IRC channel logs

<jab>damo22: if I don't update the Hurd in a while, then I always update in the Hurd console. I've gotten issues when trying to update after X starts.

<damo22>i dont have X installed on hurd

<gnucode>It's kind of fun to use X. :) It's what us Hurd newbies do!

<gnucode>now I con't get ballz to work.

<damo22>error: RPC failed; curl 92 HTTP/2 stream 0 was not closed cleanly: CANCEL (err 8)

<damo22>error: 2126 bytes of body are still expected

<damo22>fetch-pack: unexpected disconnect while reading sideband packet

<almuhs>blobby volley fails too, with same glx error

<damo22>can you paste the error somewhere

<damo22>im having trouble cloning a big repo

<gnucode>biloba fails too

<almuhs> https://pasteboard.co/HKyUQnR3kWO4.png

<gnucode>GLXBadContext

<damo22>X_GLXIsDirect

<damo22>need to find that in the source

<gnucode>biniax2 fails as well. GLXBadContext

<gnucode>blobby failed too. couldn't open audio device. GLXBadContext

<gnucode>bloboats failed as well. Same error.

<almuhs>alex4 works. But it's really retro

<gnucode>blobwars fails as well.

<almuhs>there is a bug in opengl, definitelly

<almuhs>about assaultcube, i think that the problem could be in the script

<almuhs> https://github.com/assaultcube/AC/blob/master/assaultcube.sh

<gnucode>I can't install alex4

<almuhs>from the snapshot too

<gnucode>blupimania is also not working.

<almuhs>same glx error?

<damo22>do these games all use libSDL?

<damo22>the xorg-server parses the request to open a GLX context

<gnucode>bombardier works.

<damo22>part of the request is whether the GLX context is a direct one or not

<gnucode>text based game.

<damo22>almuhs: can you try LIBGL_ALWAYS_INDIRECT=1 ./game

<gnucode>doomsday almost worked...

<almuhs>in which game?

<damo22>for a start, try: LIBGL_ALWAYS_INDIRECT=1 glxgears

<damo22>do you get the spinning gears?

<almuhs>glxgears?

<gnucode>I got a pop up window.

<gnucode>then when the game actually loaded...

<damo22>glxgears is a built in test program to try opengl

<damo22>with the flag i am trying to force it to use software rendering

<almuhs>LIBGL_ALWAYS_INDIRECT=1 glxgears shows same glx error

<damo22>ok

<damo22>what about glxinfo

<gnucode>QOpenGLWindow::beginPaint: Failed to create context

<almuhs>same

<damo22>ok that sucks

<almuhs>it seems an opengl bug

<damo22>no its not using the software renderer

<almuhs>but i can't downgrade easily

<damo22>there is a environment variable that can override it

<damo22>its not a bug its a bad default

<damo22>LIBGL_ALWAYS_SOFTWARE=1 glxinfo

<almuhs>same error

<damo22>sudo apt install libgl1-mesa-swx11

<almuhs>this package is not available in repos

<damo22>maybe no 1 in the name

<damo22>libgl-mesa-swx11 ?

<almuhs>neither

<damo22>hmm

<almuhs>there are not any package with swx11 in its name

<almuhs>the most similar is libgl1-mesa-glx

<damo22>ok is that installed libgl1-mesa-glx

<almuhs>i have to go sleep

<almuhs>you can install xorg with a little desktop in a VM and make the experiments

<damo22>ok

<gnucode>crawl works. :)

<gnucode>sudo apt install libgl1-mesa-swx11 -> no installation candidate

<damo22>what about -glx

<damo22>libgl1-mesa-glx

<damo22>is that installed already?

<gnucode>it installed just now.

<gnucode>should I go ahead and re-test all those games?

<damo22>ok, try glxinfo

<gnucode>glxinfo --> binary not found

<gnucode>sudo apt install glxinfo --> failed

<damo22>you are missing some stuff

<gnucode>probably.

<damo22>i dont know the names of all the libs but you can check the "mesa" debian source to find them

<gnucode>maybe I should try to install mesa.

<damo22>eg in here https://salsa.debian.org/xorg-team/lib/mesa/-/tree/debian-unstable/debian

<damo22>there is no meta package called mesa to install all of mesa

<damo22>you just have to install them one by one

<gnucode>kk

<damo22>you probably need libglx-mesa0

<gnucode>I got that one.

<gnucode>probably after I installed mesa-common-dev

<damo22>mesa-utils

<damo22>install that ^ then do this: LIBGL_ALWAYS_SOFTWARE=1 glxinfo

<gnucode>sudo apt install mesa-utils

<gnucode>LIBGL_ALWAYS_SOFTWARE=1 glxinfo

<gnucode>X Error of failed request: GLXBadContext

<gnucode>Xlib: sequence lost (0x10014 > 0x16) in reply type 0x0!

<gnucode>I can put the whole error message on a pastebin if you like

<gnucode>snake in emacs works, but it's really laggy

<damo22>youpi: i have a feeling that rumpdisk does not flush its cache often enough, and so if you write a lot, it buffers the writes until it fills up a cache and then somehow gets stuck

<damo22>but if i regularly type 'sync' manually, it does not get stuck

<damo22>is it true that we are caching twice, once in rump vfs and once in pager?

<damo22>youpi: i have been in contact with a netbsd kernel developer, apparently if we are using pread/pwrite on a rwd0 device, there is no caching, but we are wasting ram for no reason, we can set -DBUFPAGES=16

<damo22>when building rump

<damo22>(03:48:59 PM) mlelstv: I'm not 100% sure if 16 pages are enough, but they should be.

<damo22>(03:49:30 PM) mlelstv: sufficient to do one maximum sized I/O request, if not enough, increase slightly

<damo22>youpi: rumpkernel debian package wont compile for me on hurd unless i add -D_GNU_SOURCE

<damo22>wcwidth cannot be located

<damo22>gnumachUser.c: In function 'thread_get_name':

<damo22>gnumachUser.c:1791:37: error: comparison of integer expressions of different signedness: 'unsigned int' and 'int' [-Werror=sign-compare]

<damo22> 1791 | if (mig_unlikely (msgh_size != 36 + ((OutP->nameType.msgt_number + 3) & ~3)))

<damo22> | ^~

<youpi>damo22: is it true that we are caching twice, once in rump vfs and once in pager?

<youpi>That's possible, yes, AFAIR the people doing the rump port didn't manage to understand how to control the caching done in rump

<youpi>> we are wasting ram for no reason, we can set -DBUFPAGES=16

<youpi>That's very possible

<youpi>> wcwidth cannot be located

<youpi>Missing #include, very common issue nowadays with the requirement of having all declarations

<youpi>> comparison of integer expressions of different signedness: 'unsigned int' and 'int' [-Werror=sign-compare]

<youpi>We probably want to disable sign-compare being an error

<youpi>(and fix mig)

<youpi>(note: I can provide some quick advise about these ; I can't magically find time to work on them)

<damo22>ok

<youpi>damo22: to sync more regularly, use the --sync option of /hurd/ext2fs

<youpi>(but we want to fix the caching controlling issue of course)

<damo22>damn, i cant figure this ram usage out

<damo22>is there a command on hurd that dumps the memory map of a process?

<damo22>theres like 200mb rss

<damo22>also, running rump_init() in userspace kills the running rumpdisk

<damo22>well not kills it, but acpi reroutes the irq and causes havoc on the running disk driver

<youpi>damo22: vminfo

<bool`>is there any place for a newer programmer to contribute to gnu/hurd, or atleast learn from it? i'm very curious about it

<gnucode>hey hure people!

<gnucode>hurd*

<gnucode>bool`: yes! Feel free to jump in!

<gnucode>grrr, my T43 locked up on me again.

<gnucode>I wonder if it's 'cause this box has 1.5GB of RAM. That's not a lot of memory...

<gnucode>I don't think that's it though. I use i3, emacs, netsurf (with max 3-5 tabs open), and a couple of terminals (rxvt). When I run top (and I run it a lot to make sure I am not exhasting memory...)I normaly have .5GB of RAM free.

<gnucode>lots of ram gets used as a buffer/cache.

<gnucode>bool`: The hurd.gnu.org contributing page has a small hack entries stop.

<gnucode>geez, X just locked up on my again when I started playing ballz.

<gnucode>weirdly enough I have to start ballz like 3 times before it'll start properly.

<gnucode>weird.

IRC channel logs

2024-10-14.log