promach has joined ##openfpga
pie_ has joined ##openfpga
pie_ has quit [Quit: Leaving]
m_w has quit [Quit: Leaving]
m_w has joined ##openfpga
talsit has joined ##openfpga
digshadow has quit [Quit: Leaving.]
<rqou> hmm, somehow i needed to do some "percussive maintenance" on my chipwhisperer board
<azonenberg> Lol oh?
<rqou> something happened while breaking the break-off part i guess?
<azonenberg> Lol
<azonenberg> oh joy
<rqou> seems to work reliably for now
<rqou> might be the sma connector actually
amclain has quit [Quit: Leaving]
<rqou> ok, seems to work :P
<rqou> lain: intel ark just got completely redone
<azonenberg> lol
<azonenberg> wut?
<rqou> the website got a makeover
<azonenberg> that looks like it belongs on an ipad
<davidc___> heh; when they were working on it, they dropped a bunch of their datasheets
<davidc___> Turns out when you email them from the work domain of a large enough company, they actually respond, and respond promptly, with apologies.
<azonenberg> lol
<azonenberg> ooh big news
<azonenberg> For those of you who are afraid of tiny QFNs
<azonenberg> Silego just pushed out a new datasheet rev for the SLG46620
<azonenberg> With a TSSOP-20 package option
<lain> davidc___: and if it's a small company, they ignore it :D
<rqou> hey, just like brcm :P
kiboneu has quit [Ping timeout: 252 seconds]
scrts has quit [Ping timeout: 252 seconds]
kiboneu has joined ##openfpga
<rqou> lain: so i'm almost certain the pentium g4400 has ecc
<rqou> going to test the g4560 now
<rqou> on the g4400 using the public datasheet:
<rqou> dev 0:0.0 config register CAPID0 has the ECCDIS bit clear indicating ecc capable
<rqou> and MCHBAR+0x5004 has 11 in bits 12-13
<rqou> which are labeled "ECC" in the diagram but "Reserved" in the table :P
scrts has joined ##openfpga
<lain> lol
<rqou> idk if you can "accidentally confirm" anything about these bits :P
<rqou> also, nice inconsistencies intel
<rqou> yup, putting in my stick of non-ecc ram and those bits clear
<lain> chances are I don't have access to anything more than you do wrt ecc :P
<rqou> wait really?
<rqou> wtf intel
<lain> lemme put it this way
<lain> a friend works at intel, and he says most of his dept gets their info on what the latest stuff is /from wikipedia/
<rqou> what
<lain> yes
<rqou> i thought only apple worked that way
<lain> intel is terrible at documentation
<lain> like legitimately terrible
<rqou> even when i was at broadcom i had access to "everything"
<lain> and heavily silo'd
<lain> if you're a firmware dev at intel, you can't talk to the hw or sw devs your code interacts with
<rqou> broadcom is silo'd too
<lain> you have to go through layers of management
<rqou> and i could see "all" datasheets
<lain> it's like a fucked up corporate game of telephone
<rqou> wow even broadcom was better than that
<lain> yeah it's greaaaaaaaaaaaaat
<rqou> is this why intel can never fix cross-cutting platform bugs?
<lain> lol, likely
<rqou> hmm so the kabylake mchbar+0x5004 also sets the "ecc" bits
<rqou> it seems it does have ecc
<rqou> so wtf intel
<rqou> you f*cked up ark and caused yourselves a bunch of negative PR
<rqou> lain: do you think intel PR will realize they need to ask for retractions of articles like http://techreport.com/news/31267/kaby-lake-pentiums-gain-hyper-threading-and-lose-ecc-support ?
<lain> lol
<lain> they're too busy putting out the C2000 fire
<rqou> oh right
<rqou> that
<davidc___> rqou: no brcm in our stuff - they won't even talk to $WORK unless we want 1mil pcs
<azonenberg> lol yeah
<azonenberg> i dont use broadcom for anything
<azonenberg> or qualcomm or any of Those Places
<rqou> lol my sister just hit a "utf-8 BOM" problem
<rqou> thanks windows
<davidc___> (or at least, thats what one team told me)
<rqou> "somebody set us up the (UTF-8) BOM"
<rqou> (joke stolen from tdwtf)
<mtp> YOU HAVE NO CHANCE TO SURVIVE MAKE YOUR [clock emoji]
<rqou> emoji? are you _trying_ to break it even more? :P
<mtp> yes
<lain> lol
<lain> felix_: I like the chromanova dance stream, thanks :3 although tonight it seems to keep momentarily dropping, dunno if that's me, them, or something in between :P
<davidc___> 💩
<davidc___> some day the world will suport UTF-8
digshadow has joined ##openfpga
digshadow1 has joined ##openfpga
digshadow1 has quit [Client Quit]
digshadow1 has joined ##openfpga
digshadow has quit [Ping timeout: 264 seconds]
digshadow1 has quit [Client Quit]
<lain> rqou: so have you confirmed it does indeed support ecc?
<rqou> as far as i can tell, yes
<rqou> i don't know how to confirm for sure
<lain> neat
<rqou> but reading MCHBAR+0x5004 gives "0x3111" with the "ECC"/"Reserved" bits both set
<rqou> and yes, the bits get unset when using a non-ecc dimm
<rqou> OH SHIT I MISSED MTVRE
<rqou> google magic inbox ate my email
<fpgacraft2> <nmesisgeek> hey at least it didn't mic-drop it :p
<rqou> i thought it would be next week
<rqou> because it's usually on a day of the month with two digits :P
<rqou> bad heuristic
<lain> lol
<lain> google inbox seems like such a bad plan :P
<fpgacraft2> <nmesisgeek> /me uses thunderbird and IMAP no matter what the back end is :p
<lain> but I'm biased, google still can't figure out that when I say "do not send to spam" I mean it: https://twitter.com/lainpants/status/606838729475387392
<rqou> I wasn't even referring to the "Inbox" product
<lain> oh
<lain> lol
<rqou> but the magic Social/Promotion thing
<lain> ahhh I turned that shit off
<lain> lol one time google started sending customer tickets to spam because they contained "nonsense" (read: debug output and crash dumps)
<lain> had to disable the spam filter on the ticket account
<lain> it gets SO MUCH spam
<lain> 90% of it is in chinese
<lain> I have no idea what it says
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
<openfpga-github> [yosys] azonenberg pushed 34 new commits to master: https://git.io/vDuk2
<openfpga-github> yosys/master 87fe8ab Clifford Wolf: Merge pull request #293 from thoughtpolice/minor-cleanup...
<openfpga-github> yosys/master 6781543 Austin Seipp: passes/hierarchy: delete some dead code...
<openfpga-github> yosys/master 943389c C-Elegans: Fix issue #269, optimize signed compare with 0...
cr1901_modern1 has joined ##openfpga
cr1901_modern has quit [Ping timeout: 240 seconds]
digshadow has joined ##openfpga
<rqou> lain: if i try to force-enable the igpu by setting the DEVEN.D2EN bit, what is the probability that my hardware explodes?
<rqou> greater than 0?
<azonenberg> The probability of a meteor taking out your CPU at that instant is nonzero as well
<azonenberg> jfyi
<rqou> lol
<rqou> the intel datasheet says that vcc_gt minimum is not zero
<rqou> but i've been learning how awesome intel's datasheets are
<rqou> so idk if that is totally bogus or not
<azonenberg> lol
<azonenberg> i dont trust any datasheet these days
<azonenberg> literally every large digital IC i've worked with i've found datasheet errors
<azonenberg> (mostly FPGAs)
<azonenberg> ranging from typos to blatantly incorrect information on register values
<lain> hmm
<lain> rqou: I literally have no idea, but you should totes try it and tell us what happens :V
<azonenberg> Set up a camera first
<azonenberg> Just in case
<lain> ^
<rqou> yeah how about no :P
<lain> I wouldn't expect it to damage anything
<lain> but there could be some weird latchup effect, I dunno
<rqou> hmm i checked again and vcc_gt can go to 0
<rqou> when "idle"
<rqou> so this mobo probably doesn't supply it
<rqou> blargh, wasting the gpu transcode capability :(
<rqou> also, the xeon e3-1200v5 datasheet doesn't document the mchbar+0x5004 ecc bits either
<rqou> as in, it's also only present in the diagram and not the table
<rqou> is intel hiding it for some reason? why?
scrts has quit [Ping timeout: 256 seconds]
<lain> intel is incompetent.
<lain> also the datasheet is a boiled-down version of a NDA doc
<lain> it's probably not even checked for accuracy :V
<rqou> heh
<rqou> i know broadcom datasheets don't have very good accuracy
<lain> well
<lain> not that their NDA docs are checked for accuracy
<rqou> and they're not even public :P
<lain> I've asked for clarifications numerous times
<lain> it always gets blackholed
<lain> my engineer contact is usually eager to help
<rqou> (at least in one particular silo) broadcom datasheets are a boiled-down version of a Perl data structure
<lain> then can't get a single response from the management chain
<lain> because of course it has to go through 15 layers of management to get to the person who can answer it
<rqou> hmm at least broadcom didn't seem to silo _getting_ information very much
<rqou> fixing things however...
<rqou> hmm from a very informal test, the hottest components in the nas i'm putting together are:
<rqou> 1) the crappy nvidia g92 gpu that is being used for testing 2) the 10gbe card 3) the cpu :P
<rqou> it's just like (one particular version of) the raspberry pi where the hottest components were iirc 1) the usb+ethernet phy 2) the pmic 3) the dram 4) the soc :P
<rqou> also the stock intel cooler is *loud*
<rqou> lain: i wonder what it's like to be a mechanical engineer at intel?
<rqou> they have to have at least one
<rqou> just imagine how frustrating his/her job must be :P
<lain> lol
<lain> yeahhh
<lain> woo
<lain> now I can linux while I windows and windows while I linux
<lain> the future is weird :3
<rqou> wow
<rqou> how are you enjoying your GNU/NT operating system?
<rqou> :P
<lain> hahaha
<lain> I'm wanting to test some tools that only exist in linuxland
<lain> as part of our Visual Studio build flow
<lain> sooo now I can just... invoke the linux shit directly :D
<rqou> afaik most of the "real magic" you can do with linux doesn't work on wsl though
<lain> depends, it's improving rapidly
<rqou> can i use the ioperm syscall? :P
<lain> wuzzat :P
<rqou> it gives you access to x86 i/o ports
<rqou> from userspace
<lain> sounds scary!
<rqou> but restricted to root
<rqou> or "CAP_SYS_RAWIO" but nobody actually uses capabilities :P
<lain> looool "“[WSL] doesn’t include X Windows or any other graphical subsystem.” Oh, why did Microsoft have to go and make that a challenge?"
<rqou> but you can use ioperm and then poke io port 0x378 to your heart's content :P
<lain> 0x378, isn't that parport?
<rqou> yeah
<lain> haha
<lain> can't believe I remembered that :>
<lain> I used to do parport stuff in QBasic lul
<lain> and later in C
<rqou> also, i bet you can't call arch_prctl or modify_ldt on WSL :P
<lain> lol
<rqou> which are needed by various emulators
<rqou> e.g. wine needs arch_prctl
<lain> yeah GUI isn't supported
<lain> apparently you can hack it to mostly work but it's buggy and not advised :P
<lain> (lol wine in WSL)
<lain> windows on linux on windows
<rqou> and some xbox original emulator used an undocumented windows syscall to effectively perform modify_ldt
<rqou> but this means that it can never work on win64
<rqou> not even wow64
<lain> wow, such bit, many OS
<rqou> linux has lots of "interesting" footguns and "features"
<rqou> e.g. you can issue a 64-bit syscall from a 16-bit protected mode code segment
<rqou> in case you didn't see this coming, a security vulnerability resulted from this :P
<rqou> look up espfix/espfix64
<rqou> (btw windows also has a "espfix" but it's not as good as the linux implementation)
<lain> -.-
<lain> windows doesn't even run 16bit code anymore
<lain> it hasn't since like win7
<lain> or was it vista
<rqou> i thought the 32-bit builds still did?
<lain> do 32bit builds still exist?
<lain> does anyone care?
<lain> :P
<rqou> nah, might as well run winxp in that case :P :P
Bike has quit [Quit: assault]
<rqou> anyways, here's the windows "inferior espfix" https://blogs.msdn.microsoft.com/oldnewthing/20160404-00/?p=93261
<rqou> the linux implementation is better because it actually restores the proper upper 16 bits of esp rather than just making it zero
Hootch has joined ##openfpga
kuldeep has quit [Remote host closed the connection]
pie_ has joined ##openfpga
kuldeep has joined ##openfpga
scrts has joined ##openfpga
pie_ has quit [Ping timeout: 256 seconds]
pie_ has joined ##openfpga
<felix_> lain: yay. can't remember having dropouts in the chromanova dance stream though (using vlc)
<pie_> finally done with exams \o/
<qu1j0t3> pie_: congrats
<lain> felix_: also vlc here, but I'm in the US, pretty far from the origin, so maybe just congestion along the way or something
<felix_> pie_: congrats
* felix_ is quite happy that he finished his masters degree last year; no more exams \o/
<pie_> lol
<felix_> lain: probably that
<rvense> i miss university
<lain> >checking whether vsnprintf works... probably
<lain> way to boost my confidence, build script
* lain sleeps while this hopefully builds
<lain> doing perverse things.. trying to build ghdl with gcc backend on WSL (windows subsystem for linux), after unofficially upgrading WSL from 14.04 to 16.04
<jn__> i'm sure someone's running WSL with gentoo already :)
<lain> failing that, I'll just wait for the ghdl-llvm CI builds to start appearing thanks to this person recently setting that up for the ghdl folks
<lain> jn__: lool
<pie_> what does wsl run ubuntu
<lain> pie_: yep
<lain> it's ubuntu userland, but the kernel stuff is agnostic, just presents a linux kernel api surface, like a reverse-wine
<pie_> running inux on window,s what has the world come to, whats going to happen to FOSS market share D:
<lain> haha
<pie_> microsoft infiltrates, then kills
* pie_ fearmongers
<lain> canonical found to be complicit!
<lain> I can see the headlines now ;)
<pie_> its already happening twith systemd
<pie_> readhat is in bed with thme
<pie_> them
<pie_> redhat
<pie_> cant type
<felix_> well, being a student was awesome, but having to learn for and then write the exams was still meh
<pie_> hehe
<felix_> but yeah, i needed 4 semesters more than the standard period of study for my bachelor and master degree together, since i worked on a lot of side projects. i'd say that i learned more stuff in those side projects than in my studies at the university though...
<felix_> university was still worth the time i spent there
<pie_> i have a lot of side projects it would be nicec to work on but i dont feel like graduating later...idk
<qu1j0t3> (later in life) I have a lot of side projects it would be nice to work on but I need to pay the rent
dingbat has joined ##openfpga
pie__ has joined ##openfpga
pie_ has quit [Ping timeout: 255 seconds]
<rah> heh, true dat
<rah> except.. how much work do you really need to do to pay the rent? as much as you are doing now? ...
<qu1j0t3> pie_ | readhat is in bed with thme // It employs a good part of its developers.
<qu1j0t3> pie__: So don't expect any relief from that quarter :)
<pie__> microsoft employs redhat?
<pie__> TIL
<qu1j0t3> pie__: Oh i thought you were talking abotu systemd. nm
<pie__> kek
rvense has quit [Quit: leaving]
amclain has joined ##openfpga
digshadow has quit [Quit: Leaving.]
pie__ has quit [Changing host]
pie__ has joined ##openfpga
digshadow has joined ##openfpga
<davidc___> pie__: TBH, it'll probably be great for linux market share. Who wants to write for native windows APIs, when you can work with the linux subsystem
eduardo__ has joined ##openfpga
<davidc___> pie__: then its an easy jump to "wait why am I paying for this kernel again"
<pie__> haha
<pie__> well if you put it that way
eduardo_ has quit [Ping timeout: 252 seconds]
<davidc___> In all seriousness, I'm pretty excited about it, since I prefer to use linux for all things; but some of the SW I write, people want to run on windows
Bike has joined ##openfpga
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
promach has quit [Ping timeout: 240 seconds]
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
m_w has quit [Quit: leaving]
m_w has joined ##openfpga
m_w_ has joined ##openfpga
<azonenberg> lol
dx has quit [Changing host]
dx has joined ##openfpga
mifune has joined ##openfpga
mifune has joined ##openfpga
Hootch has quit [Quit: Leaving]
<azonenberg> Any southern cali folks here near LA?
<azonenberg> I'll probably be in the area next week for work
mifune has quit [Ping timeout: 260 seconds]
mifune has joined ##openfpga
mifune has joined ##openfpga
digshadow has quit [Quit: Leaving.]
mifune has quit [Read error: Connection reset by peer]
mifune has joined ##openfpga
mifune has joined ##openfpga
balrog has quit [Ping timeout: 276 seconds]
digshadow has joined ##openfpga
balrog has joined ##openfpga
<rqou> offtopic: I just looked into how hackintoshes on qemu boot
<rqou> it's an unbelievable clusterf*ck
<rqou> one method is that qemu boots seabios, which emulates legacy BIOS functionality which boots a bootloader that emulates EFI that then boots xnu kernel
<rqou> the other method is that qemu boots ovmf, which is a hacked up version of edk2 that does EFI
<rqou> this then boots Clover, which is a _differently_ hacked edk2
<rqou> that then boots xnu
mifune has quit [Ping timeout: 240 seconds]
<rqou> wtf
<rqou> why can't it be "a fixed up combo of qemu+ovmf just works?"
<balrog> rqou: why oh why
<balrog> why not just boot EFI directly and provide the necessary keys
<nats`> why trying to boot MAC OS ?
balrog has quit [Ping timeout: 240 seconds]
<pie__> why try to boot at all
<qu1j0t3> asks a dessert
<pie__> dont dscriminate against sweets
<pie__> we control your life
<qu1j0t3> you seem pretty sour to me
* pie__ checks the sweet and sour o' meter
balrog has joined ##openfpga
<pie__> i do seem to be in a superposition of states
<rqou> balrog: I wonder too why Clover is even necessary
<rqou> there are modules that fake the environment to be more Mac-like
<rqou> but those are just modules
<rqou> for extra bonus wtf, Clover patches the ACPI DSDT table
<rqou> for a VM
<rqou> where you can just change the code to generate a more Mac-compatible DSDT in the first place
<rqou> there's also a module that either randomly unmaps some RAM and/or PCI BARs
<rqou> or it pretends to unmap it
<balrog> rqou: OS X is very particular about ACPI DSDT
<balrog> it relies on the somewhat-broken DSDT tables in Macs
<balrog> Well, yeah
<rqou> I can't really tell
<rqou> yeah, but in QEMU you control the *entire universe*
<rqou> none of this bullshit should be necessary
<rqou> it should be possible to generate a DSDT that works with osx and other operating systems
<rqou> how else could boot camp work?
<rqou> but things like OsxAptioFixDrv amaze me that they work at all
<pie__> \o/ software
<rqou> as far as I can tell from skimming the code, it somehow *hooks the xnu kernel* so that the Mac bootstrap can run before jumping *back into the patch driver* so that it can mess with page tables or something
<rqou> in addition to messing with memory and PCI BAR allocation
<pie__> blame commit and look up in mailing list? :P
<rqou> can't find a mailing list
<rqou> hackintosh-ing seems to be centered around random forums everywhere
<rqou> I'm amazed how people who do it can know as much as they do yet write code as hacky as what I've seen
<pie__> :I
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
digshadow has quit [Quit: Leaving.]
digshadow has joined ##openfpga
digshadow has quit [Client Quit]
digshadow has joined ##openfpga
digshadow has quit [Client Quit]
digshadow has joined ##openfpga
carl0s has joined ##openfpga