<waddlesplash>
I could be misremembering but I thought that is supposed to work at this stage
<waddlesplash>
could be something locked up in an infinite loop
zdykstra has quit [Quit: WeeChat 4.5.1]
zdykstra has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
Nasina has joined #haiku
nephele_xmpp has left #haiku [Disconnected: Hibernating too long]
mmu_man has quit [Ping timeout: 480 seconds]
<Ellenor2000>
(legally this is a joke) do i need to make sure the SRBs are lit before continuing
<waddlesplash>
the system should take care of that for you (if applicable on your hardware(
<waddlesplash>
:P
<Ellenor2000>
that obviously assumes the boosters have a driver, which I don't think they do
<Ellenor2000>
when I plugged in the launch module it didn't detect anything
<Ellenor2000>
hope your day goes well
OscarL has joined #haiku
<OscarL>
My Linux installs are either too old (5.15.x, missing some needed packages), or too unstable (EndeavourOS 6.13.x, kept freezing/locking up the UI).
<OscarL>
As I don't really use them... I don't want to spend money trying to updating them (would need to buy 4G data). So... I won't be able to test build times there, sorry.
<OscarL>
waddlesplash: ^^^ (I tried for about 3 and a half hours :-D)
<OscarL>
(The shell on that older Ubuntu felt really, really snappy for some reason.)
<waddlesplash>
ouch
<OscarL>
building hrev58650 now to at least compare some builds with that last malloc related changes.
<waddlesplash>
well, thanks for trying!
<OscarL>
np. I was curious too. but not *that* curious it seems :-D
<OscarL>
"/src/system/kernel/vm/vm.cpp:4278:18: warning: variable 'lastCache' set but not used [-Wunused-but-set-variable]"
xqkuang has quit [Quit: Vision[]: i've been blurred!]
<OscarL>
a -Wint-conversion, couple of -Wmaybe-uninitialized, and a -Wint-conversion also on openbsd/malloc.c (rest of warnings there seem harmless)
phcoder has quit [Quit: Connection closed for inactivity]
nephele_xmpp has joined #haiku
talos has quit [Remote host closed the connection]
erysdren has joined #haiku
HaikuUser has joined #haiku
talos has joined #haiku
HaikuUser has quit []
nephele_xmpp has left #haiku [Error from remote client]
linuxmaster has joined #haiku
<erysdren>
i wonder what category trenchbroom would be packaged under
<erysdren>
on haikuports
OscarL has quit [Ping timeout: 480 seconds]
* coolcoder613
waves
nephele_xmpp has joined #haiku
Nasina has quit [Read error: Connection reset by peer]
<augiedoggie>
probably dev-games
<waddlesplash>
I can't believe it
<augiedoggie>
me either
<erysdren>
whats up
<waddlesplash>
$ time jam -q -j4 libroot.so
<waddlesplash>
real 0m23.448s
<waddlesplash>
the absolute best time I could get for this as little as a month ago was like 27.4 seconds
<waddlesplash>
(on Haiku)
<waddlesplash>
Linux meanwhile is somewhere around 16.0, so we're still a ways off ... but we're closing in
<erysdren>
i don't get it, what does that command do?
<waddlesplash>
"time" measures time (lol), "jam ..." is the tool we use to build Haiku
<waddlesplash>
so that tells jam to rebuild "libroot.so"
<erysdren>
oh i see
<waddlesplash>
I've cut out all the logs in the middle of course
<augiedoggie>
screen_blanker actually works pretty well as a login screen once you get it set up right
nephele_xmpp has left #haiku [Error from remote client]
<augiedoggie>
had to ssh into my VM and kill the process because the "Network password" feature doesn't work
bmp has joined #haiku
* augiedoggie
considers making a how-to post on the forums
piku has quit []
pvalue has joined #haiku
pabs has quit [Remote host closed the connection]
pabs has joined #haiku
<erysdren>
how do you use qt6 in a haikuporter script?
frkazoid333 has quit [Ping timeout: 480 seconds]
<erysdren>
im trying to package TrenchBroom, i just dunno how to tell haikuporter to allow the use of qt6
frkazoid333 has joined #haiku
Nasina has joined #haiku
nephele_xmpp has joined #haiku
<erysdren>
actually, ill be back later to work on this
<erysdren>
brb
erysdren has quit [Quit: Konversation terminated!]
OscarL has joined #haiku
<OscarL>
waddlesplash: results on VMware are a bit weird: https://bpa.st/HL4Q
<waddlesplash>
that is weird
<waddlesplash>
because here the change seems to make things faster
<OscarL>
(host is Win10, set to "High performance" mode, host *very lightly* used during builds.
Nasina has quit [Read error: Connection reset by peer]
<waddlesplash>
the difference looks like almost all sys time
<OscarL>
I think I'll be able to test against Void Linux (6.6.x kernel, gcc 13.2), but on a different machine (dual core Celeron N4020).
<waddlesplash>
ah no it's user time
<OscarL>
(need to update my Haiku install there still)
<OscarL>
(and use "sneaker-net" to move the needed packages, so it might take me at least one day to share results)
<waddlesplash>
VMware may just be too host-dependent in some ways I guess
<waddlesplash>
I test in it but I usually rerun tests multiple times because results fluctuate by 5-10%
<OscarL>
waddlesplash: tomorrow will test that hrev58650_dirty-1 on this Phenom, but on bare metal.
<waddlesplash>
excellent, that will probably be a better test
<OscarL>
(we have previous results on the post I did on the forum)
Nasina has joined #haiku
<OscarL>
bah... let's run the Phenom bare metal now... back in 40/45 minutes I hope (if I don't lose power to due to the storm :-D)
OscarL has quit [Quit: "I'll be back!"]
Nasina has quit [Remote host closed the connection]
Nasina has joined #haiku
tombhadAC has quit [Ping timeout: 480 seconds]
frkazoid333 has quit [Ping timeout: 480 seconds]
nephele_xmpp has left #haiku [Error from remote client]
Nasina has quit [Read error: Connection reset by peer]
<erysdren>
is the "middle" version required for app_resources in an rdef, btw? sorry to butt in
<Begasus[m]>
bugger ... skipping build :/
<Begasus[m]>
erysdren (@_oftc_erysdren:matrix.org) you can just set the minor to "0" or skip it I guess
<OscarL>
Begasus[m]: I had started to update openblas too at first, but decided to leave that for later (specially after noticing the issues with lapack's .pc files and the "duplicated" cblas.h)
<Begasus[m]>
should be some that use it, use the "source" Luke! :)
janking has joined #haiku
<Begasus[m]>
reinsttalled the "old" version from backup, but I guess I need to or force a rebuild or build clean
<janking>
Good morning
<Begasus[m]>
morning janking
<janking>
:)
<OscarL>
Begasus[m]: with work-dirs on RAMFS, my builds are mostly clean by default (sometimes even *too* clean :-P)
<Begasus[m]>
lots of -Wno-multichar messages for lapack :P
<Begasus[m]>
ran list_work... yesterday, so all clean here too :)
<OscarL>
yeah... that's why I opened an issue for that... pretty annoying.
<OscarL>
(gfortran is way too spammy with that warning)
<Begasus[m]>
👍
* OscarL
waves at erysdren, janking, and phschafft.
<BiPolar>
so I only have the new lapack and numpy2
<Begasus[m]>
openblas 0.3.23-2 (same here)
<Begasus[m]>
yeah 2.2.1 for numpy
coolcoder613_ has joined #haiku
coolcoder613 has quit [Ping timeout: 480 seconds]
Oscar-L has quit [Ping timeout: 480 seconds]
Oscar-L has joined #haiku
Oscar-L is now known as OscarL
<Begasus[m]>
k, something is using arpack here ... need to track that down
<Begasus[m]>
looks like lib:libarmadillo
<OscarL>
you build octave not long ago, IIRC.
<OscarL>
that and armadillo come up with "pkgman search -r arpack"
BiPolar has quit [Ping timeout: 480 seconds]
<OscarL>
still, one would expect the chroot to not be affected by the _devel packages installed system-wide, no?
<Begasus[m]>
yeah, it shouldn't be involved in chroot
<OscarL>
(in any case, I don't have any arpack related .hpkg in my system)
<Begasus[m]>
it takes a few other packages down here also, so don't want to tear it down too fast :)
<OscarL>
can you paste the lines from syslog about the missing symbol?
<OscarL>
(running "lgrep slamch_" here, just in case)
<Begasus[m]>
KERN: runtime_loader: /boot/system/lib/python3.10/vendor-packages/scipy/sparse/linalg/_eigen/arpack/ arpack.cpython-310.so: Could not resolve symbol 'slamch'
<Begasus[m]>
looks arpack is messing here :/
<OscarL>
"/boot/system/lib/liblapack.so.3.10.0:000000000017a0c0 T slamch_"
<Begasus[m]>
yeah also in lapack
<OscarL>
"boot/system/lib/liblapacke.so.3.10.0: U slamch_"
<Begasus[m]>
that too :)
<OscarL>
isn't that suppsed to read as: arpack and lapacke both USE that symbol (U-ndefined there), but liblapack actuall defines it?
nephele_xmpp has joined #haiku
<OscarL>
let me see what I get for my "arpack.cpython-310.so"
janking has quit [Quit: Vision[]: i've been blurred!]
Captain0xff has joined #haiku
<Begasus[m]>
reboot
<Begasus[m]>
bugger, no 32bit nightly in the Vm's
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
janking has joined #haiku
jmairboeck has joined #haiku
bbjimmy has joined #haiku
<Captain0xff>
hi :)
<janking>
hi
MisthaLu has joined #haiku
ADS_Sr has quit [Remote host closed the connection]
erysdren has quit [Quit: Konversation terminated!]
Captain0xff has quit [Quit: Captain0xff]
Captain0xff has joined #haiku
Captain0xff has quit [Quit: Captain0xff]
<phschafft>
hm.
<Begasus[m]>
low
janking has quit [Quit: Vision[]: i've been blurred!]
<phschafft>
high-z
<Begasus[m]>
bugger ... outranged :P
freddietilley has quit [Quit: WeeChat 4.5.0]
<phschafft>
;)
bjorkintosh has joined #haiku
phcoder has joined #haiku
<phcoder>
waddlesplash: I figured out SMP on my darp10-b. Other than x2apic another problem was that gdt and pml4 are above 4GiB and haiku tried to load it using 32-bit cider on AP
<Begasus_32>
ok, let's bump this one to nightly ;)
<Captain0xff>
I am asking the same question in the forum since I need to restart my computer to see if the new configuration is working or not and won't be able to see any replies here
Captain0xff has quit [Quit: Captain0xff]
mmu_man has quit [Ping timeout: 480 seconds]
yann64 has quit [Quit: Vision[]: i've been blurred!]
<Begasus_32>
crossing fingers :)
Begasus_32 has quit [Quit: Vision[]: i've been blurred!]
<Begasus_32>
so I guess we call the nighlies R1BP6?
DKnoto has quit [Ping timeout: 480 seconds]
DKnoto has joined #haiku
bjorkint0sh has joined #haiku
bjorkintosh has quit [Ping timeout: 480 seconds]
janking has joined #haiku
talos has quit [Quit: Konversation terminated!]
talos has joined #haiku
Begasus_32 has quit [Quit: Vision[]: i've been blurred!]
zard has joined #haiku
erysdren has joined #haiku
<waddlesplash>
phcoder: still hangs when going to desktop though?
drusatori_alt has quit [Ping timeout: 480 seconds]
gouchi has joined #haiku
gouchi has quit []
diver has quit [Quit: Leaving.]
ADS_Sr has joined #haiku
janking has quit [Quit: Vision[]: i've been blurred!]
<waddlesplash>
anyone here have a 32-bit Haiku install and can test a change?
<phcoder>
waddlesplash: yes. Any pointers?
<waddlesplash>
not really :/
<waddlesplash>
my initial suggestion would be a watchdog that enters KDL manually after a fixed time
<waddlesplash>
just add another kernel_daemon that checks system_time periodically and panic()s
<waddlesplash>
if that doesn't actually drop to KDL then we know the system has really locked up, if it does then debugging should be possible from there
nexus6 has joined #haiku
HaikuUser has joined #haiku
HaikuUser has quit []
coronavirus has joined #haiku
<x512[m]>
waddlesplash: There are no method to directly read APIC frequency?
<waddlesplash>
there is on recent CPUs in CPUID 0x15/0x16
<waddlesplash>
same as for TSC frequency
<waddlesplash>
but I didn't implement this (yet). The computations are a little confusing
<waddlesplash>
korli is the one who implemented the version of this for TSC
<waddlesplash>
I think I'll probably just leave it as-is for now and let someone else attempt that, I only have one machine where it might even work anyway
<x512[m]>
For RISC-V it is easy, timer frequency is stored in FDT.
<waddlesplash>
yeah, it's not on x86
<x512[m]>
On my AMD x86 machine there are significant time drift.
neoncortex has quit [Ping timeout: 480 seconds]
<x512[m]>
I suspect inaccurate timer frequency detection.
mmu_man has quit [Ping timeout: 480 seconds]
<waddlesplash>
time/clock drift means wrong TSC, not wrong APIC
<waddlesplash>
wrong APIC just means too many or too few timer interrupts
<waddlesplash>
TSC is what is actually used to check how much time has passed, including in timer interrupt routine
<jmairboeck>
the values are closer together than the first time
<jmairboeck>
this time I shut down the VM and booted it fresh. The first time was from a reboot (but I guess that shouldn't make a difference)
<waddlesplash>
still a very wide variation
<waddlesplash>
oh well. but at least it still "looks sane"
<Begasus[m]>
closing down here
<Begasus[m]>
cu peeps!
yote has joined #haiku
yote is now known as Yoke
MisthaLu has quit [Ping timeout: 480 seconds]
digitalbox98 has quit [Quit: Vision[]: i've been blurred!]
<phcoder>
waddlesplash: I put a panic on every billionth interrupt. It has triggered couple of times during the boot with a functional kdebug. But at a hang it doesn't happen anymore
<waddlesplash>
so, it's locked up somewhere with interrupts disabled
<waddlesplash>
there are ways of triggering NMIs, I think
<waddlesplash>
which may help here
<waddlesplash>
another possibility: create a REALTIME priority thread, disable interrupts, busy-loop in it
<waddlesplash>
and then start dprintf() when things quit responding
<waddlesplash>
actually. There is a related bug: if some CPU quits responding to ICIs then we can't enter KDL
<waddlesplash>
I have had that on my TODO list for months to fix
<waddlesplash>
for now you can make the "start kernel debugger" message ASYNC which may help
<phcoder>
Do you have a patch?
<waddlesplash>
not right now, no
<waddlesplash>
haven't worked on this yet
<phcoder>
I mean for async part
KapiX has joined #haiku
<waddlesplash>
kernel/debug/debug.cpp line 978
<waddlesplash>
change FLAG_SYNC to FLAG_ASYNC
yann64 has quit [Quit: Vision[]: i've been blurred!]