DDraceNetwork - developer

@Chairn https://bugzilla.mozilla.org/show_bug.cgi?id=1751709

1751709 - RDD VA-API sandbox crashes due to denied getpwuid_r access

NEW (jld) in Core - Security: Process Sandboxing. Last updated 2022-02-15.

07:24

i found the bug

07:29

disabling vaapi fixes it

07:29

sad

https://twitter.com/marcan42/status/1494213855387734019

Hector Martin (@marcan42)

Well, this is unfortunate. It turns out Apple's custom NVMe drives are amazingly fast - if you don't care about data integrity. If you do, they drop down to HDD performance. Thread.

07:43

(this is the spanish reverse engineer guy i mentioned, porting linux to m1)

07:43

07:43

As it turns out, macOS cheats. On Linux, fsync() will both flush writes to the drive, and ask it to flush its write cache to stable storage.

>

But on macOS, fsync() only flushes writes to the drive. Instead, they provide an F_FULLSYNC operation to do what fsync() does on Linux.

07:44

Macbook Air M1 (macOS): - flushing: 46 IOPS - not: 40000 IOPS

>

x86 iMac + WD SN550 1TB NVMe (Linux): - flushing: 2000 IOPS - not: 20000 IOPS

>

x86 laptop + Samsung SSD 860 EVO 500GB SATA: - flushing: 143 IOPS - not: 5000 IOPS

07:44

@bencie Hey bencie , can you give me your steam profile?

fstd: Certificate for paste.pr0.tips expired

я Руслан

@bencie Hey bencie , can you give me your steam profile?

it's on my discord profile (edited)

09:42

but id/benciee

bencie

it's on my discord profile (edited)

Okay, thanks, I’ll look now. My nickname is Ruslan (or Руслан)

bors[bot]

[ddnet:master] 2 new commits

14e5222 Better fix for sqlite3_expanded_sql - def- c8363bc Merge #4719 - bors[bot]

10:46

bors[bot]

[ddnet/ddnet] Pull request closed: #4719 Better fix for sqlite3_expanded_sql

Ryozuki

As it turns out, macOS cheats. On Linux, fsync() will both flush writes to the drive, and ask it to flush its write cache to stable storage.

>

But on macOS, fsync() only flushes writes to the drive. Instead, they provide an F_FULLSYNC operation to do what fsync() does on Linux.

so not surprised, changing the definition to score better on tests is what profit driven entities do best

It's not exactly news though: https://github.com/etcd-io/etcd/issues/5029

storage: consider using F_FULLSYNC for HFS/OSX systems · Issue #502...

I was talking with Mikeal Rogers who used to work on couchdb and one thing he learned is that sync on OSX lies and that you need to use F_FULLSYNC instead: https://developer.apple.com/library/mac/d...

11:34

otherwise why would they have F_FULLSYNC if fsync already did that

How do you know? Do you use it? (<@749222324980416602_=5bquakenet=5d=20deen>)

[quakenet] deen#0000

fstd: Certificate for paste.pr0.tips expired

even the manpage of fsync says so:

Note that while fsync() will flush all data from the host to the drive (i.e. the "permanent storage device"), the drive itself may not physically write the data to the platters for quite some time and it may be written in an out-of-order sequence.

11:37

Specifically, if the drive loses power or the OS crashes, the application may find that only some or none of their data was written. The disk drive may also re-order the data so that later writes may be present, while earlier writes are not.

>

This is not a theoretical edge case. This scenario is easily reproduced with real world workloads and drive power failures.

>

For applications that require tighter guarantees about the integrity of their data, Mac OS X provides the F_FULLFSYNC fcntl. The F_FULLFSYNC fcntl asks the drive to flush all buffered data to permanent storage. Applications, such as databases, that require a strict ordering of writes should use F_FULLFSYNC to ensure that their data is written in the order they expect. Please see fcntl(2) for more detail.

Well the news is that fsync on macOS doesn't satisfy atleast the POSIX rationale and when you do use the provided equivalent you end up with performance far below what is expected of an nvme ssd

But still thanks @Ryozuki for bringing this up, something I can use for my job

the question is, did apple advertise their speed? ofc they pick the best case

11:45

but i can also use a RAM fs and claim it to be faster than any other fs xD

when nearly no software does F_FULLSYNC, then you won't notice any problems in benchmarks

but a good example on why we need well written intuitive standards

11:47

bcs what programmer would think about this while not reading the manpage for all OS

I would read the POSIX manpage and learn that "The fsync() function is intended to force a physical write of data from the buffer cache, and to assure that after a system crash or other failure that all data up to the time of the fsync() call is recorded on the disk."

I wanted to make a small contribution to SQLite, but first you have to sign an affidavit and send it by letter to the U.S. Sounds a bit annoying for a small comment fix

deen

I wanted to make a small contribution to SQLite, but first you have to sign an affidavit and send it by letter to the U.S. Sounds a bit annoying for a small comment fix

lolwat?

https://www.sqlite.org/copyright.html

SQLite is open-source, meaning that you can make as many copies of it as you want and do whatever you want with those copies, without limitation. But SQLite is not open-contribution. In order to keep SQLite in the public domain and ensure that the code does not become contaminated with proprietary or licensed content, the project does not accept patches from people who have not submitted an affidavit dedicating their contribution into the public domain.

12:07

An older version of that page had "Only 27 individuals have ever contributed any code to SQLite, and of those only 16 still have traces in the latest release."

12:11

(Yugabyte also requires such a Contributor License Agreement, but you can just do it online in a few seconds, funnily enough with a service by SAP: https://cla-assistant.io/)

And some large projects ran into huge trouble by not having a CLA when they wanted to relicense: https://blog.llvm.org/posts/2021-11-18-relicensing-update/ LLVM has to reach every single contributor to get approval from them

LLVM relicensing update & call for help

In this blog post, I’d like to summarize the main points I talked about in the relicensing update presentation at the 2021 LLVM Developer’s meeting. The very short summary is that we are currently in the long tail phase of collecting relicensing agreements of past contributors.

chillerdragon

How do you know? Do you use it? (<@749222324980416602_=5bquakenet=5d=20deen>)

We use it for the discord bot: https://github.com/ddnet/ddnet-discordbot

GitHub - ddnet/ddnet-discordbot

Contribute to ddnet/ddnet-discordbot development by creating an account on GitHub.

12:19

so our moderators are not happy when it fails

12:20

but it's run by Learath2 anyway, so ping was useless

def-

[ddnet/ddnet] Issue opened: #4722 LAN server crashes when changing map

People in #bugs report this on Windows, I can't reproduce in Wine.

Jupeyy

[ddnet/ddnet] Issue closed: #4559 Hz problems in fullscreen modes with alt-tab

@bencie I'm sorry, but how can I create my own models? (Hooks, skins)

well, creating your own is way harder than editing one, but you can draw your own models onto a template, and then delete the template at the end so only your visuals remain. https://skins.tw/gridTemplates

Teeskins: Teeworlds Grid Templates

Download and share many Teeworlds grid templates. Create your own teeworlds assets with these templates. Templates show you the positioning and borders, which you have to consider so that they are compatible with teeworlds

bencie

well, creating your own is way harder than editing one, but you can draw your own models onto a template, and then delete the template at the end so only your visuals remain. https://skins.tw/gridTemplates

Sorry for taking so long to reply, got out of the house. I'm reading it now, thank you, thank you!

1

13:45

def-

[ddnet-scripts:master] 2 new commits

055c739 Update ddnet-setup.sh - def- a46d11e Add steam images - def-

15:37

def-

[ddnet-scripts:master] 1 new commit

4c6c053 Add copyright line - def-

hi! How can we get infclass maps hosted on maps2.ddnet.tw?

16:48

it would actually be nice to allow the servers communicate the full url to the map

Hi breton. Yes, would be nice. So we used to scan servers for all maps and used to host them

17:27

But Teeworlds servers used short hashes which were easy to cause collisions

17:27

and people abused this, so we stopped. Since then only the official DDNet maps are available there.

2

@Jupstar ✪ tomorrow ill install the gpu

17:29

i noticed i can enable amd SAM

17:30

and i have a pcie 4.0 motherboard

17:30

https://www.amd.com/en/technologies/smart-access-memory

17:30

def-

[ddnet/ddnet] Issue opened: #4723 Crash with failed dbg_assert: incorrect id

Crash on CHN8, log is already deleted: ``


Core was generated by

./DDRace64-Server_sql -f servers/8321.cfg'. Program terminated with signal SIGILL, Illegal instruction. #0 dbg_break () at /home/teeworlds/src/master/src/base/system.cpp:168 168 /home/teeworlds/src/master/src/base/system.cpp: No such file or directory. --Type for more, q to quit, c to continue without paging-- [Current thread is 1 (Thread 0x7f15b1137b80 (LWP 2691))] (gdb) bt #0 dbg_break () at /home/teeworlds/src/ma...

Ryozuki

@Jupstar ✪ tomorrow ill install the gpu

you are so patient xd

Jupstar ✪

you are so patient xd

well i needed a new monitor xd