<jrtc27>
my concern with dropping @plt is that binutils will do the wrong thing
<jrtc27>
(for -fno-integrated-as)
<jrtc27>
if we drop it then we need to explicitly document the option as unsupported until binutils is altered
paulio_uk has joined #riscv
jmdaemon has quit [Ping timeout: 268 seconds]
ksbedfordjr has joined #riscv
<meowray>
jrtc27: it seems that gas has been assembling call foo to R_RISCV_CALL_PLT for some time (2022-09 70f35d72ef04cd23771875c1661c9975044a749c). -fno-integrated-as with old gas will emit R_RISCV_CALL, but that seems fine as well? (i don't remember what the R_RISCV_CALL/R_RISCV_CALL_PLT discrepancy was in older gas)
<jrtc27>
R_RISCV_CALL just gets a bit weird in GNU ld IIRC
<jrtc27>
it has strange semantics for dynamic linking when mixed with R_RISCV_CALL_PLT, or something
<jrtc27>
but if gas assembles it to R_RISCV_CALL_PLT then my concerns are dealt with
<meowray>
iirc the R_RISCV_CALL difference in gnu ld was related to undefined weak symbol. now i look at bfd/elfnn-riscv.c again, i cannot find any difference
<jrtc27>
maybe it finally got fixed
<jrtc27>
some time last year when I checked it was still dodgy
<jrtc27>
anyway, approved now, sorry for not realising that bit of information and thus delaying the PR
<ksbedfordjr>
I was sent here from discord riscv chat abotu getting help to learn to conver dts files from 5.10 to 6.6.y stable Mainline kernel
<ksbedfordjr>
is there any one that can help
Tenkawa has quit [Quit: Was I really ever here?]
Tenkawa has joined #riscv
<ksbedfordjr>
There is one of them from the riscv discord
crabbedhaloablut has quit []
crabbedhaloablut has joined #riscv
BootLayer has quit [Quit: Leaving]
<ksbedfordjr>
so as I was saying the unit I am working with is the milkv-cm
zBeeble42 is now known as zBeeble
<ksbedfordjr>
and I need to bring it upto 6.6.x stable so that it can be added into mainline
<ksbedfordjr>
but I have never done a dts/dtsi
<sorear>
the device trees are supposed to be forward compatible, but the whole area is a disaster. maybe look at some of the dts files that exist in both 5.10 and 6.6 and see how they changed?
<ksbedfordjr>
and the ones I am looking at compared to whats in ml kernel dont match
<sorear>
are you working with upstream 5.10 or a milkv-specific fork? if it's a milkv fork, is there anything in there that you would need except for the dts?
<ksbedfordjr>
well there is alot in the cm module thats not in the vf2
<ksbedfordjr>
thats the duo board not the mars/ mars-cm
<mps>
that is what I see in git log
<ksbedfordjr>
yeah thats the smallerchips
<ksbedfordjr>
so its not the board/chip I am working with
<ksbedfordjr>
I pre ordered my rv-mv-duo 512 and the dock
mlw has quit [Ping timeout: 252 seconds]
<mps>
I looked only on Pioneer because Alpine Linux should get this machine in near future, I hope
<ksbedfordjr>
thats the 64 core board
<ksbedfordjr>
1,499 for board
<ksbedfordjr>
2,599 for full build wow
<mps>
yes
<ksbedfordjr>
I will have to have the UNIV look into 1 for the lab
<ksbedfordjr>
and once they do it will become a inhouse pkg builder/repo mirror
<ksbedfordjr>
thats also whay I would use to go back to the days of xterms and a server
<ksbedfordjr>
why did those days die off
<ksbedfordjr>
it was nice to keep things in 1 place and have xterms where you did not have to maintain a hd
<ksbedfordjr>
NCD xterms
<ksbedfordjr>
big ole flat pizza box units with monitors
davidlt has quit [Ping timeout: 252 seconds]
<mps>
on UNI that makes sense
s1b1 has joined #riscv
<ksbedfordjr>
when I was younger my uncle had a 10 ncd 700 terminal setup . and we all had a terminal in each house as we all lived with backyards to each other so he wired the houses together
<ksbedfordjr>
but thats how we got on the internet and did school work
<mps>
heh, old good times
<ksbedfordjr>
but we had sound and we could chat each other house to house with a console app he had called chatterbox
<mps>
RS485 network?
<ksbedfordjr>
so if it was to cold or raining out we cold still talk with out tying up the phone line
<ksbedfordjr>
no it was a coax 10 base 10 as I recall
<ksbedfordjr>
10 base T
<mps>
ah, then faster than RS485
<ksbedfordjr>
I will ask him when I call next week
<ksbedfordjr>
yeah it was burried in garden hose through the yard
<ksbedfordjr>
to keep it dry
<mps>
sure
<ksbedfordjr>
but I remeber the coax cables and the end jumers on the t at the card
<mps>
coax are sensitive to humidity
<ksbedfordjr>
but it was cool to learn on those
<ksbedfordjr>
thast what made me get into computers
<ksbedfordjr>
I would like to make a bunch of these board do the same thign
<ksbedfordjr>
where they net boot and everything is on a main server
<mps>
yes, yes, learning foundations of technology is very good for furter studiying and work
<ksbedfordjr>
so have you seen the 2 beaglev riscv boards now out
<ksbedfordjr>
rather intresting
<mps>
I think I noticed something about on some channels but forgot where
<ksbedfordjr>
but it seems the support for the beaglev fire has not gotten into the kernel
<mps>
new boards need time to be 'included' in mainline kernel
<sorear>
I have to wonder if the ground the students were connected to was the same as the circuit ground on the boards
justache has joined #riscv
<muurkha>
that's a reasonable concern, but normally static bands have a megohm or two in series so they don't pose an electrocution hazard
<muurkha>
I'd think that would be enough to keep current injection minimal at any plausible ground-loop voltage?
<muurkha>
I guess it's a different story if you are walking around with the board charged up to 200 volts through the ground loop and then plug it into the circuit ground, but that's just an HBI ESD on every pin at once, isn't it?
<sorear>
MOSFET gate oxides tolerate substantially less current than human hearts, a megohm might be enough for one but not the other if there's no adequate ESD protection on the board...
frkazoid333 has joined #riscv
<sorear>
what's a HBI?
vagrantc has quit [Quit: leaving]
frkzoid has joined #riscv
<muurkha>
sorry, I meant HBM, human body model
<muurkha>
it's been a long time since I've seen a chip that wasn't rated for a few thousand volts of HBM ESD; clamping diodes on I/O pins were already common on TTL in the 01970s
frkazoid333 has quit [Ping timeout: 252 seconds]
<muurkha>
and as far as I can tell nobody sells signal MOSFETs any more because they're too static-sensitive
<muurkha>
but it's definitely a thing Buffalo Lab (?) could have screwed up
crossdev has quit [Remote host closed the connection]
<sorear>
i've seen it on very-high-speed ADCs and similar where you are acutely aware that every off-state diode is a capacitor, but the more i think about it the less likely it seems here