DDraceNetwork - developer

DDraceNetwork

Development / developer

Development discussion. Logged to https://ddnet.tw/irclogs/ Connected with DDNet's IRC channel, Matrix room and GitHub repositories — IRC: #ddnet on Quakenet | Matrix: #ddnet-developer:matrix.org GitHub: https://github.com/ddnet

Between 2023-01-26 00:00:00Z and 2023-01-27 00:00:00Z

[ddnet-maps:master] 1 new commit

e03081c A DonoWall - ddnet-maps

ChillerDragon

Yo bois I did my first run through documenting all system messages of the 0.7 protocol any feedback to the layout or content? Suggestions? https://chillerdragon.github.io/teeworlds-protocol/07/system_messages.html

nice documentation chillerdragon

Sänks :)

looks nice

07:34

if it was for the 0.6 it would be cool to have it on the wiki

07:35

style wise, i think u are missing some margins in places, like after tables xd

It is linked on the wiki twice :D Most of the 0.7 protocol is relevant/used in ddnet since it’s so close to 0.6But yea you are right. I sometimes do a Tag after tables and sometimes not. I think I put margin: 0 on everything. Good catch will probably fix that later thanks!

07:41

But yea if I manage to finish one day I’ll continue with 0.6 then ddnet specific extensions

yes, very interesting documentation

Faulty

@deen I'm not familiar with your workflow yet, do I need to do anything else with my PR, or should I wait for approval from someone?

wait for approval, basically

1

This is a reply to some rust stuff right? (@heinrich5991)

heinrich5991#9827

wait for approval, basically (@Faulty)

chillerdragon

This is a reply to some rust stuff right? (@heinrich5991)

No. It was a reply to a new dev asking if they need to do anything more on their PR

o thank

Learath2

No. It was a reply to a new dev asking if they need to do anything more on their PR

btw, could you reply pls in PR thread when you have time (https://github.com/ddnet/ddnet/pull/6293)

1

Image attachment

10:59

How i can delete this options on my server

In your config file? Maybe?

@Jupstar ✪ did u check if its not reported or if its the case that its reported using the 0.7 protocol?

11:19

the layout changes if(!Server()->IsSixup(ID)) depending on this (edited)

11:19

on a per player basis

11:20

if(!m_apPlayers[ID]) return there is also this check, which should always pass i guess xd (edited)

I've no idea. I added it where the other json stuff was parsed.. so I assume that it's already check if it's 0.6

no it doesnt check

Return sounds like it must be continue

u gotta check it urself

Ryozuki

no it doesnt check

Mh weird, so it randomly parses 0.6 and 00.7 stuff in the client?

11:21

Thought 0.7 registry is a complete new entry

game/server/gamecontext.cpp:4315 for the json creation

lalala soytari 卍

Click to see attachment 🖼️

go to the autoexec_server.cfg and Search Line Vote Delete This Code In CFG

@Jupstar ✪ it doesnt randomly parse it, it just that the server reports a json object however it wants to the master

11:22

and the ddnet server parse it

11:22

i guess

11:22

cc @heinrich5991

11:23

u can check if the skin is 0.6 by checking for the existance of the color_body key iirc

11:23

also a player may be 0.6 and another 0.7

11:23

i would just display a default tee for 0.7

11:24

for now

11:25

another thing may be that the buffer is too small for the data

11:25

i havent rly calculated exactly

11:26

each player has a buffer of 512 bytes char aExtraPlayerInfo[512];

11:26

total for server info is 16384

Mh yeah I dunno. If that's the case I'd expect that the master server warns about it

11:27

But maybe our impl doesn't.

yeah i would guess the master would reject invalid json

11:27

which probs does if it uses serde

11:28

anyway, the server is free to report whathever it wants

And then it uses vanilla fetching instead, which obv has no skins?

within a size limit idk which one is rn

11:28

idk, i think it fetches with json from the server, and if its invalid its vanilla?

11:29

heinrich should know

[ddnet-web:master] 1 new commit

d6f3a17 Frontpage video - def-

[ddnet/ddnet] Pull request closed: #6309 Add buttons to move envelopes left and right, improve envelope UI layout, add tooltips for editboxes

Span<int> numbers = stackalloc int[length];

14:35

till u can stack allocate on c#

14:35

i wonder if u can do that in java rn

Ryozuki

heinrich should know

I don't quite get what you're asking

14:37

if the server doesn't provide valid json, the register request is rejected

@heinrich5991 if the server reports invalid json to the master, does it instead fetch using the vanilla protocol

only accidentally

do u have logs of this

14:37

specifically from ddnet official servers

14:37

if they reported invalid json

14:38

would be useful to know if this is the cause

not sure, probably not

the problem is apparentl sometimes skins arent reported

14:38

according to jupstar

you can check whether it's vanilla server info

14:39

by checking e.g. for the presence of the map sha256

14:39

ah yea

14:39

you could do that for the historic server info

14:39

I should make sure to never request legacy server info for servers registering on https masters

Everyone can use /r how can I fix this

[ddnet/ddnet] Issue opened: #6310 Select Persian language on first start = Crash

Reported by Quick, I can reproduce, will fix: ``` ------------------------------------- Translated Report (Full Report Below) ------------------------------------- Process: DDNet [33423] Path: /Users/USER/*/DDNet Identifier: DDNet Version: ??? Code Type: ARM-64 (Native) Parent Process: zsh [33321] Responsible: iTerm2 [999] User ID: 501 Date/Time: 2023-01-26 16:32:23....

15:37

[ddnet/ddnet] Pull request opened: #6311 Fix Persian language crash (fixes #6310)

Checklist

[x] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-ad...

https://www.calligrapher.ai/

Image attachment

Calligrapher.ai: Realistic computer-generated handwriting

Convert text to handwriting using an in-browser recurrent neural network. Choose from various print and cursive styles. Customize the outputs and download as SVG.

15:38

this looks rly gud

lalala soytari 卍

Everyone can use /r how can I fix this

https://ddnet.org/settingscommands/ sv_practice 0 in your server config

does someone know a good buffer ringbuffer library for rust?

15:47

I want to have a 2048 byte buffer

15:47

and read network packets into it

hmm

15:47

stack?

and then remove them again and pass them to the application

iirc a vecdeque can do

it seems I can't get a reference to the uninitialized part with vecdeque

15:48

which I need for reading from the network

Vecdequeue is a growable Ringbuffer

15:49

But I assume u mean static sized

dynamic is fine

15:50

but I need access to the uninit parts so I can pass it to a recv function

15:50

let read: usize = recv(ringbuffer.uninit_part()); ringbuffer.assume_written(read);

15:51

I suppose I could write it myself if I can't find a library

I also didn't find a multiset like cpp offers

what's a multiset?

15:52

ah, hashmap<something, int>?

A set that allows dublipcate keys

15:52

But sorted

hmm what about a vec of maybeuninit that u transmute

15:53

or smth

15:53

https://doc.rust-lang.org/nightly/std/mem/union.MaybeUninit.html#initializing-an-array-element-by-element

MaybeUninit in std::mem - Rust

A wrapper type to construct uninitialized instances of T.

https://doc.rust-lang.org/std/vec/struct.Vec.html#method.spare_capacity_mut

Vec in std::vec - Rust

A contiguous growable array type, written as Vec, short for ‘vector’.

15:53

vec allows access to uninitialized parts

and vecdeque not?

not as far as I can see

Saddest

so ima roll my own I guess

Ez

@heinrich5991 doesnt recv accept a writer?

15:56

isnt that better

recv usually doesn't accept a writer

I can give u cpp version of u want to port

gimme

In 45 min I'm home

@Ryozuki https://doc.rust-lang.org/std/net/struct.TcpStream.html#method.read-1

TcpStream in std::net - Rust

A TCP stream between a local and a remote socket.

Or i give u access to my gitlab

secret jupstar projects?

15:57

sign me up

Yes xd

15:58

Give me 5min

user is https://gitlab.com/heinrich5991

heinrich5991 · GitLab

GitLab.com

It's self hosted. I'll pm u

is it necessary to separate proc-macro crates?

16:00

like i can't do this in one file

16:00

can i?

[ddnet:master] 2 new commits

7d984ec Fix Persian language crash (fixes #6310) - def- fbc3d9b Merge #6311 - bors[bot]

16:02

[ddnet/ddnet] Issue closed: #6310 Select Persian language on first start = Crash

16:02

[ddnet/ddnet] Pull request closed: #6311 Fix Persian language crash (fixes #6310)

heinrich5991

let read: usize = recv(ringbuffer.uninit_part()); ringbuffer.assume_written(read);

correct me if im wrong, but once the ringbuffer is filled, u wouldnt get more data

16:04

what u trying to make exactly?

16:04

u can always have the vecdeque, and a stack array where u recv from and then u push it to the deque

16:06

https://github.com/rust-lang/rust/pull/102689

Add a tier 3 target for the Sony PlayStation 1 by ayrtonm · Pull Re...

This adds a tier 3 target, mipsel-sony-psx, for the Sony PlayStation 1. I've tested it pretty thoroughly with this SDK I wrote for it. From the tier 3 target policy (I've omitted the subpoi...

16:06

next release

Jupstar ✪

Or i give u access to my gitlab

why secret

16:11

are u hiding the next gen graphic algorithms? xd

In the past I wrote a game engine

16:12

Now I'm free and believe in open source

16:12

But i consider it educational anyway (edited)

16:12

Even if I wrote few games xd

https://blog.rust-lang.org/2023/01/26/Rust-1.67.0.html

Announcing Rust 1.67.0 | Rust Blog

Empowering everyone to build reliable and efficient software.

Ryozuki

https://blog.rust-lang.org/2023/01/26/Rust-1.67.0.html

I hope some nightly features are in xd

16:19

I use way to many

std::sync::mpsc implementation updated Rust's standard library has had a multi-producer, single-consumer channel since before 1.0, but in this release the implementation is switched out to be based on crossbeam-channel. This release contains no API changes, but the new implementation fixes a number of bugs and improves the performance and maintainability of the implementation.

16:20

--

16:20

something unbelievable in the c++ world

Well nowadays it is believable. Bug wg21 would give it a new name

https://github.com/rust-lang/rust/blob/stable/RELEASES.md#version-1670-2023-01-26

Ryozuki

std::sync::mpsc implementation updated Rust's standard library has had a multi-producer, single-consumer channel since before 1.0, but in this release the implementation is switched out to be based on crossbeam-channel. This release contains no API changes, but the new implementation fixes a number of bugs and improves the performance and maintainability of the implementation.

But it has no API change

So you'd end up with std::sync::mpsc_ng or sth

Jupstar ✪

But it has no API change

no, but it improves perf

But why should cpp not allow that

ABI

Jupstar ✪

But it has no API change

Some C++ containers have complexity requirements in the standard

https://cor3ntin.github.io/posts/abi/

16:22

i remembered this xd

So you are stuck with a specific implementation of unordered_map

Learath2

So you are stuck with a specific implementation of unordered_map

But i assume only if u want strict standard?

16:23

GCC can wrap their impl in agcc namespace and still link it

Jupstar ✪

But i assume only if u want strict standard?

Which is what all the compilers implement. There have rarely been intentional departures from the standard

this one seems important for ffi https://github.com/rust-lang/rust/pull/97971/

Enable varargs support for calling conventions other than C or cdec...

This patch makes it possible to use varargs for calling conventions, which are either based on C (efiapi) or C is based on them (sysv64 and win64). Also pinging @phlopsi, because he noticed first t...

16:23

was added

Learath2

Which is what all the compilers implement. There have rarely been intentional departures from the standard

Isn't there a GCC standard

The only one I can remember off the top of my head is the SSO until it became allowed

oh look

16:24

since field ordering is rust impl defined

16:24

they optimized it in this release

16:24

https://github.com/rust-lang/rust/pull/102750/

optimize field ordering by grouping m*2^n-sized fields with equival...

use std::ptr::addr_of; use std::mem; struct Foo { word: u32, byte: u8, ary: [u8; 4] } fn main() { let foo: Foo = unsafe { mem::zeroed() }; println!("base: {:p}\nword: {:...

what else does gnu++17 mean?

Jupstar ✪

Isn't there a GCC standard

Yeah, but they usually extend the standard, they don't really override it

If it's just cpp17

gnu++17 has VLAs like C and a couple other things

16:25

I think they also add restrict

The layout of repr(Rust) types now groups m*2^n-sized fields with equivalently aligned ones. This is intended to be an optimization, but it is also known to increase type sizes in a few cases for the placement of enum tags. As a reminder, the layout of repr(Rust) types is an implementation detail, subject to change.

Learath2

Yeah, but they usually extend the standard, they don't really override it

mhh thats kinda sad imo, i think it would be an easy work around against c++ standard requirements

https://news.ycombinator.com/item?id=34532983

ginzunza

Tell HN: IBM and SAP are cutting thousands of jobs

IBM (IBM) announced the cuts Wednesday, saying they were related to the previously announced spinoff and sale of two business units. Some 3,900 positions, or 1.5% of its global workforce, are expected to go. The move will cost IBM (IBM) about $300 million this quarter, a spokesperson confirmed.SAP (SAP), Europe’s largest software company, will l...

18:33

@deen your old company this time :o

18:33

not many eu companies laid off but sap did

Ryozuki

correct me if im wrong, but once the ringbuffer is filled, u wouldnt get more data

that's correct

Ryozuki

what u trying to make exactly?

I'm trying to minimize data copies

sap bankrupt since deen is not carrying anymore

Ryozuki

u can always have the vecdeque, and a stack array where u recv from and then u push it to the deque

this introduces more copies which I'm trying to avoid

heinrich5991

I'm trying to minimize data copies

Why? You should maximize safety

19:34

And code prettiness

@heinrich5991 are u writing a quic backend?

https://www.youtube.com/watch?v=MWBfmmg8-Yo hmm, two traits that we could use for templated mem_zero and mem_copy

C++Now 2018: Arthur O'Dwyer “The Best Type Traits that C++ Doesn't ...

Jupstar ✪

@heinrich5991 are u writing a quic backend?

yes

with quinn?

Learath2

Why? You should maximize safety

I'm also trying to maintain safety

Jupstar ✪

with quinn?

I decided to try quiche

mh ok, i used quinn

what were your considerations on which library to choose

19:45

?

i guess tokio integration? i read some stackoverflow or reddit

19:45

cant remember

19:46

is quiche pure rust?

afaict, none of the libraries were pure-rust, unfortunately

i think quinn is (at least for compiling)

19:47

"Quinn is a pure-rust, async-compatible implementation of the IETF QUIC transport protocol."

19:49

only thing i still dislike is tokio not offering a condition variable xD rest was okish i guess i also use tokio for async file io

i think u need to roll ur own ringbuffer impl

Jupstar ✪

"Quinn is a pure-rust, async-compatible implementation of the IETF QUIC transport protocol."

it depends on ring, which is not pure-rust

the good thing is that u cannot mix async with normal code, so u kinda force to use async io

the good thing is, u can maintain code prettyness and safety thanks to rust encapsulation of unsafety cc @Learath2

heinrich5991

it depends on ring, which is not pure-rust

is it not?

19:50

i know it has some assembly

yep

but everything uses ring iirc

19:50

rustls does right

"Pluggable cryptography, with a standard implementation backed by rustls and ring"

19:50

u can choose apparently

19:51

i used the rustls from their example

ring exposes a Rust API and is written in a hybrid of Rust, C, and assembly language.

ring is audited tho btw

Jupstar ✪

u can choose apparently

I don't see other options

why did they mix c there tho

19:51

oh maybe its cuz its a port from something old?

19:51

idk

yes, openssl

19:51

→ boringssl

19:51

→ ring

there is a dalek thing in rust

19:52

i find it pretty funny

because openssl has good and fast implementations of crypto primitives

and awesome

19:52

im a doctor who faan

19:52

https://dalek.rs/

dalek cryptography

(written in perl, which generates assembly code, I heard)

https://github.com/dalek-cryptography

dalek cryptography

Fast, safe, pure-rust elliptic curve cryptography. dalek cryptography has 11 repositories available. Follow their code on GitHub.

19:53

its awesome

heinrich5991

I don't see other options

what do i need to search for in the symbols?

19:53

to check if its linked with my project

dalek-cryptography is a collection of libraries providing safe, mid-level APIs for cryptographic primitives, aimed at making it faster and easier to implement cryptographic protocols such as key agreement, signatures, anonymous credentials, range proofs, and zero-knowledge proof systems.

>

Other libraries aim to provide well-designed APIs for protocol users, but do not aim to provide well-designed API for protocol implementers. The result is that implementing new protocols often requires using unsafe internal APIs which were never intended for reuse.

Jupstar ✪

to check if its linked with my project

check Cargo.lock, search for name = "ring"

try cargo tree

heinrich5991

check Cargo.lock, search for name = "ring"

but if its unused its dead code isnt it

cargp tree | grep "ring"

i'd prefer to check stripped version

idk how

  _ZN85_$LT$ring..io..writer..LengthMeasurement$u20$as$u20$ring..io..writer..Accumulator$GT$10write_byte17h5028fc7b2b977ff8E
00000000000fc4e0 l     F .text  0000000000000004              _ZN85_$LT$ring..io..writer..LengthMeasurement$u20$as$u20$ring..io..writer..Accumulator$GT$11write_bytes17h0a8d19af71ae01d3E
00000000000fc4f0 l     F .text  0000000000000035              _ZN74_$LT$ring..io..writer..Writer$u20$as$u20$ring..io..writer..Accumulator$GT$10write_byte17ha0b1ab21820904a2E
00000000000fc530 l     F .text  0000000000000060              _ZN74_$LT$ring..io..writer..Writer$u20$as$u20$ring..io..writer..Accumulator$GT$11write_bytes17hddfc4814984ae122E
0000000000320018 l     O .data  0000000000000020              _ZN4ring4rand7urandom4fill4FILE17hefd278b0303a3e1cE
00000000000fcac0 l     F .text  0000000000000579              _ZN4ring3rsa12verification93_$LT$impl$u20$ring..signature..VerificationAlgorithm$u20$for$u20$ring..rsa..RsaParameters$GT$6verify17h5ff374f4b5b4fe5dE

i see some symbols that are probably related to ring, so its in indeed

19:58

ah ok

19:58

but it seems it doesnt expose any dynamic symbols

19:59

these are from debug symbols i assume, after stripping they are gone

20:00

but yeah maybe its static linked do ppl ship stuff in crates that requires a c compiler (to build) @heinrich5991 ?

yes

i'd have thought that isnt the nicest style xD

20:00

ok

20:01

thats a bit sad

and all rust quic crates fall into that category, as far as I could tell

20:01

which is annoying

Ryozuki

https://codedoc.ddnet.tw/classCUIRect.html

WHERE ISSSSSSS

but i guess for quinn its a matter of time? i mean if they already give an alternative

Image attachment

cant be too hard to get away from ring/ make it optional

ah ye

20:02

@deen did u update the dns for codedoc?

brownbear

Jupstar ✪

cant be too hard to get away from ring/ make it optional

famous last words. not really; constant-time code can only be done in assembly, I think

20:02

I guess you could do them in rust inline assembly

20:02

that would work

heinrich5991

famous last words. not really; constant-time code can only be done in assembly, I think

are such packages somehow flagged in crates?

20:03

i'd like to know if i have other packages like this

using C compiler?

20:03

no

@heinrich5991 https://crates.io/crates/subtle

20:03

made by the dalek guys i told u

20:03

https://docs.rs/subtle/2.4.1/subtle/

subtle

Jupstar ✪

i'd like to know if i have other packages like this

break your C compiler and check whether the clean build still works

20:03

e.g. put a broken C compiler in your path

heinrich5991

break your C compiler and check whether the clean build still works

no need, i use rustls and they also use ring

This code is a low-level library, intended for specific use-cases implementing cryptographic protocols. It represents a best-effort attempt to protect against some software side-channels. Because side-channel resistance is not a property of software alone, but of software together with hardware, any such effort is fundamentally limited.

>

USE AT YOUR OWN RISK

Ryozuki

https://codedoc.ddnet.tw/classCUIRect.html

@deen tell me if u will update :p

i cant do nothing from my side

20:08

cloudflare dns needs to be updated

is there no pure rust openssl/boringssl alternative? i'd have thought thats a good rust use case

@heinrich5991 do u have access too?

Jupstar ✪

is there no pure rust openssl/boringssl alternative? i'd have thought thats a good rust use case

rustls iirc

Ryozuki

cloudflare dns needs to be updated

i know

Ryozuki

rustls iirc

it depends on ring

@Jupstar ✪ doing something like openssl/ring is a tremendous effort and requires a lot of audit

20:09

so the only h ope is probs they update ring removing more and more C

Ryozuki

@Jupstar ✪ doing something like openssl/ring is a tremendous effort and requires a lot of audit

they could start with the most common implementations

ring is the closest u have rn i think

20:10

i mean, doesnt ring do that

20:10

@Jupstar ✪ well

20:10

there is the dalek group i sent

20:10

https://github.com/dalek-cryptography

20:11

looks like they focus mostly on elliptic tho (edited)

20:12

there is also https://github.com/RustCrypto

well i just find it weird that its not easy to find out which package uses which compiler

which has a lot of stuff

if i read "pure rust" i'd think its pure rust xD

Jupstar ✪

if i read "pure rust" i'd think its pure rust xD

xd

20:14

honestly its possible to not use ring

20:14

just use some stuff from rust-crypto and subtle

well i hope they get this settled in ~1 year

why 1 year xd

20:14

its just that the stuff is not all in 1 crate

20:14

but in various

Ryozuki

@heinrich5991 do u have access too?

I have access

Ryozuki

why 1 year xd

bcs i want to finish my project in 1 year

what do I need to do?

i migrated my server so u need to change the codedoc.* ip

what project?

@heinrich5991 copy it from wiki.ddnet.org

20:15

A and AAAA

20:16

ty

why do I add both A and AAAA?

you update them

it's proxied by cloudflare anyway

yes

20:18

u change the proxy xD

20:18

the dns

20:18

?

I just wondered why I need to add both IPv4 and IPv6, one would suffice

hmm maybe its faster with ipv6 from their side

20:19

why not?

20:19

maybe the proxy has even better routing with ipv6 to my server

because two entries could get out of sync

cmon

20:19

ur being pedantic

20:20

anyway i dont care dont wanna argue

20:20

u do u

I add both

20:20

no worries

20:20

but it seems weird to me

maybe one day i lose the ipv4

20:20

they expensive these days

then we won't notice the problem because we added both

20:21

only slow site loads, perhaps

i need to get a cert on my side for it to work

20:22

done

20:22

hmm maybe i lost the old theme xd

20:22

will do for now i guess

20:23

i wanna try https://datasette.io/ with the ddnet db

Datasette is a tool for exploring and publishing data. It helps people take data of any shape, analyze and explore it, and publish it as an interactive website and accompanying API.

20:23

tomorrow ill check

not sure if you actually needed certificates tbh

probs not but on the wiki it didnt work otherwise

20:24

actually the thing to do here would be u give me a origin cert

20:24

but 2 lazy anyway

20:25

@heinrich5991 i think ddnet has the full strict config on cloudflare

20:25

so proxy - origin needs to be encrypted too

no, unfortunately not

20:25

only "full"

20:26

I changed it to full (strict)

20:26

did anything break?

ah well its "Encrypts end-to-end, using a self signed certificate on the server"

20:27

thats why it requires me to have a cert on my side too

20:27

altho i just use lets encrypt a self signed would work

heinrich5991

did anything break?

no it works fine

20:27

i just forgot to port my theme

20:27

it was dark themed before

heinrich5991

I changed it to full (strict)

oh

20:28

if u change it to full strict wiki and codedoc will break

20:28

because u havent provided me a origin cert

20:29

Encrypts end-to-end, but requires a trusted CA or Cloudflare Origin CA certificate on the server

20:29

ahh

20:29

wait

20:29

it doesnt break cuz im using a trusted CA xd

yes

im stupid

heinrich5991

famous last words. not really; constant-time code can only be done in assembly, I think

constant-time code doesn't exist on most, if not all, architectures

20:34

memory access on multicore design is non deterministic

20:34

interruptions are non deterministic

20:34

context switches are non deterministic

sounds irrelevant for cryptography

20:35

the interruptions don't depend on the data

20:35

memory access does not depend on the data being encrypted

20:36

context switches don't happen during encryption, except interrupts which don't depend on data

heinrich5991

sounds irrelevant for cryptography

is very relevant for cryptography, that's how most hardware attacks work

can you give an example?

https://mimoza.marmara.edu.tr/~msakalli/cse466_09/cache%20timing-20050414.pdf

20:55

Substitution boxes are often the target of such attacks

is the rest of my argumentation valid?

21:07

interrupts, context switches?

you need context switches on monocore for timing attacks, context switches are based on timer interruptions

21:16

for multicore, you don't need those

heinrich5991

memory access does not depend on the data being encrypted

but this point is false for AES for instance

21:17

depends on implementation, you can do it without lookup table, but then code path is also data dependant

actually, context switches can be triggered by any external event that the OS needs to handle

yep

21:29

but since that doesn't depend on the data, it can't be used to extract the data

21:30

it needs data-dependent timings from elsewhere

21:30

it can only be used to measure the data-dependent timings (edited)

i proudly present https://ddstats.org/

22:02

(its not the old project)

22:03

i setup a instance with datasette

22:03

so u can do all the queries u want

22:03

against today ddnet database

22:03

and generate graphs!

Epic

22:07

Show some examples xd

https://ddstats.org/ddnet-0587c26?sql=select%0D%0A++Map%2C%0D%0A++count%28*%29+as+Count%0D%0Afrom%0D%0A++race%0D%0Agroup+by%0D%0A++Map%0D%0Aorder+by+Count+DESC%0D%0Alimit%0D%0A++20#g.mark=bar&g.x_column=Map&g.x_type=ordinal&g.y_column=Count&g.y_type=quantitative

22:11

xd

22:11

Image attachment

Image attachment

22:12

it won't respect order for some reason

you can also query it and it returns json

22:12

or wiht graphql

Learath2

Click to see attachment 🖼️

ye idk why :x

22:13

if you click "copyable"

22:13

and then fancy

22:13

╒═════════════════╤═════════╕
│ Map             │   Count │
╞═════════════════╪═════════╡
│ Multeasymap     │  847102 │
│ Sunny Side Up   │  688313 │
│ Tutorial        │  615787 │
│ Kobra           │  494704 │
│ LearnToPlay     │  398745 │
│ Just2Easy       │  273273 │
│ 4Beginners      │  256242 │
│ Epix            │  250523 │
│ Kobra 2         │  238256 │
│ run_g6          │  202859 │
│ StepByStep      │  169370 │
│ Kobra 4         │  162996 │
│ Tsunami         │  146537 │
│ NUT_short_race6 │  145741 │
│ run_ankii       │  145453 │
│ run_blue        │  138420 │
│ Linear          │  135617 │
│ 4Nubs           │  128939 │
│ Gold Mine       │  119653 │
│ Tangerine       │  117432 │
╘═════════════════╧═════════╛

22:13

easy table

22:13

https://ddstats.org/ddnet-0587c26.copyable?sql=select%0D%0A++Map%2C%0D%0A++count%28%2A%29+as+Count%0D%0Afrom%0D%0A++race%0D%0Agroup+by%0D%0A++Map%0D%0Aorder+by+Count+DESC%0D%0Alimit%0D%0A++20&_table_format=pretty

22:13

@deen maybe ur interested in promoting this :D

Ryozuki

i proudly present https://ddstats.org/

.

i set it up so it autoupdates everyday

i added another visualizer that works kinda better

22:43

Image attachment

gotta purge cache for it to appear tho xd

why are the bars of different widths?

[ddnet/ddnet] Pull request opened: #6312 Increase relative joystick movement based on input update time, disable relative joystick when console is open or window not focused

Checklist

[X] Tested the change ingame
[ ] Provided screenshots if it is a visual change
[ ] Tested in combination with possibly related configuration options
[ ] Written a unit test (especially base/) or added coverage to integration test
[ ] Considered possible null pointers and out of bounds array indexing
[ ] Changed no physics that affect existing maps
[ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addres...

heinrich5991

why are the bars of different widths?

i set it to that, its relative to the count

23:08

by default it isnt xD

so the area is proportional to count²?

i guess

that's a bit misleading

hmm idk

23:10

Image attachment

23:10

https://ddstats.org/ddnet-0587c26?sql=select%0D%0A++Map%2C%0D%0A++count%28*%29+as+Count%0D%0Afrom%0D%0A++race%0D%0Agroup+by%0D%0A++Map%0D%0Aorder+by+Count+DESC%0D%0Alimit%0D%0A++100#dataExplorer.metadata=%7B%22dx%22%3A%7B%22view%22%3A%22bar%22%2C%22lineType%22%3A%22line%22%2C%22areaType%22%3A%22hexbin%22%2C%22selectedDimensions%22%3A%5B%22Map%22%5D%2C%22selectedMetrics%22%3A%5B%5D%2C%22pieceType%22%3A%22bar%22%2C%22summaryType%22%3A%22violin%22%2C%22networkType%22%3A%22force%22%2C%22hierarchyType%22%3A%22dendrogram%22%2C%22trendLine%22%3A%22none%22%2C%22marginalGraphics%22%3A%22none%22%2C%22barGrouping%22%3A%22Clustered%22%2C%22colors%22%3A%5B%22%23DA752E%22%2C%22%23E5C209%22%2C%22%231441A0%22%2C%22%23B86117%22%2C%22%234D430C%22%2C%22%231DB390%22%2C%22%23B3331D%22%2C%22%23088EB2%22%2C%22%23417505%22%2C%22%23E479A8%22%2C%22%23F9F39E%22%2C%22%235782DC%22%2C%22%23EBA97B%22%2C%22%23A2AB60%22%2C%22%23B291CF%22%2C%22%238DD2C2%22%2C%22%23E6A19F%22%2C%22%233DC7E0%22%2C%22%2398CE5B%22%5D%2C%22chart%22%3A%7B%22metric1%22%3A%22Count%22%2C%22metric2%22%3A%22none%22%2C%22metric3%22%3A%22none%22%2C%22metric4%22%3A%22none%22%2C%22dim1%22%3A%22Map%22%2C%22dim2%22%3A%22none%22%2C%22dim3%22%3A%22none%22%2C%22timeseriesSort%22%3A%22array-order%22%2C%22networkLabel%22%3A%22none%22%7D%7D%7D

2

23:10

xd

23:10

(you can click the view 100 rows in explorer)

23:11

ah there is also a graphql api

23:11

https://ddstats.org/graphql

23:12

dont ask me how it works idk graphql

23:12

fancy

Image attachment

Exported 426 message(s)