<_ngn>
when a filesystem is mounted, is it a good idea to load the entire filesystem tree into the memory?
<geist>
not generally no, since it can be massive
<geist>
and that would take a long time
<geist>
except for maybe specific situations you pretty much always read what you need from a FS on demand
<_ngn>
geist: i see, thank you
<geist>
generally fs drivers when mounting search for the superblock/master block/etc (whatever defines the 'root' of the filesystem), validates that it's right, finds the root directory and then more or less hands that to the OS
<geist>
at that point the fs driver is basically waiting for calls to search directories and open files/etc
<geist>
it's an oversimplification, but the gist is you only read what you need as you need it
<_ngn>
so you only store information about the root of the filesystem, and you load other information about all the directories and stuff on demand?
<clever>
_ngn: yeah
edr has quit [Quit: Leaving]
qubasa has joined #osdev
<geist>
yep
<_ngn>
alright thank you guys
goliath has quit [Quit: SIGSEGV]
<geist>
sure thing!
x64S has quit [Quit: Leaving]
cloudowind has quit [Ping timeout: 260 seconds]
Arthuria has quit [Ping timeout: 248 seconds]
cloudowind has joined #osdev
agent314_ has joined #osdev
agent314 has quit [Ping timeout: 246 seconds]
<kof673>
well i mean you could load stuff from disk into RAM....but then it is either temporary or ...should be written back to disk at some point. either it is an ephemeral "ram disk" that disappears on shut down, or it gets written back to disk
<kof673>
you would still have to work out "caching" basically
<kof673>
either there is a save point or there is no save point lol
agent314_ has quit [Ping timeout: 260 seconds]
agent314_ has joined #osdev
GeDaMo has joined #osdev
GeDaMo has quit [Client Quit]
GeDaMo has joined #osdev
qubuepe24 has joined #osdev
sidcha has joined #osdev
<clever>
kof673: writing it back to disk only comes into play if the data in ram becomes dirty, so you could just read the entire fs into ram, but still issue writes on every write to files
<clever>
depends on how agressive you want your read-ahead to be
qubuepe24 has quit [Remote host closed the connection]
<kof673>
yeah, and if it is read-only and "small" it might make more sense
ThinkT510 has quit [Quit: WeeChat 4.4.3]
ThinkT510 has joined #osdev
pjals has joined #osdev
<kazinsal>
One of my more recent experiments (a couple years ago now -- shows how long I've been out of actually doing any osdev hackery) used a small ramdisk filesystem for critical logging that would flush to a specific contiguous reserved space at the beginning of the system partition regularly
<kazinsal>
In the event of a crash it could be read back live to the debugging console, dumped over serial, or if the system was "stable" enough for it, primitively blatted to disk using as few interrupts etc. as possible
cloudowind has quit [Ping timeout: 252 seconds]
netbsduser` has joined #osdev
cloudowind has joined #osdev
<kazinsal>
Part of the overall idea was being able to snapshot enough system state that the "supervisor" part of the kernel could be restarted on another CPU with minimal downtime and minimal loss of in-progress jobs, and having a persistent buffer would make that a bit easier (as well as improving kernel debugging)
<kazinsal>
(The idea was to write a paper about it. I eventually abandoned the idea after realizing that I had basically poorly reimplemented an OS-specific hypervisor from first principles.)
spare has joined #osdev
cloudowind has quit [Ping timeout: 276 seconds]
cloudowind has joined #osdev
zhiayang has quit [Quit: oof.]
zhiayang has joined #osdev
cloudowind has quit [Remote host closed the connection]
zhiayang has quit [Ping timeout: 244 seconds]
cloudowind has joined #osdev
zhiayang has joined #osdev
cloudowind has quit [Remote host closed the connection]
cloudowind has joined #osdev
pjals has quit [Ping timeout: 244 seconds]
hwpplayer1 has joined #osdev
FreeFull has joined #osdev
pjals has joined #osdev
hwpplayer1 has quit [Remote host closed the connection]
<kazinsal>
I thought today was the day that will live in infamy
<Ermine>
heat: what happened?
<heat>
<Ermine> day of the linux kernel desktop?
<heat>
<pog> that's tomorrow
<heat>
tomorrow is today
<Ermine>
tomorrow is tomorrow
<heat>
tomorrow was tomorrow, now it's today
<heat>
tomorrow it'll be yesterday
<heat>
and so on
<Ermine>
that's breaking space time continuum
zhiayang has joined #osdev
<Ermine>
anyway, should they merge a couple more wayland protocols (including colour management one), and everyone will be installing linux in no time
<heat>
ONE MORE WAYLAND PROTOCOL BRO PLEASE PLEASE JUST ONE MORE WAYLAND PROTOCOL BRO I SWEAR IT'LL FIX THE DESKTOP I SWEAR IT'LL FIX IT PLEASE ONE MORE WAYLAND PROTOCOL BRO
<zid`>
Good news though, they're doing that tomorrow
cow_ has joined #osdev
agent314_ has joined #osdev
cow has quit [Ping timeout: 252 seconds]
<heat>
i assume creating and merging a wayland protocol is a nasty mashup of a horrible committee-based design process plus all the pendatry of an open source code review
<zid`>
heat
<zid`>
why is my cpu insane
<heat>
is it intel? that would explain it
<zid`>
no
<zid`>
amd zen3
<zid`>
I changed my recursion to a divide
<zid`>
and now it's 5x slower
<heat>
lol what
<heat>
show coden
<zid`>
and that's including some fgets overhead, so it's probab like, 100x slower
foudfou has quit [Remote host closed the connection]
foudfou has joined #osdev
dalem has joined #osdev
dalem has quit [Ping timeout: 252 seconds]
liana has quit [Ping timeout: 260 seconds]
obrien has joined #osdev
spare has quit [Read error: Connection reset by peer]
heat_ has joined #osdev
heat has quit [Read error: Connection reset by peer]
bauen1 has quit [Ping timeout: 245 seconds]
cow_ is now known as cow
obrien has quit [Remote host closed the connection]
Left_Turn has joined #osdev
pjals has joined #osdev
bauen1 has joined #osdev
op has joined #osdev
heat has joined #osdev
heat_ has quit [Read error: Connection reset by peer]
melnary has quit [Remote host closed the connection]
melnary has joined #osdev
Turn_Left has joined #osdev
Turn_Left has quit [Remote host closed the connection]
Turn_Left has joined #osdev
Left_Turn has quit [Ping timeout: 276 seconds]
op has quit [Remote host closed the connection]
hwpplayer1 has joined #osdev
Left_Turn has joined #osdev
Turn_Left has quit [Ping timeout: 248 seconds]
_ngn- has joined #osdev
_ngn has quit [Ping timeout: 264 seconds]
Left_Turn has quit [Quit: Leaving]
Left_Turn has joined #osdev
SupUser has joined #osdev
Turn_Left has joined #osdev
Left_Turn has quit [Ping timeout: 276 seconds]
<heat>
vt-d is a mess
<heat>
classic intel
_ngn- is now known as _ngn
SupUser has quit [Ping timeout: 244 seconds]
pjals has quit [Ping timeout: 244 seconds]
qubasa has quit [Remote host closed the connection]
<heat>
finished taking a quick look at this pdf, and i found: Intel(r) 64, 50 quadrillion abbreviations, page tables two electric boogaloo and, the best of all, ACPI tables galore
<heat>
The Device Scope Structure is made up of Device Scope Entries
<heat>
YES LETS GO
<heat>
i'm sure i've completed the intel bingo twice just now
<nikolar>
What was amd's thing
<nikolar>
For virtualization
<nikolar>
I forgor
<heat>
amd-vi
<heat>
well to be clear vt-d is the intel iommu stuff, not strictly virtualization
craigo has joined #osdev
<nikolar>
Right fair enough
xenos1984 has quit [Ping timeout: 246 seconds]
Left_Turn has joined #osdev
xenos1984 has joined #osdev
Turn_Left has quit [Ping timeout: 246 seconds]
SupUser has joined #osdev
Arthuria has joined #osdev
bauen1 has quit [Ping timeout: 248 seconds]
xenos1984 has quit [Ping timeout: 246 seconds]
xenos1984 has joined #osdev
op has joined #osdev
<Ermine>
is vt-x virtualization?
<CompanionCube>
yes
<heat>
vt-x, vt-d, vt-g, vt-c
<heat>
these are legitimately intel feature names
<nikolar>
Nice
<zid`>
I have
<zid`>
avx
<ring0_starr>
you'd swear there's some kind of ethereal force trying to make intel look/act stupid
<heat>
zid` do you have avx512jadujanwj
<zid`>
no, I'm not a cool laptop :(
<heat>
dont forget amd-vi named after that editor that no one knows how to use
<ring0_starr>
it goes deeper than feature names
<ring0_starr>
how about the new "ultra" branding
<ring0_starr>
killing off all kinds of development projects for no real reason
<zid`>
I also have sysret_ss_attrs null_seg spectre_v1 spectre_v2 spec_store_bypass srso
<zid`>
suck it
<heat>
dude
<heat>
i have like 5x that
<ring0_starr>
i was interviewing at intel right around the time they killed knights corner, the atmosphere was really miserable
<ring0_starr>
so imagine all these other things that gelsinger murdered
<ring0_starr>
they make their employees wear rainbow badges
<heat>
kabylake is era appropriate for "full of speculation bugz" and not too new for "we found that shit and posted a paper with a funny name"
<heat>
RAINBOW BADGES?
<ring0_starr>
yes
<heat>
WOKE INTEL
<zid`>
I don't have as many pokemon gym badges as you heat :(
<ring0_starr>
i know it sounds like i'm making it up
<heat>
VERSUS BASED RIGHT-WING 4CHAN LOVING AMD
<zid`>
amd beats a minority for you with every purchase
<ring0_starr>
overvolting their newest chips for higher clock speeds might not be a dumb decision
<heat>
jensen drops a bomb in ukraine for every RTX4000 they sell
<heat>
like, personally
<ring0_starr>
it might cut the life span of something that could theoretically last many decades down to a couple years, but that's nbd, tech is so replaceable these days that it's practically throw-away like fast fashion
<zid`>
except heat scales with like the 10th power of voltage
<zid`>
so now they're 350W chips running the exact same silicon from 5 years ago
<heat>
i dont scale
<zid`>
I could have just bought a 2011 and overcl- oh wait that's exactly what I did
<the_oz>
you couldn';t pay me enough to stand there jeeeeez
<the_oz>
why did the chicken cross the road
<the_oz>
to get away from this craziness!
<heat>
they were watching a group b really and a car was coming
<heat>
rally
<the_oz>
lol
<heat>
i do have to wonder how anyone actually starts rallying
<heat>
like
<heat>
what teen picks a car up and thinks that hmmm maybe driving really fast in the middle of the trees and shit is a genius idea
<nikolar>
really car
<heat>
this isn't even CTE
<heat>
if you're a rally driver you're probably actually insane
<nikolar>
kek
<nikolar>
fair assesment
<the_oz>
Competition?
<Ermine>
colin mcray rally 2.0 is my first vide game
<Ermine>
not counting windows xp games
<heat>
this isn't a fucking track with tons of safety stuff, it's a forest, if you're 50cm out, you're dead
<heat>
all-in-all, good career prospects
<nikolar>
probably not dead realistically
<nikolar>
like the whole car is built like a tank
<heat>
definitely dead
<nikolar>
but it certainly wouldn't be pleasant
<Ermine>
yes, I've been destroying my cars by the end of the first track
<heat>
the car isn't that tough and you're going reaaaaaaaaaaaaaally fast through some very tight sections
<Ermine>
actually, some tracks in this game happen on the actual roads
<heat>
yep
<Ermine>
but those roads are in mountains, so turns are very steep
<heat>
take a random rally game, realize how easy it is to fucking die, but imagine it being permanent
<heat>
not to mention random stuff like mechanical failures
<Ermine>
like 140° to 180°
<heat>
if you fuck up a bump you might be going 120km flying into a tree
<heat>
you'll be human soup
<Ermine>
I had to go to an actual driving school to get an idea how to go through this
<the_oz>
don't worry, you'll have a passenger driver to tell you how badly you fucked up moments before you both die
<nikolar>
Ermine: did you never figure out how to play the game before that
<Ermine>
no
Turn_Left has joined #osdev
Left_Turn has quit [Ping timeout: 246 seconds]
Turn_Left has quit [Read error: Connection reset by peer]
zhiayang has quit [Ping timeout: 248 seconds]
op has quit [Remote host closed the connection]
npc has joined #osdev
<heat>
i have really important things to do but decided to just move code around for the day
<nikolar>
relatable
<nikolar>
i usually mess with my makefiles and such when i procrastinate lol
Celelibi has quit [Ping timeout: 265 seconds]
vdamewood has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
Celelibi has joined #osdev
vdamewood has joined #osdev
<geist>
heat: yeah i've found the more code you write on a project, the larger the project is, the more miscellaneous code-moving-around tasks you can do on any given day
<geist>
so it's easy to just do that instead of 'real work'
<geist>
though you can argue taht a lot of those tasks are also important, and i think that's precisely how large software teams at large companies always find work to do, the amount of miscellaneous 'surface area' of the code grows sort of exponentially and tere's always something to move around
<the_oz>
*runs around in factorio*
<the_oz>
decision paralysis
agent314_ has quit [Ping timeout: 248 seconds]
<geist>
yah, and even if you're decisive about what you do on any given time, there's whether or not what you're doing is te best thing to be doing
<geist>
it's so easy to pick a series of 30 minute projects for days, weeks, months
<the_oz>
I don't need to expand iron lines, I'll juat redo the fuel lines again to be even more efficient
<geist>
ummhmm yep
npc has quit [Remote host closed the connection]
<geist>
i got into that with satisfactory until i decided to just go ahead and push to win so i can put it down
<the_oz>
undervolting killed satisfactory for me, but that was before swooping bsack in like 0.4 or 0.5