<nekobot>
[haiku/haiku] f72d1684e521 - kernel/debug: Initialize the frame buffer console much earlier, if possible.
mmu_man has quit [Ping timeout: 480 seconds]
SLema has quit [Ping timeout: 480 seconds]
bjorkintosh has quit [Remote host closed the connection]
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
erysdren has joined #haiku
AlienSoldier has quit [Quit: Vision[]: i've been blurred!]
erysdren has quit [Quit: Konversation terminated!]
bbjimmy_64 has quit [Ping timeout: 480 seconds]
rexbinary_ has joined #haiku
rexbinary has quit [Ping timeout: 480 seconds]
rexbinary_ is now known as rexbinary
rexbinary_ has joined #haiku
rexbinary has quit [Ping timeout: 480 seconds]
rexbinary_ is now known as rexbinary
Begasus has joined #haiku
<Begasus>
morning peeps
_justin_kelly9759578656 has quit [Read error: Connection reset by peer]
_justin_kelly9759578656 has joined #haiku
freddietilley has joined #haiku
<mbrumbelow>
Crickets... 🦗
<dovsienko>
hello all
<dovsienko>
waddlesplash: 20 seconds to complete "git status" in the Haiku git repository after a fresh boot of hrev58344 in VirtualBox 7.1.4. judging by the Tracker indicator, each of the four CPU cores was about 25% busy
<Begasus>
hi mbrumbelow dovsienko
<Begasus>
it's snowing here! :)
<mbrumbelow>
Yikes!
<mbrumbelow>
Too cold for me.
<Begasus>
about 3°C atm, by the weekend about 15°C, so it's a few hours at most I think we'll see it :)
<dovsienko>
maybe grab you camera then and have some fun away from the computer?
<mibi88[m]>
And the KDL I just got is very similar to the one I got a few days ago.
<matthewcroughan>
That's awesome if so
<zard>
matthewcroughan: No idea ;)
Begasus has joined #haiku
<mibi88[m]>
Oh what I got doesn't look like a KDL then… But what is it?
bbjimmy has quit [Ping timeout: 480 seconds]
<zard>
It looks similar to the output of the command `Debugger -c path/to/a/program`
<mibi88[m]>
I just rebooted, I want to continue coding :)
Mibi88 has joined #haiku
<zard>
Maybe you can grab a backtrace with 'bt' and then quit Debugger 'quit'
<Mibi88>
I tried, pressing enter did nothing
<Mibi88>
So I just rebooted
<zard>
Mmm. You're screenshot doesn't show any "debugger> " line, which I would expect to exist
mmu_man has quit [Ping timeout: 480 seconds]
_-Caleb-_ has left #haiku [#haiku]
_-Caleb-_ has joined #haiku
freddietilley has quit [Quit: WeeChat 4.4.2]
mmu_man has joined #haiku
san2ban has quit [Quit: Vision[]: i've been blurred!]
<augiedoggie>
looks like an app_server crash
mmu_man has quit [Ping timeout: 480 seconds]
Begasus has quit [Quit: Vision[]: i've been blurred!]
Begasus has joined #haiku
<waddlesplash>
bbjimmy_64: still there?
Mibi88 has quit [Quit: Vision[]: i've been blurred!]
mmu_man has joined #haiku
chaosDNA has joined #haiku
OscarL has joined #haiku
<OscarL>
mibi88[m]: that doesn't looks like a KDL. More like your app_server just crashed.
<OscarL>
(I had a couple of those lately)
Anarchos has quit [Remote host closed the connection]
<OscarL>
ah... should read complete log first (augiedoggie already mentioned it being an app_server crash :-D)
HaikuUser has joined #haiku
HaikuUser has quit []
<mibi88[m]>
Why couldn't the appserver just restart. It would be useful (instead of requiering a full reboot).
<OscarL>
no idea what's the limitation there :-/
<OscarL>
mibi88[m]: try looking on the syslog for signs of that "debugger for team 145" message. IIRC, at least on one occasion, I got some info "survive" in the syslog.
<OscarL>
Begasus: save some of that snow, and mail it to me near xmas? (most likely to be >40 C down here :-P)
mmu_man has quit [Ping timeout: 480 seconds]
xet7 has quit [Remote host closed the connection]
<waddlesplash>
app server restarts should be possible
<waddlesplash>
I have a todo list item to investigate why they don't work
boody24[m] has joined #haiku
<boody24[m]>
Hi
<boody24[m]>
I'm interested in the ARM port and device tree support in the list ideas here
<OscarL>
you would also search for "arm" (but that seems to only get older results). Also, not all changesets related to the arm port might have the "topic" set :-D
<OscarL>
s/would/could/
<zard>
So, my guess is that Zhihong Niu (unfortunately) failed
FreeFull has quit [Ping timeout: 480 seconds]
<OscarL>
boody24[m]: looking at the latest merged arm64 changesets and their reviews, will also show you who are the people more active on that front.
<zard>
They don't appear to be on the list of people who successfully completed their projects in 2022
<zard>
Yeah, I kinda doubt that there are many people on IRC that are active on arm
<zard>
I certainly don't know much about it ;)
<boody24[m]>
hmm
<OscarL>
13m24.700s for an *unoptimized* build of Python3.14a2 on 32 bits. Not bad. (was done while using haikuporter with RAMFS for workdirs).
FreeFull_ is now known as FreeFull
FreeFull has quit []
FreeFull has joined #haiku
<boody24[m]>
OscarL: yes, I was looking for answer to what's missing rather than what exists so I can connect the dots
<boody24[m]>
thanks anyway :)
<OscarL>
IIRC, on the forum was at least some discussion regarding ARM port.
<zard>
Perhaps the easiest way to find it out is just to build it and try to run it yourself
<OscarL>
maybe you'lll have better luck asking there.
<zard>
Doubtless, you would run into problems. Maybe it doesn't boot. Or maybe it does but a lot of programs don't work
<zard>
But I think the problems would be very obvious
imrahil has left #haiku [Error from remote client]
<waddlesplash>
OscarL: how's "git status" performance after the recent changes?
<waddlesplash>
it should at least be way better once the disk cache is hot
MrSun has quit [Ping timeout: 480 seconds]
<OscarL>
hi waddlesplash! On my "current" setup, I'm on beta5, using mostly RAMFS, plus every "git speed up trick" I already had... AND I'm on SSD, so even cold-boot "git status" has massively improved for me (10s on HaikuPorts).
<OscarL>
But I still have my "early" setup on HDD available, so I will definitively give your latest changes a spin! Thanks for your work!
<waddlesplash>
np :)
<waddlesplash>
"git status" is now 5s for me on the Haiku tree
<waddlesplash>
however, almost all that is lock contention in the block cache
<waddlesplash>
it's kind of nuts how much contention actually, there's over a minute of waits total
<OscarL>
ah, let me see what I get on Haiku tree now, for comparison (SSD still).
<waddlesplash>
fixing that problem won't be too easy due to how the block cache lock works
<waddlesplash>
but at least once we have all the inodes read in, it's very fast (0.3-0.5 seconds)
<OscarL>
32.156s for "git status" on haiku-tree here (VBox / SSD / beta5 32 bits)
<waddlesplash>
and second run?
<OscarL>
second run: 5.474s
<waddlesplash>
ok. will be interesting to hear how much faster the second run is after recent changes
<OscarL>
mmm, and both numbers are even with /me using PowerSaving mode on the Windows host (forgot to change it before testing the cold-start, oops)
<OscarL>
I'll switch to hdd soon, and share the numbers I get there.
<waddlesplash>
no need to test too extensively, just the quick numbers are interesting
<waddlesplash>
I am most interested in the second run performance difference, which should not hit disk IO at all
<OscarL>
rebooting :-)
<Begasus[m]>
Got it build by disabling the plugin for cmake, but not ready for any project so far :) https://0x0.st/Xnfu.80.png
<OscarL>
k, gathered data for hrev58287, now will update to latest nightly, and run things again.
<bbjimmy_64>
waddlesplash i'm back
<bbjimmy_64>
The error is still on screen
<waddlesplash>
ok
<bbjimmy_64>
now what?
<waddlesplash>
first thing to do: "threads | grep I/O"
<waddlesplash>
is one of these threads "scsi notifier"?
<waddlesplash>
ok. can you try and filter out all the ones I mentioned above?
<waddlesplash>
most of them may be "umtx", in which case you can add another grep -v
<waddlesplash>
"threads | grep -v port | grep -v umtx"
_-Caleb-_ has left #haiku [#haiku]
<bbjimmy_64>
still a long list
<waddlesplash>
what are some of the "cvar:" types remaining?
<waddlesplash>
page, obje, etc.?
_-Caleb-_ has joined #haiku
<bbjimmy_64>
page , unde, obje, dcp, I/O, hid and unnamed
<waddlesplash>
how many I/O? only one, "scsi notifier"?
<waddlesplash>
and how many "page"?
<bbjimmy_64>
only 1, scsi_scheduler
<bbjimmy_64>
2 page, page writer, page demom
OscarL has joined #haiku
<bbjimmy_64>
*daemon
<OscarL>
Anarchos: thanks (lost internet while trying to reply to you). No idea why it doesn't works. It is a public repo, and shows up fine if I'm logged in. Sigh.
<waddlesplash>
bbjimmy_64: get backtraces of those, I suppose
<waddlesplash>
but odds are they're not doing anything of note
<bbjimmy_64>
how?
<waddlesplash>
bt <id>
<waddlesplash>
id is 2nd column
<waddlesplash>
bbjimmy_64: oh, there's one other thing you can check
<waddlesplash>
see if anything else is calling BFS
<waddlesplash>
"image | grep bfs", this will have 1 output line
<waddlesplash>
the 2nd column "(number)" is an ID
<waddlesplash>
then run: image <ID>
<waddlesplash>
this will have more output, there will be 2 fields called "text.start" and "text.size"
<waddlesplash>
run the "calling" command, like this: "calling text.start text.start+text.size"
<waddlesplash>
you can leave the "+" in the command, it will do the math for you
<waddlesplash>
e.g. if text.start is 0x1234, and .size is 0x1000, then "calling 0x1234 0x1234+0x1000" will do the trick
<waddlesplash>
these will be 64-bit kernel addresses, so they will start with 8 "f"s though after the "0x"
<OscarL>
Anarchos: btw, IIRC, some time ago you asked me if I had success building Python on RAMFS. At the time I said I had only tested it ouside of HaikuPorter...
<OscarL>
earlier today I was able to build Python 3.14.0-alpha2 on beta5 32 bits, using HaikuPorter and RAMFS. Build went just fine!
<OscarL>
was an unoptimized build (and run inside a VirtualBox VM), so not sure if (or how much) it affected build timings, but it sure helps a lot with init / cleanup / repeated builds.
<Anarchos>
OscarL always nice to see improvements, i was expecting a x15 speedup :)
<OscarL>
success rate is not 100% though, as in previous days I couldn't complete a build of mozc using that setup on 32 bits (worked fine on 64 bits).
<bbjimmy_64>
I took some photos of the backtraces, but will have to upload them after work.
<bbjimmy_64>
waddlesplash ^^
HaikuUser has joined #haiku
<waddlesplash>
OK
HaikuUser has quit []
<OscarL>
Anarchos: I guess my CPU is too slow, and RAM too (DDR2-800 :-D) to get much of an speed up by any means :-D
gouchi has quit [Remote host closed the connection]
imrahil has left #haiku [Error from remote client]
_justin_kelly9759578656 has quit [Read error: Connection reset by peer]
OscarL has joined #haiku
_justin_kelly9759578656 has joined #haiku
<OscarL>
re: RAMFS and possible optimizations... Understood. In any case, when it works, still a huge help to reduce HDD/SSD trashing with HaikuPorter.
<Anarchos>
is there a way to know which process is using hdd a lot ?
zardshard has left #haiku [Disconnected: Replaced by new connection]
zardshard has joined #haiku
<OscarL>
(on VoidLinux, seeing what updates are available) mmm, 1/3 of the total 850 MBs of updates are of firmware for hardware that this netbook will never have. /me declines the update.
hightower3 has joined #haiku
Anarchos has quit [Quit: Vision[]: i've been blurred!]