'm going to use right hand cables for this batch of leap 
frogs
because i seem to have a few more of those or i did before i just
just grabbed five of them
okay
so check off power cable sticker sheets tarot cards
i've got the happy user damn it
i
something i really want for all mia installations of the the devon 
system they're just popped up at screensaver or after
much too short of a delay
something i really want is to just turn off the screensaver except 
still have it available for manual inculcation like sometimes i want to
locklear or freeze the screen and i want to have a button for doing that
but i don't ever really want it to happen just because of a delay
and i have yet to figure out how to do that because on a linux system 
there are so many different places screensavers can be configured
it doesn't seem possible to turn them all off and
and if you do still want to run the screensaver team and if you want to
be able to invoke it manually then it frequently does not have a do not 
pop up option now
i need five tarot cards here


we can be surprised and a few minutes when i turn those over and see 
what they are
and also five
five stickers
every time you buy secrets from these people sticker mule
they send you like a bar mail to a little coaster
and as i've been
as as its been summer
i've been drinking iced drinks of different kinds more often and using 
those coasters and they get soggy and eventually they get moldy so i 
had to throw them out i now completely out of coasters and i think i 
may actually have to buy or make some new coasters because i still have 
plenty of these stickers i know
not going to be due to put it another order with the mule for awhile 
okay so i have everything on my list here
and i can start putting it all together
oh
i do these in a fairly specific order
to try and to make the nicest possible package
so
the panel goes pretty much in front
and then i build up a little stack with the boards
and the terror occurred
i guess i should look at this
each of wands
and
the terror card and the stick are part of the purpose they serve is 
they going between these boards so that the boards won't rub against 
each other
because if they do rub they may scratch
and its pretty unlikely that they would scratch themselves badly 
enough to actually do any electrical damage but
again because of this as i do diy kit
the customers are going to be seeing those sports up close and
so i want them to look as nice as possible
alright next time guess i'll put in
these smaller bags

i have reusable or reused
plastic bags for the small ones as well which i have not yet decided 
whether i'll be attempting to put them in here in
in kid so on i guess i probably will
but they raise the same issues really that the
the used ones are a thinner
thinner plastic film than the ones i would buy if i were buying them


so they don't look quite as nice and they tend to be a bit scruffy
fit this in here
guess i should have put the cable in the first really
go

and the two
two cable ties
its going alright
how does this work
to da one leapfrog kit
and now i just have four more of them






so this whole stack of
boards and panel goes in the bottom there its the heaviest thing and 
that's where it would shift too
a hundred different motion of
things going in there you know these go through a fair bit of
a fair bit of abuse and shipping and storage and whatever else
so i tried to make a solid package here that long to
won't be a problem as it shifts around


yeah




i know what i don't think i even looked at that tarot card
i kind of would like to see it
yeah
or maybe not no i don't think i bought her
but we'll look at the next couple
let's see the empress
and the emperor
and the hero from okay
bunch of big names there
okay























i was the screensaver came up again
i've been streaming thirty minutes and basically not touching the 
computer so i think it must be on a fifteen minute
delay so we'll feel indulge me for a minute here i'm going to try to
find whether i can turn that off or something
maybe settings
display
i know that's just for setting like clock speed and so
settings desktop
as a background thing and menus and icons no

screensaver
as i've mentioned there's like fifteen different places

say fourteen forty
okay see there's an arbitrary limit here i cannot set it to
fourteen forty minutes the maximum it will accept is seven hundred 
twenty why there's no good reason
there are four for there to be any limit on that at all and and there's 
no option for just not having
a screen say an automatic pop up with a screensaver
i mean i could remove the screen server software from my system 
entirely but that wouldn't
that wouldn't be what i want i want to have it i just don't want it to 
happen automatically
i mean maybe this is
maybe i'm reading too much into it but it sure seems like there's been 
a trend toward less and less user control
nothing is allowed to be manual anymore it all has to be automatic this 
and now i'm really dating myself by saying this but this was my gripe 
with katie e for that
there used to be a button for manually hiding the the taskbar
and then there wasn't any more and they said oh well you can configure 
the automatic hiding no i don't want automatic hiding i want to have a 
button for manually hiding it all but you can invoke the automatic 
hiding by moving to the side to know i don't want to know
a symptomatic of the general issue with katie e of not listening to 
users but
it was
i guess the printer thing was worse but
the taskbar auto hiding thing
its pretty bad too
just zero comprehension of the idea that someone might want to be able 
to actually control it
it shouldn't be always an automatic thing
i mean really
something that is underrated is the importance of users being able to 
have a mental model of a
how a computerized system works
and we see that with google
i mean i don't know how old you are and if you remember but there was a 
time when google search results didn't suck there was a time when that 
was actually what google was good at
and what they were famous for and how they became
a major corporation
and that time was a long time ago
because the the search engine results have just gotten smarter and 
smarter its no longer just a
glorifying keyword search its
its now some survey i think that attempts to guess what question you 
are asking
and
there is no way to control you can't there's no way to know
what it will conclude from what you enter

and so its impossible to build a mental model any more like the 
original google was basically a keyword search
i think it would search for maximum number of keywords you entered
with the possibility to override it with necessary and forbidden 
conditions you could say this word must be done by adding a plus or 
this word must not be in by adding a minus
and then it would rank according to the
the value of a page in the i can vector of the principal eigenvalue of 
the adjacent see matrix
so perfectly straightforward right i mean that concept actually is a 
much simpler thing than it may sound
you don't really need to know what an eigenvector is its just that the 
idea that some sort of energy flows along the links and things that 
have a lot of incoming links from a lotta places are probably valuable 
webpages worth looking at
so i mean google once worked that way
and now there's no way of knowing how it works
and its secret
its easy to say oh well they had to change their algorithm because
because of spam
but they've done a whole lot of things that they didn't need to do 
because of spam and given that they're in the business of selling 
advertising you know any
anything that they say about the scourge of spam has to be taken with a 
grain of salt
and the result is that now you can't find what you wanted in google 
anymore especially not if what you want is something that doesn't make 
them money
and there's no replacement there's no other new search engine that 
actually works that is available because everybody
he's in it to make money
and if they think they want to compete with google then they're going 
to try to do the same things that they see google doing
just try to somehow do better at making money
we never should have allowed commercial use of the internet



i mean look at twitter look at twitch
case closed basically
alright so that is five
leapfrog kits
i still have some concerns about these bags that look kind of wrinkly
yeah well
that's the way it goes
just see if i when i sell them if i get any complaints
i
so i'm going to move these to this table over here
i don't really want to dig through the
the bin that they're stored in which is ana
high shelf back there
and then we're going to have the interesting adventure of me trying to 
hook up my other laptop here with screen capture
and we'll see how that works part of the issue with me not wanting to 
do software to debug on stream
is that i end up having to run you know multiple layers of
ah
the the the microchip development tools and
until i had the separate laptop for it also
twitch is a stream controller
and screen capture now i don't actually have to do the real the real 
serious and coding company the laptop that is on a desktop machine or 
out of out of view here
but it needs to at least be able to
screen capture itself
and ship that over the network
to the bigger computer
and screen capturing is pretty expensive
and the microchip tools are very happy
so its rather a tight fit
let's see though let's see how this works

and i can't like do this development on the big desktop machine
partly because i don't want to trust those microchip tools to run on 
the big desktop machine they have a boatload of telemetry and so on and 
also because i need to be at my electronics workbench with the power 
cable and power supply and so on in order to run the
in order to run the usb host that i'm going to be
working on
so plug this in
boot it up and
did i put that
this thing
i'm going to need a network cable as well
and i probably should hook that up through the
through the usb hub


just change views here
okay
call me paranoid but i don't really want to have my
keyboard on screen while typing passwords

okay so
what if i got here can i get away without having a web browser
probably better keep the web browser
and now i have to
find the command for streaming
ah
streaming this laptops
whew
its just done sh alright
okay so i'm gonna i don't know
if my big computer will
oh actually accept this but let's try
can i just run the fm pig command
yeah alright so the
the laptop is now
sending its screen
two to the desktop
and
let's see if
desktop can pick it up
ooh that looks good for me
you can watch ff mpeg produces so many error messages
i
you know there's a strong element of black art in getting this to work
alright so
with that working
i
now i've got here various get goodies and gizmos let me give you the
camera again
can play a little show until
i
okay so
here's the the laptop
and its got the
el cheapo usb hub with the
usb ethernet
through which its streaming i mean there are so many variables here i 
didn't want to also be trying to do this wirelessly
also on that hub is the microchip pic kit for which is four up
controlling and debugging
the microcontroller chips that i'm working on
and then i also have this i call it the
my first easy bake usb protocol analyzer because its its the cheapest 
usb protocol analyzer that money can buy and and i got it used
so just plug that in here
because i want to run the usb connection between my
my device and the
in this case usb usb flash drives through that so that i can capture 
the traffic as they talk to each other
alright so
i mean that may not even really be necessary i mean not its not the 
low level stuff i'm worried about right now but it'll probably help to 
have it okay so now i've also got this thing which is the
the actual machine that i'm working on
this is an early prototype of the year
msk or fourteen gracious host
oh
and basically its just a thin wrapper around this microchip
twenty eight pin microcontroller which you know its its a sixteen bit 
computer with however much ram eight k of ram and sixty four k of 
programmable rom programmable through this thing
and its got a whole bunch of built-in peripherals so its a nice chip 
to program for unfortunately i ordered a mauser said they had forty in 
stock and i ordered all forty and they delivered thirty two and 
microchip the manufacturer
says that they will have more stock
the last several times i've checked its always been one year minus one 
day from the current date so on september second they said they'd have 
it september first of next year and september third they said they'd 
have it september second of next year and so on so i don't know if i 
will ever be able to buy more of these like
controller chips and if i can't then i'd better sell all the ones that 
i have because otherwise i won't even make
make back the development costs as this has taken a long time to 
develop and a heck of a lot of programming on the firmware
and the firmware is not even finished yet
although its getting pretty close basically the the missing piece is i 
needed to be able to read files off a usb key and use them to reflash 
its own firmware so that someone will be able to reflash the firmware 
in the field without using one of these machines
and i have all the parts for that except i think there are some 
incorrect calculations in the fat thirty two file system code that 
cause it to not read the correct
sector when its following the chain of links through the fat thirty 
two now i mean that's you know a fairly specific and esa tierra kind of 
thing but it does need to to actually be right or else i can't ship 
this song
so what i'm looking at doing today and i don't know how well this will 
work because it was kind of a last minute thing
i've been working on other things i shelved this for awhile and then 
they all while i have to have some stream content so
what i'm going to try to do is find the find the mistake
so let's go back to that screen capture now we're looking at the the 
screen of the laptop
you can see my pretty desktop picture there
i
and okay i guess i'm gonna need to boot up the power supply
that's off screen there
let's let's open mp lab

just update from subversion most of these recent changes have been on 
the documentation i haven't touched the code in a little while


and it'll start that
that's gonna take a while to start though not quite as bad when i'm not 
also running the
the fucking stream manager
so well that's working
i have to get the voltage right on my power supply
my power supply is a cheap chinese one
and it has single single turn pots for
controlling the voltage and its also a little bit temperature 
sensitive i think
so i guess i better plug this one in first
incidentally if you saw me
unpack the box of
the box of circuit boards from india they sent me this key chain
which is a little ls usb light so if i plug that into the usb port on 
here it lights up whoop de do
not sure how useful it is to have a light that you know is
just plugs into a usb port them has to be plugged directly into the 
port but that's what that is
alright so i can plug that in and i guess i
this will be next
i can't remember which end is which on this thing
yeah
and i
the the my first easy bake
protocol analyzer its got to this uplink port that goes to the actual 
computer that will be analyzing stuff but then it also is designed to 
splicing between host and device solve this little cable goes to my my 
machine which is the host and then i will be plugging devices into her
here and the analyzer in theory doesn't interfere with the the traffic 
on the bus it just records it something i could view who said what on 
the bus
and learn whether there's something wrong with my drivers
now you can get a usb analyzer that is just a
just a piece of software that runs on your pc or whatever that's fine 
if you're debugging connections between that pc and the device
but it doesn't work very well when
when you are actually debugging the host
okay so now
this thing is actually still not fully booted up it says its 
background scanning of projects
but its basically started and you know watching a minute or two would 
also pop up a stupid message about wanting to be upgraded
i
no i have a bunch of
usb usb memory sticks that have been
loaded with different things
and i'm going to
i guess swap camera again
you know
this is this one has a fat thirty two file system on it with a file 
that my firmware ought to read
but i don't think it really will
so i'm going to plug it in here
and then we can step through the code and see what happens


i have got my own makefile system
and its only an uneasy fit with the mp lab mp lab as the microchip 
yeah here we go pack updates available there it seems to be no way to 
turn that off short of going to extremes like firewalling the software 
to prevent it talking to the net at all
i
the mp lab x id he really wants to generate your makefiles for you and 
that's not a good plan because my makefiles is a whole lot of other 
things
so i had to go to some lengths to create sort of a stub project file 
that would allow it to run my makefile and not mess with it
and in the near future i'm going to have to write up how to do that to 
go in the documentation of my module
which i'm not looking forward to but it does work so that's fine
i
you know having having run that
yeah i should have said build for debugging i think i did

now i'm going to program the device
so this is just sending that to
firmware that i just compiled through the year
the link to the the chip


alright its programmed
so now i should be able to watch the debugger

and run it
in a second
yeah ok
so if i run that
see what happens

now it says its running but
its not blinking its a little light switch
is a bit of a surprise
the no lights here
so let's ah
interrupted and see where we're at
okay so it is
hanging up waiting for a bite to be read



that's interesting it wasn't something i expected
so let's see
i think
its maybe a bug that i've already seen somewhere once before
right
send the right command okay so the thing is that with its talking from 
the the cpu to the to the sram chip using an interface called s p i
and the thing is with spr you have to read and write simultaneously
so what you end up doing is you queue up the bike you want to write and 
then you ask to read a bite and then that will have the effect of 
writing your bite as well and there's a buffer so you can queue up 
several bytes in advance
and then
and then do the corresponding reads
and if you actually did want bright usually the read will be ignored 
it'll be garbage data so you just do the reads to make it iterate 
through its buffer
so i did have an issue at one point where i was reading one less bite 
than i was writing so that would mean that the buffer would slowly fill
i guess its also possible that this is not an error and its just that 
because it was reading badly from the fat thirty two file system its 
got more data than it thinks it should have
and so
its just taking a long time i mean its having to do all kinds of 
skenazy and usb transactions in the background here
so actually that may be more likely so let's reset this
and get me another terminal
we're going to actually use the the beagle

i'm starting up the
the analysis software for them my first easy bake
usb analyzer here
and
now this thing is
a little flaky
so it may just give me a bunch of errors
but i've told her to start capturing
so now i go back to mp lab
reset and run

yeah its just giving a bunch of errors the thing is this thing is it 
doesn't seem to have a clear idea of whether it wants to be talking to 
them
in usb high speed or low speed
or or full speed to usb has four speeds it has high
i know it has low speed full speed high speed and superspeed
now my hardware can only actually do
low speed and full speed
but it appears that the the analyzer doesn't have it sometimes
defaults to the wrong one
so let's clear that out
i'll try unplugging this and unplugging that
and re plugging both
actually i should start the capture before i plug it in
cause that seems to have it make it work okay devices can close 
communication error that's because i unplugged it
so i'm going to have to restart this software
which fortunately runs relatively fast its not like m p lab witches
a real pain if you have to stop and restart it
okay
now i start this
plucking this
okay now with a little luck at me
it may capture at the correct speed
right
alright
no that didn't help
so stop
start
oh there we go
there are some real transactions going by
and you can see its a whole lot of them
so i think this is just symptomatic of the trouble with the the fount 
driver because there are some large files on the usb key and i think 
what's happening is that because it miscalculated where
i like it it goes through the
the directory structure finds the the proper cluster number
and then it has to find i'm just thinking out loud here and then it has 
to find the right sector for that or the right entry in the
in the fat
and then it follows the linked list through and i think what's 
happening is its somehow getting into the wrong file so its ending up 
in the middle of some other file that happens to be huge
and so its reading through all of that file
and dumping it into the sram chip
which won't hurt the sram chip but it its just the wrong data
so
i think maybe what i need to do
his prepare better
i
a better usb key because this one has all kinds of weird crap on it it 
was a a key that was used as like the extra storage during a windows 
upgrade and its actually a very weird fan file system on there it has 
a whole bunch of sort of forbidden data
so i think the thing to do is create a smaller usb file system
and then i can
step through that and see if if this code i currently have ever works 
and if it does at some point work
then i won't be able to you know sorta to would diff

its interesting that this is still running


i guess just the fact of having an i mean if its miscalculating how to 
follow the links that may well have gotten into a loop anyway

so let's pause this
and
stop this clearer
we can leave them running i think
i
alright i'm going to need to
another
oh
another usb key

and i've got several here i'm not sure which ones i can safely override 
so i'm going to have to take a look at what's on them

just plugging another usb key into the hub here



okay all that's on that one
is
this bmp directory which is from using it for ourself on the scope 
scope shots
so that's one i can safely put things on and its probably a fairly 
well behaved
file system
so let's let's do that
firmware data from ram
if i remember correctly its one of the files that
that this code is supposed to be looking for



yeah from where
this may not show well on the stream but
i've got a text constant here which is the filename that is searching 
for witches from where dot frm

so
i just copied that to another usb key
and we'll see if
if the gracious hosts can recognize it
solid that's been copied and unmanaged don't unplug that
pluck this one
this one in

i basically trust the code up to this point
so let's
let's break at the point where it actually recognizes the the filing
that is here
if i run up to this point

yeah really
i should have had the capture running and i should have had the camera 
on
so let's reset that
and turn on the capture
and i'll just push us back on the camera briefly


now if i run into that point you should see the lights blink on the 
host as it goes through usb enumeration and so on
alright
now onto the laptop
so at this point in the code it has recognized the file name i wanted
and it just sends a reset command to the sram to make sure that
that is in a known state
now
get the i know that reading the first cluster works i've had that 
succeed before
so w zero
its to cluster
and i ordered the clusters zero
and then we check is this the fat thirty two i believe it will be


what its trying to do is read the file and write it into this sram chip







so its going to try to get the first block of the file
working register true here
contains values zero nine eight two which is a
pointer to a structure that describes
a cluster number

in that contains the data that we that we wrote its a byte zero or i 
guess sector zero within cluster zero four fifty
and its more serious after that for higher bites
and then when it calls we'll get kava
it may be that i took too long
to do that so its up
the machine has gone into an error stage here
both let's try
resetting that
and running jew after that point

if i tell it to run here
actually let's look at the capture as well
it just says singer
its not critical right now



alright
we ran up to here yeah get kava has succeeded
and w for now contains a pointer to the buffer that it just filled
just address twelve forty six




and if we look here the first bites of that r m s k space on all four 
space
from where copyrights okay so it has correctly loaded the first the 
first block of the file
because its those those bites are the ones that occur at the start of 
that file
then it goes through
sets up a write command then its going to send all the bytes of that 
block to the sram
and then there's this loop test
and the spr transaction
etc
so
all that looks good up to this point
increment camera
is a a function that then points that that address at the next block 
and that is inside theories where it actually hits the the file 
allocation table and finds the next cluster number and so on
so
i guess
gonna have to run up to that point




the first layer of this is just that its
goes to the next block within the same cluster

arab again
part of the fun with usb is that if you don't send a command or don't 
do anything for a while the device will decide that the host is dead 
and just reset itself
so you you have to keep things moving through through the usb bus

let's try this again with a reset




going to set a breakpoint here
the thing is i wanted to be able to catch it when its actually
going to the next cluster in a year
i saw a file so iterating through the
the sectors in the first cluster is not very very useful because i know 
that part works




okay so now i've run
through to the end here
we're getting the correct cluster
there's a fair bit of multi precision arithmetic that has to be done 
here because of this is found thirty two at least potentially and the 
its only a sixteen bit microcontroller
so we check is this fell twelve
don't think so
sep sixteen or thirty two
so we're gonna take the
cluster member the design registers w four in w five
that's all for fifty may recall
what just happened there
oh yeah its because if you hover i hate when that happens
things should not pop up just because you hover the mouse but that was 
what happened
okay so anyway
double that multiply by two shift left and rotate left
and then test is this really fat sixteen or thirty two
its thirty two so then we have to multiply by two again we're taking 
the
we're trying to find a byte offset inside the the fat so fat entry if 
its three two bits
its four bytes


so at this point
w three w two
its become one one forty which is twice for fifty in hexadecimal or 
four times four fifty i guess
fat block size is two hundred texts that's five twelve bytes
and then let's skip over that deficient instruction
one of the funny things about this microcontrollers that if you want to
do division there's a division instruction but you have to run it 
eighteen times
that's how they implement the somewhat complicated binary long division 
algorithm



so
w five to ducking for
contains zero zero two latex which is the block index
of
the the corresponding sector of the file application table
epithet is going to fail yup
alright i'm gonna have to do this with the break points again





so i set a breakpoint after that call it failed and i'm gonna just run 
full speed until just after it because the issue is that if i 
interrupted with the with the debugger
then
it won't come back when i restart




so at this point it has read
the sector of the
the sector that contains the relevant part of the file allocation table 
so its going to have to look in that table find the entry and it 
currently has a pointer to it
and then it gets the cluster number from there







you know its going to check for whether that question number is a 
special value sent indicating for instance the end of the table
there's the number that was in there the first cluster of this file is 
cluster number four fifty and after going through all that work neural 
and reading the the fount
it now thinks the next clusters for fifty one which sounds very 
plausible






the ok
the if i do this its probably if i go to the next one its probably 
going to crash again
yeah
but that means that it is successfully going through
and read the the second cluster of the file
no
let's see if we can figure out whether it actually read the file 
correctly
the second question correctly
i can do that by ah
putting on a fancy break points here
i will make it break not on the first time around but on the second
or actually
its going to have to be even fancier

i want to break here only after it has hit the other break point


this breakpoint point here is one i didn't set up this session but its 
if there's a memory alignment error it should break


sears line five ninety something

i want to set up a complex break point
and two new sequence
i don't know why it insists on having a name for the sequence


move to fool
okay so i think what this means
is that if it hits line seven or three which is inside that
process for loading the next job
cluster if it does if it goes to the next cluster and then after 
attacker has read the first block
it'll break
i may have gotten those in the wrong order but we can deal with that
okay
yeah i did get those in the wrong order
so let's try and changing that

move up


alright so at this point it should have gotten the first block the first
block the first sector let's say
of the second cluster
and w four
has the value two for eighty
and that should be a pointer to where the block is stored in memory

so if i go there the
i can read out the bytes six four two eight e f six b
now we'll look at the young
we'll look at that from our data from file



this is the the assembler generated listing
and it contains hex dump
six four two eight e f

maybe i really want to look at image target list too
six four two eight iaf

six four two eight e f six b four eight
six for trade and six b four c e f six p four h e one o two o all

iwan o'toole so that's definitely where we're at
see if we can find the address
this is a zero to six zero
and somewhere below that
next addresses thirteen something
it looks like its about halfway between the two
so this is probably address one thousand we can check that more 
carefully in a minute
ah
and address one thousand to be four kilobytes
right
sixteen times geophysics yeah its plausible that that too
the file system has a cluster size of four kilobytes so i think at this 
point it is correctly going onto the next
cluster
and i think the problem with my other one is that when the cluster 
number is much larger so it doesn't fit in sixteen bits
there's something weird about the carry between the low and the high 
words
so that is
yeah
i
see if i remember
the syntax for the odi command

not quite right

i wanted it to be single bytes
tc
next
there we go
oh not quite





come on i want i want to see hex for single bytes









there we go
i


alright here we go here is
address one thousand
and its a six four to a gf six b four eight that's the same thing 
we're seeing here
six forward to wait e f sixty four eight
so i think what we've established here is that it is correctly reading 
the second cluster of the file and it did not do that on the other
on the other usb key
by way of verification
let's try swapping the keys
and running again

now here we are at this point where it has read what it thinks is the 
second cluster of the file
and register for four points at twelve forty six

two
and its totally different data its fifty thirty nine b six
which might not necessarily be an issue if the cluster size is different
but its not found in this dump at all
so with the one
file system that's fairly well behaved and has the file near the start 
of the disk its correctly reading the second cluster of the file
and with the other file system that's weirdly organized and has the 
file pretty deep into the disk because its a large large usb key
its not correctly reading the file
now i'm not sure that i'm going to be able to get much further
ah with that today
i


just trying to leading all
and i think i may have just crashed mp lab

but i am glad that i was able to get back into touching that at least a 
little because its something that i've been sort of overdue on

i think it being a holiday monday and that i don't really want to
try to keep poking at that right now that i'm going to probably call 
the stream at this point
unless there are any critical questions that people want me to answer
and i will say
i hope you enjoy
and thanks for watching and have a nice day
yeah
yeah the
the
the
the