open source crispr cat ears @haskal@cybre.space

ok i need to sleep,,,

i'm not even mad about the proprietary software licensing being proprietary, like, on top of that it literally just does not work

what if our shitty drm scheme for our own software actually worked???
that would make it too easy for people to use our software now wouldn't it

ok so here's the other thing that happened today (yesterday)
so
we have license server at school with licenses for partial reconfiguration in vivado, so i set my machine up with the license server and then open vivado, and then from inside vivado i open the license manager and it's like yeah, you have floating license for partial reconfiguration and there are currently 0 out of 5 seats used
cool
back to vivado
enable partial reconfiguration for this project
error: you don't have a license

xilinx is just, critically incapable of doing literally anything in any sort of reasonable manner and it kind of hurts

it rly doesn't have to be this way

while reading xilinx kernel source for this it also turns out under normal usage it just leaks memory
because, lol,

anyway fuck xilinx fuck engineers, like,
engineers write a single line of code that is actually not hot garbage challenge 2020

xilinx has a linux driver for writing fpga configuration on zynq platform
this driver actually has a classic trivial concurrency bug in it that prevents it from working, and i was stuck on this for literal hours

so what's the bug
well, when you write configuration it does a DMA transfer, so it initiates the transfer and then it spins on a flag waiting for it to complete
well actually it turns out the flag is not spun on correctly so the compiler optimizes it out and boom every write actually errors
well kind of
it errors but in the meantime the dma has actually completed successfully, but because of this error you need to write all the data at once, you can't do it in chunks or you'll error on the first chunk and you're done
so basically the xilinx manuals tell you to do cat bitstream > /dev/xdevcfg
this is wrong
what i did was
created a basic C program that reads a file fully into memory and then issues a giant single write call to /dev/xdevcfg which errors but ends up successfully programming the fpga

ok so i'm switching to pub for this part because i need to rant out loud

vivado
more like vivadon't
~ iitalics

spirited awoo,,,,

i use linux because i literally just do not have the time to deal with windows bullshit for hours
i don't want to install gigabytes of crap and a microsoft terminal to be able to write code
i don't wanna install "cable drivers" to use ftdi serial devices
i don't want to get interrupted by bluescreens and then reboot to find the file is now overwritten by all null bytes and it wasn't checked into git yet, whoops! microsoft is reawwy sowwy for your data loss!!!

like when you install manjaro on this laptop it just works, nvidia drivers with bumblebee, wifi, everything, 0 effort, you get your ez development environment
and as a bonus it looks fuckin pretty too

it annoys me that manjaro is turnkey as fuck but nobody knows what a linux is or they're too afraid to use it

meanwhile windows has literal ads built into the desktop ui and also bluescreens all the time and people r like, yeah, this is fine, this is normal