<digshadow> azonenberg: someone at work is interested in FOSS VHDL synthesis
<digshadow> do you think yosys could support this or would it better be a separate tool
<azonenberg> digshadow: there's work ongoing to add a vhdl front end to yosys afaik
<azonenberg> how far it's progressing, i dont konw
<digshadow> k, I'll pass that along
<azonenberg> yosys is very modular, most of the work is done on abstract netlist stuff
<azonenberg> basically once you get past the parser it's language independent
<digshadow> thats what I figured
<digshadow> yeah an intermediate representation
<balrog> azonenberg: someone here was interested in that
<balrog> I forget who
<balrog> egg|egg might know since he has a lot of Ada background
<balrog> and was talking to that person I think
<lain> digshadow: it's very very early stages, but: https://github.com/tgingold/ghdlsynth-beta
<lain> digshadow: that's yosys + ghdl for foss vhdl synth
digshadow has quit [Quit: Leaving.]
clifford has quit [Ping timeout: 260 seconds]
<whitequark> rqou: you can congratulate me on having to download six hundred android repositories
digshadow has joined ##openfpga
promach has joined ##openfpga
digshadow has quit [Client Quit]
LoveMHz has quit [Ping timeout: 260 seconds]
digshadow has joined ##openfpga
LoveMHz has joined ##openfpga
[X-Scale] has joined ##openfpga
X-Scale has quit [Ping timeout: 260 seconds]
[X-Scale] is now known as X-Scale
digshadow has quit [Quit: Leaving.]
<rqou> congrats whitequark :P
<rqou> were you also trying to compile adb/fastboot?
<rqou> azonenberg, balrog, lain: i'm the one that was working on Yet Another Attempt at a VHDL Frontend for yosys
<rqou> i just finished the basic parser a few hours ago
<rqou> next will be the giant task of semantic analysis
<whitequark> rqou: no
<whitequark> I bought a new screen for my phone
<whitequark> it's identical to the one it came with from the fctory but the subpixel order is reversed
<whitequark> this is aggravating
<azonenberg> rqou: lol
<azonenberg> so you have bgr instead of rgb?
<rqou> have you considered buying a new phone? :P
<whitequark> that or the other way around.
<whitequark> rqou: that's expensive
<qu1j0t3> whitequark: oh, nasty.
<rqou> aren't screens expensive too?
<whitequark> not on taobao
<whitequark> it ws like 100hkd? vs $500 for a new phone
<whitequark> ah ok 260 hkd
<whitequark> that's still extremely cheap
<rqou> hrm, maybe just idevice screens are more expensive
<whitequark> duh
pie_ has quit [Ping timeout: 240 seconds]
<qu1j0t3> I sense whitequark would pay more for the right subpixel ordering though
<whitequark> qu1j0t3: nah
<whitequark> I got a screen from the exact same vendor in the past, and it was fine
<qu1j0t3> rollthe die again?
<whitequark> lol
<whitequark> no I am going to complain to them (through the broker)
<rqou> ah, you're one of those "break your phone all the time" people :P
<whitequark> rqou: yes
<whitequark> this is at least the fourth time
<azonenberg> rqou: lol
<azonenberg> The only phone i remember "breaking"
<azonenberg> is my old droid 1
<azonenberg> i still have it
<azonenberg> it was literally falling apart
<azonenberg> the key caps had all come off b/c the adhesive degraded from time and heat
<azonenberg> i had to touch-type
<azonenberg> Some of the buttons didnt work
<whitequark> maybe you can figure the order from that
<azonenberg> i had to do some soldering to swap the power button with one of the volume buttons
<whitequark> but I'm too lazy
<azonenberg> b/c i needed the power button more :p
<azonenberg> and in the process i lost the power button cap
<azonenberg> so i had a little hole in the side of my phone
<azonenberg> covered with electrical tape for some degree of weatherproofing
<azonenberg> the battery lasted half a day if i was lucky
<azonenberg> i figured it was time to put it down
zino has quit [Remote host closed the connection]
ChickeNES has quit [Ping timeout: 260 seconds]
kmehall_ has quit [Read error: Connection reset by peer]
kmehall has joined ##openfpga
ChickeNES has joined ##openfpga
<rqou> huh, @scanlime is using migen. is this your doing whitequark?
<whitequark> partially?
<whitequark> sb0 also met her in sf
DocScrutinizer05 has quit [Disconnected by services]
DocScrutinizer05 has joined ##openfpga
amclain has quit [Quit: Leaving]
digshadow has joined ##openfpga
Bike has quit [Quit: sleep tho]
<egg|egg> balrog: you rang?
<egg|egg> rqou: how goes the parser?
<rqou> i was coding all night last night, and now it seems more-or-less finished (for just parsing, no semantic analysis)
<rqou> except i possibly found a new bug
<rqou> so consider it "about 90% done, where the last 10% takes 90% of the time" :P
<rqou> it parses real designs though
<rqou> and old parallax propeller clone i wrote ages ago as well as the J-core sh2 project
Hootch has joined ##openfpga
<rqou> egg|egg: does ada have fooarray'element ?
<egg|egg> rqou: uuh I'm confused
<rqou> an attribute named "element" on an array object
<rqou> in vhdl this is supposed to be a predefined attribute
<rqou> and it's a subtype
<egg|egg> 'Element isn't an attribute
<rqou> well it is in vhdl
<egg|egg> hmm
<rqou> before i go down this rabbit hole i'm going to write some unit tests
<egg|egg> I guess whenever you might need that you just feed it as a generic parameter in Ada
<rqou> the problem is that "fooarray'element" might be usable as a type_mark
<egg|egg> i.e. generic type T is private; type A is array (<>) of T;
<rqou> which isn't really ambiguous anywhere except... subtype_indication
<egg|egg> hm
<egg|egg> I wonder whether there are attributes that can be types in Ada
<egg|egg> I don't think so? but then there are so many attributes
<rqou> oh this is definitely a type
<rqou> it says "Kind: Subtype"
<rqou> i don't think you can create your own "Kind: Subtype" attributes though
<egg|egg> huh, you can make your own attributes?
<rqou> you can, but afaik they can only by of types that you can define using a type definition
<rqou> they cannot be other built-in "kinds of objects"
<rqou> "attribute_declaration ::= attribute identifier : type_mark ;"
<rqou> and type-mark cannot be "a meta-type"
<egg|egg> huh
<egg|egg> afaik in Ada you have the attributes given to you by the compiler and that's it
<egg|egg> (some from the language, some implementation-defined typically)
<rqou> so T'BASE is a predefined attribute with the following note: This attribute is allowed only as the prefix of the name of another attribute; for example, T'BASE'LEFT."
<rqou> but none of the other ones have that
<egg|egg> O_o
<egg|egg> so 'Base exists in Ada, but that's just a type
<egg|egg> (and that answers my question, some attributes are types)
<rqou> it's a type here too
<egg|egg> but it's usable on its own
<egg|egg> X : T'Base is a good declaration
<rqou> aaaargh
<rqou> that's exactly what i don't want
<egg|egg> that's not legal in VHDL? but... why :-p
<rqou> does ada have subtype_indication?
<rqou> so now i can have excitement like X : T'Base(0 to 2)
<rqou> or even better, generic map (T'Base(0 to 2))
<egg|egg> generic map is such a weird syntax
<egg|egg> but yes, you can take T'Base and constrain it, I think? maybe?
<egg|egg> I'm mildly confused as to whether you can do so anonymously
<rqou> so in VHDL you can't do that to base
<rqou> but you can (from what i'm reading) do it to fooarray'element
<whitequark> rqou: it's been 7 hours and I'm still downloading Android
<rqou> sounds right
<rqou> hopefully you're not just trying to build adb+fastboot? :P
<whitequark> no
<egg|egg> rqou: yup, you can do this https://ideone.com/OrpctY
<rqou> does ada have array index constraints?
<egg|egg> so, background so that you understand what this does (in Ada, what the hell is VHDL thinking it's doing anyway). 'Base is "the underlying type"
<rqou> i am thiiiis close to just saying "f*ck it" and special-casing 'elements
<egg|egg> and the standard mandates little about it, but 1. it must contain the type (so 1 .. 7), it must contain 0, and it must be symmetrical around 0 (with possibly one more negative value)
<egg|egg> rqou: array index constraints? in which context?
<rqou> or i might just not implement this feature
<egg|egg> why the fuck did I get that from adacore and not adaic
<rqou> right, so vhdl has something very similar
<egg|egg> yeah, we encountered it the other day iirc
<rqou> now what happens when T'Base is an array type? you can give it an index constraint
<rqou> and this newly-constrained type is indistinguishable from a normal "calling a function that lives in a built-in attribute"
<rqou> i think i might need to do some refactoring of "name"
<egg|egg> rqou: prefix of base attribute must be a scalar type, in Ada
<rqou> VHDL says "Prefix: Any type or subtype T."
<egg|egg> "the underlying type" isn't a very useful concept for an array
<egg|egg> welp
<rqou> and fooarray'elements exists too
<rqou> i might just say "f*ck it, you cannot parameterize an object this way. you must use a generic"
<rqou> i might revisit it if i refactor how "name" works
<egg|egg> back in a bit
<rqou> hrm, problem
<rqou> hey armchair laywers: will i have a problem publishing my parser if the repo contains a copy of the VHDL EBNF?
<rqou> i don't want to have to go through the effort to scrub it
<jn__> git filter-branch is actually quite easy to use
<jn__> rqou: are you going to integrate your VHDL parser into yosys?
<rqou> that's the eventual goal, yes
<jn__> nice
<rqou> i haven't even started on semantic analysis, so don't get too excited
<rqou> yeah worst case i'll just filter-branch out the verbatim copy of the EBNF
<rqou> stupid ieee copyrights
<egg|egg> blarg
<rqou> (says someone who is technically a member of the IEEE)
<egg|egg> it's probably not ok either to dump copies of the ARM into things, but at least you can stare at it for free, so that's less likely to be the object of daft litigation
<rqou> yeah ieee standards aren't free either
<egg|egg> oh, actually This document may be copied, in whole or in part, in any form or by any means, as is or with alterations, provided that (1) alterations are clearly marked as alterations and (2) this copyright notice is included unmodified in any copy. Compiled copies of standard library units and examples need not contain this copyright notice so long as the notice is included in all copies of source code and
<egg|egg> documentation.
<egg|egg> (for the ARM that is)
<egg|egg> see http://www.adaic.org/resources/add_content/standards/05rm/html/RM-TTL.html for all the iterations of the legalese
<rqou> VHDL says: "This document is copyrighted by the IEEE. It is made available for a wide variety of both public and private
<rqou> uses. These include both use, by reference, in laws and regulations, and use in private self-regulation,
<rqou> standardization, and the promotion of engineering practices and methods. By making this document
<rqou> available for use and adoption by public authorities and private users, the IEEE does not waive any rights in
<rqou> copyright to this document."
<egg|egg> it seems that in all aspects VHDL takes Ada and does profoundly daft things to it
<egg|egg> (and by aspect I do not mean the Ada 2012 technical term)
<rqou> and the watermark says: "Copyright © 2009 IEEE. All rights reserved. Authorized licensed use limited to: Univ of Calif Berkeley. Downloaded on <snip> from IEEE Xplore. Restrictions apply."
<rqou> egg|egg: i suppose you weren't here for the discussion about (verilog's) vpi_user.h?
<egg|egg> no?
<egg|egg> [do I want to know]
<rqou> it's a normative part of the verilog spec for defining the interface for code to interact with verilog simulators
<rqou> it's a c header, obviously
<rqou> but it's also under the ieee copyright
<rqou> icarus verilog has an (afaik) clean-room reimplementation of it
<rqou> but that reimplementation is under the copyleft GPL, which is not always desirable
<egg|egg> aosentudansoetudsath
<rqou> at this point i believe whitequark can contribute some strong opinions about copyright? :P
<whitequark> the institution of copyright is unethical and should be destroyed
<rqou> but it wasn't originally
<rqou> companies like disney (and tech) made it so
<whitequark> intellectual property always was
<whitequark> it was less obviously harmful with smaller terms, sure
<rqou> what about the original goal of protecting and encouraging creators/inventors?
<egg|egg> rqou: an annoying side-effect of all this VHDL & Ada talk is that now I keep thinking about language design >_>
<whitequark> (well, this depends on your definition of "intellectual property". I'm not particularly opposed to, say, trademarks or trade secrets)
<whitequark> rqou: was it achieved?
<rqou> for disney it was :P
<rqou> but overall, not really
<whitequark> well, no, that's not even the right question
<whitequark> do you actually know the history of copyright?
<rqou> vaguely?
<whitequark> "protecting creators" wasn't the original goal, the original goal of copyright was censorship
<whitequark> you could maaaaaybe argue that protecting inventors was the original goal of *patents*
<whitequark> although it's lost on me how encouraging rent-seeking does anyone any good except for, well, rent-seekers
<whitequark> it's a good time to mention that I don't consider creation or invention an unconditional good
<whitequark> if you invent something and patent it and therefore prevent someone else from disseminating the same thing freely, it's a net negative
<rqou> in general i'm much more against patents than a (properly limited) copyright
<rqou> especially with all the bullshit software patents that have recently appeared
<whitequark> I'm against anything that restricts what should be human advancement to advancement of bank accounts of a few guys
<whitequark> this is especially bad with publicly funded research and publicly funded standards bodies, among other things
<whitequark> but also, fuck xilinx for keeping their software to themselves
<rqou> especially with the current state of the internet, i'm inclined to agree
<rqou> e.g. "but i want people to buy my book/game/software" doesn't work so great because they all get instantly pirated
<rqou> and it turns out that people who would have bought it often still buy it
<rqou> and people who pirate often wouldn't have bought it anyways
<whitequark> it's not even that
<whitequark> the way you *should* respond to piracy is by emphasizing the human cost of it in today's economy
<whitequark> instead, once you encourage this adversarial relationship, you don't get any sympathy back
<whitequark> it's really the same problem as entitlement to the time of open-source maintainers
<rqou> but then this gets into "capitalism is pretty broken"
<rqou> and that's a much bigger problem :P
<whitequark> but it's not like the issue of copyrights or patents or rent-seeking in general or the plight of indie artists is separable from the issue capitalism
<whitequark> of*
<whitequark> you can't treat them in isolation, and in all likelihood you can't change them in isolation too
scrts has quit [Ping timeout: 240 seconds]
<whitequark> can you really conceive a rollback of the copyright ratchet under neoliberalism?
scrts has joined ##openfpga
<rqou> yeah that's probably true
<whitequark> wtf is up with android buildsystem
<lain> s/ buildsystem//
<whitequark> it has "breakfast", "brunch", "lunch", targets "eatme", "omnom" and "bacon"
<lain> lol
<jn__> someone was really hungry :D
<rqou> answer: android sucks :P
<whitequark> not to mention their release codenames
clifford has joined ##openfpga
<jn__> has anyone tried tizen instead? :)
<whitequark> is that still alive?
<egg|egg> whitequark: what's wrong with the codenames?
<jn__> o hi clifford
<rqou> my housemate pointed out an interesting observation: android on a phone (a fully controlled and unchangeable environment) takes longer to boot than a white-box PC
<rqou> booting windows
<egg|egg> (Ok, kitkat was really shit, that got a lot of internal backlash, but aside from that)
<whitequark> egg|egg: food
<egg|egg> food \o/
<jn__> whitequark: i'm in the tizen irc channel and the most activity comes from the news bot. i'm not sure how alive it is.
<rqou> isn't tizen the samsung heap of crap? :P
<whitequark> rqou: it's interesting (boot times)
<jn__> rqou: pretty much, i guess
<whitequark> I think a large part of that is initialization of vendor blobs
<lain> tizen webpage says it's a Linux Foundation project
<whitequark> but another part of it is just that phones don't reboot often
<rqou> they just crash instead? :P
<lain> ^
<whitequark> hm?
<whitequark> a PC (not a laptop) would be booted at least once per day, in a typical office setting
<whitequark> a phone? once per month if you don't let it discharge?
<rqou> um, not for me?
<lain> the last N times my android phone has (re)booted, it's been because of a crash
<whitequark> are you in "a typical office setting" ?
<rqou> not right now, but even when i was i tended to avoid rebooting
<lain> my desktop stays on 24/7, my office computer boots when I arrive and shuts down when I leave
<lain> but yeah, android boot times are shameful
<rqou> i would usually reboot every monday when i was working
<whitequark> this is an iphone phone case that runs android
<whitequark> yes, really
<whitequark> they went way through their goal too
<lain> modern windows, and like, sony digital cameras, use a known-good hibernate-style image for fast boot: they do a full clean boot after an update, and then essentially hibernate
<lain> all future boots, unless you force a full reboot, are just loading that hibernate image
<rqou> even a clean boot of windows seems to be not slower than booting android
<lain> yes
<lain> it's way faster, for sure
<jn__> meanwhile some linux boxes do a cold boot in about five seconds
<lain> so does windows, from a full clean boot without the image
<rqou> even poetteringware boots faster :P
<lain> and my android phone takes like .. actually I wonder how long it takes
* lain reboots it
<rqou> egg|egg: you can clearly tell what era VHDL comes from: "In this standard, the VHPI information model is described using the UML notation."
<egg|egg> lɒl
<rqou> VHPI is a C interface btw
<lain> a friend is learning UML in uni right now
<lain> even they recognize it's awful :P
<rqou> so it's manual OO in C
<lain> vhpi is neat though
<egg|egg> Rational Software! \o/
<egg|egg> (phl worked there)
<lain> verilog has the same thing
<rqou> VPI doesn't require UML, does it? :P
<rqou> also, i'm not aware of a single simulator that implements VHPI
<rqou> even GHDL doesn't
<rqou> GHDL implements their own thing called AVHPI, which is like VHPI except a) an Ada interface and b) actually OO because of (a)
<rqou> for bonus points, GHDL _also_ implements VPI
<rqou> Tristan Gingold seems to be an... interesting person :P
<lain> so my moto x pure edition ( https://en.wikipedia.org/wiki/Moto_X_Style ) takes 1m20s to reach the launcher or whatever, and it took another minute to finish launching the stuff I have running at boot (Twilight [like f.lux], Pebble smartwatch app, Accuweather)
<lain> afaik ghdl has vhpi now, some docs are outdated
<rqou> it does?
<rqou> nah, that's different
<rqou> that's just a hacky FFI
<lain> FFI?
<rqou> foreign function interface
<lain> ah
<rqou> it allows vhdl to call out to c code
<rqou> but only with "actual values"
<rqou> you can't e.g. capture the signal objects themselves and interact with them
<rqou> in fact afaik there is currently no way to do that
<rqou> hence why my hack myhdl-vhdl-verilog-test uses magic names
* egg|egg stares at backlog
<egg|egg> hah, I had forgotten that my client rewrote lɒl in IPA
<rqou> so wtf
<rqou> ghdl really does allow attribute_name in a type_mark
<rqou> but trying to use it seems to just raise an error
<rqou> hrm, i can't actually understand why ghdl is barfing here
<rqou> bug report time? :P
<rqou> it seems to accept 'subtype in this context and then barfs at a later stage
<rqou> it doesn't allow 'element
<rqou> ooooh
<rqou> subtype is a keyword
<rqou> hrm, that's not the reason
<whitequark> why.... why does android build unrar
<whitequark> did they decide the attack surface was not large enough or what
<rqou> what if you want to open your downloaded warez on your phone? :P
<whitequark> why warez?
<rqou> what else is stored in rars?
<rqou> :P
<whitequark> pretty sure most russians still install rar the first thing
<whitequark> (and the oldschool ones, far the second)
<rqou> anyways, so ghdl seems to accept 'subtype in a type_mark so i definitely have a bug
<rqou> exciting
<rqou> i really need to refactor how "name" works
<rqou> but i'm pretty burned out from my last coding sprint
<rqou> i might not get to this immediately
<rqou> i see ghdl has a hack too
<rqou> in the form of an Allow_Indexes argument to the "name" parser
<rqou> i probably should have done that from the beginning
<rqou> argh, i'm going to need to write unit tests too
<lain> it deeply bothers me that Sort By on youtube is under Filters
<lain> sort is not a filter.
<whitequark> great, now it's building chromium
<whitequark> why is it building chromium. it doesn't HAVE chromium
<whitequark> is the builtin browser chromium but shitty.
<rqou> i wonder if my 5000-line bison file is better or worse (in terms of maintainability) than ghdl's 9000-line parse.adb
<rqou> OH I SEE WHY GHDL HAS IT SO EASY
<rqou> it can't have generic types!
<rqou> i really need to write a fuzzer
<rqou> i should fuzz ise, ghdl, and my parser simultaneously and see what happens
<jn__> that reminds me of vloghammer
<rqou> who wants to help make vhdlhammer?
<jn__> bibor: ^
scrts has quit [Ping timeout: 240 seconds]
<rqou> hrm, for april fools i should implement vhdl protect directives :P
<rqou> egg|egg: this part of VHDL has aged poorly: http://i.imgur.com/vXVsWw0.png
<lain> lol
<egg|egg> rqou: is there a part that has aged well?
<rqou> not really
<rqou> anyways, idk if you saw, but GHDL can't have types in generics
<rqou> they punted on the really hard parsing problem :P
<rqou> hrm, the normative vhpi_user.h has #if defined (_MSC_VER) in it
<bibor> jn__: really not my field of expertise, but: yay vhdl frontend? \o/
<rqou> yup, i'm slowly working on yet another attempt at one
<bibor> 10/10 would write code for
<bibor> i'm still sad yodl died :(
<rqou> oh wtf
<rqou> guess how vhpi maps std_logic to integers?
<rqou> (int => std_logic) 0 => U, 1 => X, ...
<rqou> so we're obviously off to a good start :P
<rqou> 2 => 0, 3 => 1, 4 => Z, 5 => W, 6 => L, 7 => H
<bibor> lulz
<rqou> so yeah, 0/1 map to 2/3
<rqou> hrm, verilog doesn't have W/L/H does it?
<rqou> i wonder how that behaves in a mixed-language simulation?
<bibor> nope verilog is 0/1/x/z
<bibor> (bit type)
<rqou> so what happens to W/L/H?
<bibor> idk. only way i used verilog is in testbenches for digital stuff
promach has quit [Ping timeout: 256 seconds]
<lain> you would probably resolve it to 0/1 before then, or something
promach has joined ##openfpga
zino has joined ##openfpga
<rqou> hrm: "If the attribute designator denotes a predefined attribute, the expression either shall or may appear, depending upon the definition of that attribute (see Clause 16); otherwise, it shall not be present."
<rqou> so special-casing the builtin stuff seems reasonable
<rqou> hrm, it's 5am but my sleep schedule is totally shot. what do?
<jn__> drink beer?
<rqou> no beer, store's closed
<egg|egg> coffee?
<rqou> i hate coffee
<egg|egg> but how do you produce theorems
<rqou> i'm not a mathematician :P
<egg|egg> instead of coffee, cotheorems then?
<egg|egg> (to produce ffee, like a comathematician)
<egg|egg> (terrible joke is terrible)
<jn__> co-ffeorems?
nurelin has quit [Read error: Connection reset by peer]
<bibor> a guy at our hackerspace made caffeine chocolate. thats a reasonable alternative
<rqou> i already stayed up all night yesterday while doing the final push getting the parser working
<rqou> hrm, verilog doesn't have hierarchies the way vhdl does, does it?
<rqou> from GHDL documentation: "Furthermore, you should not put units in the ieee library. "
<rqou> hmm i wonder why? :P
<rqou> so nvc doesn't accept attribute_names in a type_mark either
pie_ has joined ##openfpga
pie_ has quit [Changing host]
pie_ has joined ##openfpga
scrts has joined ##openfpga
<rqou> wait wtf afl (the fuzzer) is designed by the same guy that does polyurethane resin hacks and micro gearboxes?
<lain> but can it fuzz the resin
pie_ has quit [Ping timeout: 268 seconds]
m_w has joined ##openfpga
<jn__> i actually used https://upload.wikimedia.org/wikipedia/commons/0/08/Rabbit_american_fuzzy_lop_buck_white.jpg as the first slide of a talk about afl-fuzz, once :)
<lain> :3
pie_ has joined ##openfpga
Bike has joined ##openfpga
pie_ has quit [Ping timeout: 268 seconds]
massi has joined ##openfpga
m_w has quit [Quit: leaving]
m_w has joined ##openfpga
amclain has joined ##openfpga
nurelin has joined ##openfpga
pie_ has joined ##openfpga
<balrog> looks like cypress is really pushing the hardware peripherals for PSoC 6
digshadow has quit [Quit: Leaving.]
<pointfree> balrog: So do you think they are trying to minimize the use of UDB-based peripherals?
promach has quit [Quit: Leaving]
digshadow has joined ##openfpga
massi has quit [Remote host closed the connection]
pie__ has joined ##openfpga
digshadow has quit [Quit: Leaving.]
m_w has quit [Read error: Connection reset by peer]
<rqou> hrm, i now have 120 reduce/reduce conflicts after (i think) fixing the type_mark bug
<rqou> this definitely looks like a "too much lookahead required" problem
<rqou> but now it's really hard to be sure
m_w has joined ##openfpga
digshadow has joined ##openfpga
<pie_> azonenberg, i dont suppose you know anyone that keeps a bibliography of IC reversing stuff
<pie_> also dumb question, can you learn everything you need from openly available stuff or does one need to be in the in-group? :P
<pie_> access to equipment totally doesnt count
m_w has quit [Quit: Leaving]
<azonenberg> pie_: siliconpr0n has a lot of links to papers and presentatoins
<azonenberg> plus the de facto Wikipedia of silicon RE
<pie_> ok
<pie_> same place as usual then xD
<azonenberg> Lol we're pretty much the authority on the subject at this point, as far as lots of info in one place
<azonenberg> there's a ton of loose academic papers, con talks, etc
<azonenberg> but its all scattered around the net
<azonenberg> we're trying to collect as much as we can in one place
<azonenberg> The only other resource i can think of recommending is the archives of the CHES conference
<azonenberg> it's a very math-heavy / crypto con but they cover embedded stuff too
<azonenberg> including both side channel and invasive attacks
<pie_> cool
<azonenberg> and often particularly good ches papers get linked on siliconpr0n at some point
<pie_> much props \o/
<azonenberg> But we need to do a better job of making a bibliography
<pie_> on the topic of
<pie_> FPGA, i need to really get past the led blinker at some point....
<azonenberg> individual pages are full of citations but we dont have a great "getting started" reading list that i know of
<azonenberg> Yes, that helps
Hootch has quit [Quit: Leaving]
<pie_> but i keep tellingmyself i need to focus on learning linear algebra first (for something completely different)...on that note... >P
<pie_> * :P
<azonenberg> So many yaks, so little shaving cream?
<rqou> so I mentioned this before, but I specifically signed up for a microfabrication course to learn silicon RE :P
m_w has joined ##openfpga
<rqou> not recommend unless you're confident in your skillz because your grade gets attached to it :P
<qu1j0t3> azonenberg: yessss
<pie_> heh
<pie_> rqou, wish we had something like that
<pie_> but i know theres a course up on youtube iirc
<rqou> it's different from actually going into a cleanroom
<pie_> sure
<pie_> also, m$urica :P
<pie_> well who knows, maybe one of my next lives
<rqou> our teaching fab is 5um and supposedly cost "only" about USD$1million :P
<pie_> lol
<pie_> thats just the tuition of about 5 students :PPPPP
<rqou> although it has a really fancy Marvell-donated research fab to provide support staff
<rqou> nah, UCB is a public university, so tuition isn't quite *that* high
<rqou> although it might be getting there if the state of California continues to not budget well
<pie_> :I
digshadow has quit [Quit: Leaving.]
<pie_> btw rhme2 finished a couple weeks ago and there should be writeups
<pie_> im not looking at them because i want to do the challs lol
pie_ has quit [Ping timeout: 256 seconds]
<rqou> oh lcamtuf works for google
<rqou> not too surprising i suppose
digshadow has joined ##openfpga
digshadow has quit [Client Quit]
<rqou> hrm, anyone know how to report (minor) afl bugs?
<rqou> oh nvm, email the guy
<azonenberg> lol
<rqou> yeah the bug is that when it tries to pin to a cpu it doesn't take cpuset into account
<jn__> or send a mail to the afl mailing list
<rqou> anyways, i modified my parser to abort() when it detects an ambiguity and i'm now throwing afl at it to see what happens
<jn__> lcamtuf is really helpful when he replies to mails
<rqou> it's already managed to exhaust the parser stack a number of times
<rqou> which apparently show up as hangs
<jn__> hangs are simply timeouts
<rqou> no, i ran it manually and you hit this case: "yyerror (YY_("memory exhausted"));"
mIKEjONES has joined ##openfpga
<rqou> hrm, no actually they just seem to be flakiness
<rqou> they haven't managed to make the parser abort yet, so that's a good sign
<rqou> hrm, all hangs detected by afl involve lots of open parens
<rqou> i wonder why :P
<rqou> also a number of them seem to be frivolous
<rqou> herp derp i haven't actually been giving afl the keyword dictionary
<rqou> at least afl hasn't immediately blown up my parser with an ambiguity yet :P
<rqou> (i added a hack to abort() on ambiguities so afl can find them)
GreeningGalaxy has joined ##openfpga
<GreeningGalaxy> whois _whitelogger
<GreeningGalaxy> um. well.
<GreeningGalaxy> hi everyone
<rqou> hi
<GreeningGalaxy> I swear to noob I am not god
<egg|egg> rqou: how goes the parser?
* egg|egg stares at backlog
<rqou> fixed the "attribute in type_mark" issue
<egg|egg> rqou: ah :-p
<rqou> now i have over 100 reduce/reduce conflicts :P
<egg|egg> :D :D :D
<rqou> but afaik this is due to requiring more than 1 token of lookahead
<rqou> not because of ambiguity
<egg|egg> yeah
<rqou> right now i'm trying to see if i can get afl to blow it up
<rqou> even though afl isn't really designed for this
<rqou> afl has managed to exhaust the parser stack, but it hasn't managed to find an ambiguity
<egg|egg> hah
<rqou> idk if i need to fix those
<rqou> they're totally bogus to start with
<rqou> at some point if someone is feeling bored they can try to fix the grammar to be truly LR(1)
<rqou> rather than LR(k)
<rqou> there is an algorithm to do this
digshadow has joined ##openfpga
<rqou> hrm, so afl still hasn't blown up my parser yet