bradd has quit [Remote host closed the connection]
Matt|home has joined #osdev
Matt|home has quit [Remote host closed the connection]
Matt|home has joined #osdev
bradd has joined #osdev
vdamewood has joined #osdev
pog has quit [Ping timeout: 260 seconds]
goliath has quit [Quit: SIGSEGV]
gbowne1 has quit [Read error: Connection reset by peer]
GeDaMo has joined #osdev
Matt|home has quit [Quit: Leaving]
pog has joined #osdev
mantahydra13 has joined #osdev
mantahydra1 has quit [Ping timeout: 260 seconds]
mantahydra13 is now known as mantahydra1
jack_rabbit has quit [Ping timeout: 256 seconds]
vdamewood has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
vdamewood has joined #osdev
jack_rabbit has joined #osdev
nortti has quit [Remote host closed the connection]
pallas has joined #osdev
kfv has joined #osdev
kfv has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
elderK has joined #osdev
<kazinsal>
hey for anyone who's awake, do we have anyone with electronics knowledge here? need to figure out if something is safe to short or not
<Mutabah>
Fire away?
<Mutabah>
I mean, ask - I may know
<kazinsal>
GND ping of a 74xx chip is attached to the GND rail of the PCB it's on by a now blown 4.7uF tantalum capacitor
<kazinsal>
I'm not sure what the purpose of that cap is so would it be theoretically safe to just bodge wire the two pins together until I have a replacement cap to solder in?
<Mutabah>
Strange... a cap between two grounds is odd
<kazinsal>
Yeah, that's what's confusing me
<Mutabah>
A cap between ground and power makes sense, to absorb transients
<kazinsal>
It's the riser board for an IBM PS/2 floppy controller -- the controller is on the motherboard and the logic chips on the riser are to select which of the three drive connectors on the riser to work with at any given point in time
<kazinsal>
But I compared it against some other chips and indeed there's continuity between the ground rail and the ground pins on other chips as well as the input leg of that blown cap
<kazinsal>
Or, should be
<kazinsal>
So it's definitely a cap on two grounds
<kazinsal>
I just can't figure out *why*
<kazinsal>
Which is why I wonder if it's safe to bodge
<Mutabah>
It's _probably_ good
<Mutabah>
Couldbe something funky with negative rails
<Mutabah>
(Negative voltages) - but even then, it should be safe
<kazinsal>
Right, thanks. That's what I figured, I just wanted to get another opinion
<Mutabah>
hmm... 4.7uF isn't too small
<kazinsal>
The problem is I'm *thinking* it's a 4.7uF
<kazinsal>
I can't confirm because the bit with the markings is no longer a solid
<kazinsal>
So based on the pinouts of the 40pin connectors and the silkscreening on the other side of the board, the bottom left pin that you can see the rear of in that photo of the edge connector directly to the right of that 74125 is pin 39 of the 40-pin connector
<kazinsal>
And the pinout doc in the tech ref for that one says every odd numbered pin is ground
<kazinsal>
So now I'm even more confused
Cindy has joined #osdev
<GeDaMo>
I see a lot of stuff online about people making adaptors for standard floppies to use in the PS/2
<kazinsal>
Yeah, that's for the actual drives themselves
<kazinsal>
This is the riser board from the motherboard to the drive ports
<kazinsal>
The drives plug into this board, this board then plugs into the motherboard
<kazinsal>
Or "planar" as IBM insisted on calling it
<kazinsal>
This appears to be one of the few parts of the Model 50 that nobody has ever documented.
<kazinsal>
Which means of COURSE it's the one part that I need to repair.
<kazinsal>
Even the bloody 40 year old ST506 drive still works.
<kazinsal>
> : (
<kazinsal>
But I can't set the CMOS options without the floppy drive working to boot the refdisk
<kazinsal>
Okay, this might be a cap between +12VDC and GND
<kazinsal>
I need to go take some damn courses or something before I try to fix this thing
<Mutabah>
If it's to 12V, then sounds like a decoupling cap - absorbs small fluctuations in input power
<Mutabah>
those are usually safe to remove, but it might do odd things
<kazinsal>
Probably worth scrounging through other non-working machines I have for a cap matching the voltage and capacitance
mavhq has joined #osdev
GeDaMo has quit [Remote host closed the connection]
<kazinsal>
Christ. Looking at this whole thing and tracing some more +12VDC leads this cap does double duty as a decoupling cap and the negative leg is just a straight through from the GND of that 74LS125.
<kazinsal>
Goddammit, IBM, your parts saving measures are making my life miserable
Goodbye_Vincent has quit [Ping timeout: 240 seconds]
<GeDaMo>
Damned if I know :P
Goodbye_Vincent has joined #osdev
<zid>
maybe your initials are GDM and you're an emo
<zid>
Gordon Damian McScoot
Goodbye_Vincent has quit [Ping timeout: 252 seconds]
<sham1>
M68K
Goodbye_Vincent has joined #osdev
Goodbye_Vincent has quit [Ping timeout: 264 seconds]
<nikolar>
PDP
voidah has quit [Ping timeout: 268 seconds]
Goodbye_Vincent has joined #osdev
Goodbye_Vincent has quit [Quit: Ping timeout (120 seconds)]
<geist>
PAAAAAARISC
<geist>
honestly i got kinda interested in doing some kernel sparc a few months ago
<mjg>
:)
<geist>
and then i re-read all the nonsens about the register window and got kinda uninterested pretty quickly
<mjg>
there is some genuinely offputting to me about sparc assembly though
<mjg>
right
<geist>
yah, user assembly wise it just seems to be a fairly bog standard risc machine, the register window is meh
<mjg>
and delay slots
<mjg>
maybe i would do some sparc if i was a NERD
<geist>
system wise the whole needing to take exceptions and how it affects the way register frames are laid out, etc seems exceedingly complex
<mjg>
what do you think about software telb
<mjg>
tlb
<geist>
oh i think that's probably kinda fun
<mjg>
o'rly?
<geist>
lets you do your page tables, etc howeve ryo uwant. the first cpu i ever ported to other than x86 nback in the day was SH4, which also is software TLB
<geist>
and i did some hackery where i baically continued to use x86, then wrote some asm routine in the TLB handler to fill from the page table
<geist>
which worked reasonably well
<geist>
not gonna say it was necessarily fast or whatnot, but it was reasonably interesting
<mjg>
i thought everyone complains about tlb miss costs on that sucker
<mjg>
right
<geist>
sure
<geist>
but you know, speed isn't everything
<mjg>
la la la
<mjg>
can't hear you
<mjg>
so shit talk aside, what would be an interest arch to write for today
<geist>
hmm
<mjg>
no matter how vintage or otherwise
<mjg>
i guess by interesting in this context i mean as different from x86 as it gets
<mjg>
but not annoying
<geist>
yeah. hrm
<nortti>
at&t hobbit, maybe?
<mjg>
huh i have to say i never heard of it
<netbsduser>
risc doesn't really vibe with me
<netbsduser>
that's why i picked m68k as the 2nd port for my kernel
GeDaMo has quit [Quit: That's it, you people have stood in my way long enough! I'm going to clown college!]
<zid>
pog: My spider keeps running out of electricity :(
<nikolar>
I kinda want to make sparc kernel
hoshi has quit [Ping timeout: 276 seconds]
hoshi has joined #osdev
Goodbye_Vincent has joined #osdev
gbowne1 has joined #osdev
alexander has quit [Read error: Connection reset by peer]
rb has quit [Ping timeout: 264 seconds]
pog is now known as gog
<gog>
poggie goggies
vdamewood has joined #osdev
rb has joined #osdev
alexander has joined #osdev
<sham1>
goggers
<sham1>
gog: may I pet you
netbsduser has quit [Ping timeout: 268 seconds]
loosipher has joined #osdev
Matt|home has joined #osdev
Arthuria has joined #osdev
<kof123>
> at&t hobbit, maybe? wasn't there a person who sometimes shows up here that was writing an emulator?
<nortti>
puck?
<puck>
meow
<puck>
yes
<zid>
puckers
* mjg
pets gog
<sham1>
meow to you too
<mjg>
long time no pets
<kof123>
so wait...if the beos ppc was code-warrior-ish IIRC, x86 was gcc, where did the hobbit beos c compiler come from?
<puck>
it's hch. sec
<puck>
MetaWare High C
<kof123>
ok thanks. > the first cpu i ever ported to other than x86 nback in the day was SH4 send me your dreamcast code ppl, i can test
<zid>
yea the co-coroutine/generators sare the thing I am most jealous of
<bslsk05>
gist.github.com: ftest.c · GitHub
<puck>
i don't see => tho
<bslsk05>
gist.github.com: ftest.c · GitHub
<bslsk05>
gist.github.com: ftest.c · GitHub
<puck>
yup, named parameters work
pretty_dumm_guy has quit [Quit: WeeChat 3.5]
<puck>
nortti: okay, so i was a biiiit worried but it looks like there in fact do exist non-hobbit versions of this era of High C
Goodbye_Vincent has quit [Ping timeout: 260 seconds]
frkazoid333 has quit [Ping timeout: 260 seconds]
jack_rabbit has quit [Ping timeout: 246 seconds]
frkazoid333 has joined #osdev
Goodbye_Vincent has joined #osdev
<heat_>
geist, nah i think .. is still an extension
<heat_>
more commonly GNU
<geist>
well, looks like you have to get use from metaware to do anything with it anyway
<heat_>
hm?
<geist>
the code pasted by puck into pastebin
<geist>
the copyright
<heat_>
i dont know what any of that is
<geist>
well i wasn't talking to you heat!
<heat_>
dang
<heat_>
sorry just woke up
<geist>
heh
* geist
gets out the slapping trout
<puck>
geist: these disk images are out there :p
<heat_>
it's really funny they do #if 0 for the initial comments
frkazoid333 has quit [Ping timeout: 256 seconds]
<puck>
wait. does hobbit not have alloca
<puck>
huh. Huh. i'm not even sure you *can* implement alloca
<geist>
hmm, how not?
<puck>
geist: there's no registers
<puck>
everything is stack
<geist>
ah it's push/push/operate/pop based?
* gog
prr
<puck>
no, it's weirder
<puck>
lemme grab the list of addressing modes
<heat_>
gog: solaris sparc kernie
<gog>
alpha
<geist>
i remember hobbit being weird,but dont remember any details
<heat_>
vaxen
<heat_>
yesterday i was reading an al viro email and he said vaxen
<heat_>
sorry, not vaxen, boxen
<puck>
geist: you have absolute, immediate, stack offset, stack offset indirect
<heat_>
linux does not support vaxen
<geist>
i see
<geist>
so not a fully stack based arch, but kinda some hybrid
<puck>
geist: to e.g. add two locals together, you do add r4, r8; which is *(sp + 8 bytes) += *(sp + 4 bytes)
<geist>
ah, hmm
<puck>
the issue with alloca here is it'd require basically moving all your locals to *above* your new alloca'd buffer, or doing self-rewriting code to adjust stackoffsets
<geist>
so you can kinda think of it as having registers, they're just relative to a moving pointer
<puck>
yeaoh
<geist>
since it's not push/pop based like smoe stack arches
<puck>
in fact, r4294967295 would be a valid "register"
<geist>
ie, add doesn't pop two things and push the result sort of thing
<CompanionCube>
'register windows' but literally
<geist>
yah
<puck>
i think the reason AT&T designed it this way is now it's super easy to create stack frames
<puck>
each C variable just gets assigned a "register"
<geist>
nortti: probably one of the reasons linux-vax still exists in gcc as a valid triple
<geist>
or at least did a few years back
<heat_>
vaxen boxen
<geist>
puck: yah i remember that being the talk about it, it was all about being easy for C
<sham1>
linuxen on vax boxen
<puck>
geist: i'm so glad they called it a CRISP
<puck>
(C reduced instruction set processor)
<geist>
but yeah you're right, alloca would be really difficult, seems like you'd need to compile code that either remembers some sort of anchor pre-alloca or actually literally copy locals after it, or something
<sham1>
That's annoyingly close to CRISPR
<nortti>
puck: do you get shorter encodings for smaller numbers, or does every "register reference" take up 4-bytes?
<puck>
sham1: C Reduced Instruction Set PRocessor
<puck>
nortti: former
<nortti>
makes sense
<heat_>
omg linuxen vaxen boxen
<Ermine>
wheren?
<nortti>
I think if I were implementing alloca on hobbit I'd make any function using alloca have a special prologue and epilogue that set up a pointer in some other buffer, and use that for alloca
<sham1>
Müss Linuxen geboxen
<puck>
nortti: e.g. add3 can be represented as wai5/stk5, imm5/stk5, stk5/stk5 (all 2 bytes), gen16/gen16 (6 bytes), and gen32/gen32 (10 bytes)
<sham1>
I mean, alloca would require some kind of a special boi buffer
<sham1>
And epilogues
<gog>
i put my veggies in the crispr and then they come out all mutated
<puck>
nortti: where wai is "word aligned", stk is "stack offset, operand size of word"; and gen are generic (16/32 bit)
<puck>
word aligned immediate, even
<heat_>
gog, you put your veggxen in the crisperen and they camen outen all mutateden??
<gog>
no
voidah has joined #osdev
<heat_>
crappen
MrBonkers has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]