Results 1 ... 122 found in all logged channels for 'valgrind'

(pest) asciilifeform: the only place it is ever of practical use is occasions when you wanna run e.g. 'valgrind', where you gotta hijack malloc and free
(asciilifeform) asciilifeform: let's take concrete case -- two proggies, say, are trb and valgrind.
(asciilifeform) asciilifeform: trinque: afaik, the only conceivable use for even having dynamic build option, is debuggism (e.g. valgrind, requires it)
(trilema) mircea_popescu: http://btcbase.org/log/2018-05-16#1814305 << i was thinking, valgrind is more of a "let's see what the githubheads did here" tool than a "i wonder if i can write code" tool.
(trilema) a111: Logged on 2018-05-16 14:42 asciilifeform: ( retaining some knob for dynamic linkage isn't totally useless, it enables such things as valgrind ; but i'm quite prepared to lose valgrind, it is not really so necessary when writing asciilifeform-style -- heapless -- ada )
(trilema) asciilifeform: (ftr : if you heap-allocate from within ada, you will not be able to valgrind, on static-musltronic gnat.)
(trilema) asciilifeform: ( retaining some knob for dynamic linkage isn't totally useless, it enables such things as valgrind ; but i'm quite prepared to lose valgrind, it is not really so necessary when writing asciilifeform-style -- heapless -- ada )
(trilema) asciilifeform: ( and yes, i'm quite aware that, e.g., valgrind, won't work on such a machine )
(trilema) sina: asciilifeform: not valgrind for valgrinding, but ctgrind patched valgrind for checking constant timeness of existing code
(trilema) asciilifeform: and if you even so much as considering valgrind, your ada is ill-conceived
(trilema) sina: asciilifeform: btw I was going to ask you about your feelings on this https://github.com/cforler/Ada-Crypto-Library ...obviously hasn't been impl for constant time/space but regardless. may be possible to ctgrind it using that valgrind patch I linked in the logs
(trilema) sina: (Checking that functions are constant time with Valgrind)
(trilema) mircea_popescu: cachegrind is a valgrind plugin eh.
(trilema) phf: valgrind i take it from what asciilifeform is saying instruments the code, in an equivalent of putting printf after every statement
(trilema) mircea_popescu: "valgrind [when used for this purpose through this module] simulates" etc.
(trilema) mircea_popescu: yes. it is a valgrind plugin is it not.
(trilema) asciilifeform: valgrind doesn't simulate cpu
(trilema) mircea_popescu: if you're interested in either gprof or valgrind's callgrind, or both or whatever, his thing is useful. but i'm not. specifically because bitcoin is too large and slow to be run on valgrind-simulated cpu, and moreover if we could do that we wouldn't have any of the problems we currently approach anymore.
(trilema) mircea_popescu: asciilifeform anyway, either approach discussed is ridiculous. i don't need valgrind to emulate a processor and then guess how long things' take ; i don't want gprof's sampling either. just awk the codebase to add a printf after each instruction spitting out time and recompile. that's it.
(trilema) Framedragger spent longer than wants to admit sorting out his heap and valgrind'ing. too much python is bad for a person
(trilema) Framedragger: myeah #trilema is basically valgrind.
(trilema) asciilifeform: ( if you recall , that was when we had a trb node, offline, eat the complete chain to date from disk, with timer and valgrind going )
(trilema) mircea_popescu: why the everloving would you valgrind on a laptop ?
(trilema) davout: asciilifeform: didn't experience that, what i did experience was "oh no, can't install valgrind because fuck you"
(trilema) davout: OSX, totally the platform sane people develop on "valgrind: This formula either does not compile or function as expected on macOS" hurrrr
(trilema) ben_vulpes: asciilifeform: i'd have to restart the process to get data out of valgrind, right?
(trilema) asciilifeform: ben_vulpes et al: what happened to actually instrumenting with valgrind etc, as i described in agonizing detail on the ml last year ? the linux footprint thing is not very useful.
(trilema) asciilifeform: sorta like a special-purpose 'valgrind', for debugging rng.
(trilema) gribble: Ignoring whining about OpenSSL in Valgrind - rachelbythebay: <https://rachelbythebay.com/w/2012/12/14/quiet/>; openssl / valgrind - Stack Overflow: <http://stackoverflow.com/questions/472809/openssl-valgrind>; #299 (Second memory leak) – libssh2: <http://trac.libssh2.org/ticket/299>
(trilema) ascii_field: who doesn't believe this - get thee to valgrind, and see for yourself.
(trilema) decimation: probably valgrind 'callgrind' too
(trilema) decimation: I think you can massage valgrind to dump callgraph
(trilema) assbot: valgrind - Interpreting callgrind data - Stack Overflow ... ( http://bit.ly/1IHTQrD )
(trilema) decimation: ben_vulpes: http://valgrind.org/docs/manual/cl-manual.html < valgrind has a gadget for callgraph creation too
(trilema) decimation: asciilifeform: I bet you could control valgrind gdb with ida
(trilema) asciilifeform: decimation: valgrind is not a debugger
(trilema) decimation: mats: you could probably do this with valgrind too
(trilema) mod6: heh, i built valgrind on deb6 lastnight as well. didn't work with bitcoind either statically built, or dynamically. might have to give this a try tonight on the gentoo aws instance.
(trilema) ascii_field: mircea_popescu: btw i regard valgrind, igprof, etc. as 'c cleaners' rather than cpp. they only kinda, sorta work on cpp
(trilema) mircea_popescu: oh ok. so i didn't hallucinate valgrind igperf etcv
(trilema) mircea_popescu: imagine a 1802 valgrind
(trilema) ascii_modem: this is not valgrind
(trilema) ascii_modem: forget the valgrind-style run log, it is worthless
(trilema) ascii_modem: otherwise 'valgrind' would've sufficed
(trilema) mircea_popescu: somehow i thought we're still looking at valgrind stuffls.
(trilema) decimation: asciilifeform: your complaint is that valgrind doesn't catch all the memory use?
(trilema) asciilifeform: this is -exactly- my complaint with valgrind.
(trilema) asciilifeform: this is not valgrind! it does not need the process to terminate.
(trilema) asciilifeform: decimation: all i noticed is the same 'pocket change' leaks as valgrind shows
(trilema) mod6: in addition to the valgrind tests that ascii did earlier this month, i was going to dig into some of that myself in june.
(trilema) asciilifeform: instead of waiting for the thing to die, as in valgrind.
(trilema) asciilifeform: 1) and thus, valgrind will not see this as a leak, because the memory is theoretically reclaimable
(trilema) mircea_popescu: notrly valgrind's problem, this. if system reports it as allocated, it's allocated as far as its concerned
(trilema) davout: can someone explain to me how i'm able to malloc into existence more than 1tb, fill the first byte with some random int, and have valgrind report the massive allocated space. all this with a whopping 4gb ram and 512gb hdd?
(trilema) asciilifeform: -nor- has any detectable (through the usual automatic means, e.g. valgrind) leaks.
(trilema) asciilifeform: mod6: i assume you read my 'valgrind' outputs ?
(trilema) decimation: asciilifeform: re: valgrind < are you running valgrind on the poor little pogo or on an x86?
(trilema) mod6: and as far as valgrind, I'll do some stuff with it as I get through these other things.
(trilema) nubbins`: stan, one of these days i'll read up on valgrind and pick sense outta your last couple ML posts
(trilema) ascii_field: for the record, i'd rather wash my car than run (esp. read the output of) the valgrind. but who is going to do it if i didn't, l0l
(trilema) mircea_popescu: i dunno if you think valgrinding bitcoind is sexier than fixing phuctor or something, but a) it is not, and moreover b) doing it while the other waits is actually negative, it's not even zero. moreover c) this nonsense is exactly how the power rangers got to be power rangers.
(trilema) mircea_popescu: stop fucking around with valgrind and whatnot omg.
(trilema) asciilifeform: ^ on first sight, very perverse findings which appear to contradict those of ordinary 'valgrind'
(trilema) asciilifeform fired up valgrind-massif with --pages-as-heap=yes
(trilema) ascii_field: later tonight i might do a valgrind with 'massif' with --pages-as-heap=yes
(trilema) mircea_popescu: asciilifeform (Fragging; or dynamic crud somehow not covered by valgrind, etc.) << i toldja it doesn't only leak in that place.
(trilema) asciilifeform: [BTC-dev] Full Orphanage Thermonuke Valgrind Output.
(trilema) asciilifeform: will post valgrind output tomorrow
(trilema) asciilifeform started valgrindized run of thermonuked-orphanage tester
(trilema) ascii_field: tonight i'ma valgrind it
(trilema) punkman: valgrind's dynamic substitution for memory allocator -does not cover all of bitcoind << I'm guessing it's that custom allocator for some privkey data types
(trilema) asciilifeform: mircea_popescu, chetty, et al: in yet another - arguably inverted! - manifestation of The Problem, it appears that valgrind's dynamic substitution for memory allocator -does not cover all of bitcoind!-
(trilema) asciilifeform: beginning to suspect that valgrind isn't seeing the whole picture
(trilema) asciilifeform: am i the first to run full-bore valgrind on it or wut
(trilema) asciilifeform: BingoBoingo: don't take my word for it, either. 'valgrind.'
(trilema) mircea_popescu: asciilifeform neat. this the pc valgrind ?
(trilema) asciilifeform: mircea_popescu: no restraints on that box, just valgrind
(trilema) mircea_popescu: davout afaik it's a pc with memory restraints and valgrind
(trilema) asciilifeform: in my 'valgrind' experiments, i've found that it is responsible for virtually all growth in footprint since powerup
(trilema) assbot: dpaste: 231ZHHC: valgrind output of bastard node. ... ( http://bit.ly/1Die6yC )
(trilema) asciilifeform: bastard node is running under 'valgrind', incidentally
(trilema) thestringpuller: they'll just run valgrind until it tells them its okay
(trilema) mod6: i'll have to look back. i spent a bunch of time running gdb & valgrind earlier in the month.
(trilema) assbot: 26 results for 'valgrind' : http://s.b-a.link/?q=valgrind
(trilema) asciilifeform: mod6: a few days ago, I had been looking mapOrphanBlocks as a possible reason that it runs itself out << see my earlier 'valgrind' experiments.
(trilema) thestringpuller: valgrind is the most amazing program ever made for debugging memory
(trilema) asciilifeform: valgrind, yes, as printed on the box
(trilema) decimation: heh did you run the whole thing in valgrind? must have been slow as hell
(trilema) Adlai: libccoin is a valgrind clean bitcoin node, written in C
(trilema) asciilifeform: here's another tidbit from 'valgrind' runs - leak is very slow prior to wedge, then appears to speed up
(trilema) asciilifeform: mircea_popescu: if you want polished turd, buy a girl who knows cpp, 'valgrind', gdb, understands memory discipline, can implement bdb replacement ab initio, knows ecdsa mechanics, etc, etc.
(trilema) asciilifeform: but if it turns out that i am the only one here who even knows how to operate 'valgrind', 'gdb', etc. - you're SOL, fellas.
(trilema) ben_vulpes: forgive my valgrind/c naivte, but i'm assuming that indirectly lost refers to memory that cannot be freed?
(trilema) assbot: Logged on 24-10-2014 20:23:36; mircea_popescu: nobody that knows how to use valgrind.
(trilema) asciilifeform: so, no one ever put the cursed thing through 'valgrind' or the like?
(trilema) asciilifeform: and valgrind is not a magic pill
(trilema) thestringpuller: asciilifeform: what's an alternative to Valgrind?
(trilema) asciilifeform: even something as pedestrian as 'valgrind' is laborious for a large turdatronic system.
(trilema) decimation: asciilifeform: it would be interesting to run turd through valgrind
(trilema) mircea_popescu: nobody that knows how to use valgrind.
(trilema) asciilifeform: anybody ever bother to valgrind?
(trilema) asciilifeform: valgrind << will run inside. it attaches to a process. if you want out-of-band (say, kernel diddling) both emulators present 'gdb' interface to the outside, stop cpu, play
(trilema) jurov: *valgrind
(trilema) BingoBoingo: ben_vulpes: Valgrind?
(trilema) BingoBoingo: Valgrind definitely has higher resolution than resedit for sure.
(trilema) BingoBoingo wonders if there is a 68k port of Valgrind
(trilema) decimation: asciilifeform: try running 'valgrind' on an application that requires high-bandwidth/speed performance
(trilema) asciilifeform: profilers << 'valgrind.'
(trilema) asciilifeform partial to similar 'valgrind'