Re: please test pcm patch
On Thu, 2003-11-27 at 02:54, Mathew Kanner wrote: > Please test a pcm patch that releases the channel lock around > calls to uio move. This is a more complete patch than the previous > one as it also does the _read routine. I will ask the RE to commit > this if I hear a couple of "it works". I didn't have the panics, and the patch works fine so far on my hardware. pcm0: port 0x530-0x537 on acpi0 device_probe_and_attach: pcm0 attach returned 6 pcm0: port 0x530-0x537 on acpi0 device_probe_and_attach: pcm0 attach returned 6 -- FreeBSD 5.2-BETA i386 5:30pm up 8 mins, 1 user, load averages: 0.65, 0.76, 0.43 signature.asc Description: This is a digitally signed message part
Re: Was: More ULE bugs fixed. Is: Mouse problem?
On Wed, 2003-11-05 at 17:04, Jeremy Messenger wrote: > I don't get any mouse lag anymore in the 'cd /usr/src ; make clean ; make > cleandir' and the beginner of 'portupgrade -ra'. I did the hard test; I > have Gnome2, Opera 7 (linux version), several gvim, several gnome-terminal > tabs, pan and even use totem to watch a movie (700mb) in full screen mode > ;-).. While watching the movie and it doesn't get lag that much only while > unpack the large tarballs like mozilla. While the build and clean, it only > has very very little lag like one to two time(s) but the rest are pretty > smooth. The mouse doesn't get any lag at all. Yup.. Latest SCHED_ULE commit did it for me. My system now is fast and responsive. I ran similar tests, and everything ran quite smoothly with Gnome Desktop environment. Quite drastic change, system was practically unusable before this with SCHED_ULE. However, under higher CPU usage (90%+) and load (>2) the mouse starts to feel sluggish. While still responsive enough to use, it still slows down enough to be noticeable. psm0: irq 12 on atkbdc0 psm0: model IntelliMouse, device ID 3 Dual Pentium III Intel 440BX > BTW: I don't use KSE because of Nvidia driver. Confirmed. It now works just fine (so far) with KSE. signature.asc Description: This is a digitally signed message part
Re: ULE top(1) times...
On Wed, 2003-10-29 at 17:45, Sean Chittenden wrote: > A ULE UP kernel compiled fresh tonight (2003-10-28). Not critical, > but certainly interesting. -sc Anybody running ULE + KSE on an SMP kernel without any problems? My symptoms are as such: - idle and total cpu usage never goes above 54% - system is extremely slow (compared to 4BSD) - I'll provide follow up timings. - jerky mouse in X (as reported by others), better but unlike others, it's constant. What info should I give to help in tracing this problem? -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 6:07PM up 7:01, 4 users, load averages: 0.51, 0.31, 0.23 signature.asc Description: This is a digitally signed message part
Re: -pthread deprecated, but when?
On Tue, 2003-09-09 at 16:07, Doug Barton wrote: > Any occurences of -lc_r should be changed to ${PTHREAD_LIBS} too. Seems that -lc_r is set by default in bsd.port.mk for ${PTHREAD_LIBS}. Does it mean that we should be able to specify a pthread library in future as a make option for installing ports? Or use libmap.conf(5) to relink libc_r to one of the threading libraries? -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 8:57PM up 21:31, 2 users, load averages: 0.09, 0.16, 0.16 signature.asc Description: This is a digitally signed message part
kernel build failure: mga_state.c
cc -c -O -pipe -march=pentium3 -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -fformat-extensions -std=c99 -nostdinc -I- -I. -I/usr/src/sys -I/usr/src/sys/dev -I/usr/src/sys/contrib/dev/acpica -I/usr/src/sys/contrib/ipfilter -I/usr/src/sys/contrib/dev/ath -I/usr/src/sys/contrib/dev/ath/freebsd -D_KERNEL -include opt_global.h -fno-common -finline-limit=15000 -mno-align-long-strings -mpreferred-stack-boundary=2 -ffreestanding -Werror /usr/src/sys/dev/drm/mga_state.c /usr/src/sys/dev/drm/mga_state.c: In function `mga_g400_emit_state': /usr/src/sys/dev/drm/mga_state.c:278: warning: inlining failed in call to `mga_g400_emit_pipe' /usr/src/sys/dev/drm/mga_state.c:387: warning: called from here /usr/src/sys/dev/drm/mga_state.c:163: warning: inlining failed in call to `mga_g400_emit_tex0' /usr/src/sys/dev/drm/mga_state.c:397: warning: called from here *** Error code 1 Stop in /usr/obj/usr/src/sys/DAEMON. *** Error code 1 -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 11:23AM up 1 day, 2:15, 1 user, load averages: 0.65, 0.56, 0.39 signature.asc Description: This is a digitally signed message part
Re: background processes stuck in locks with ULE
On Sun, 2003-07-20 at 15:09, Kris Kennaway wrote: > The process stats are not updated properly under ULE at the moment. > If a process takes up 60% of CPU and then sleeps, top will continue to > show it at 60% until the next time it runs. It does not in fact > continue to use CPU. In my case if I kill the process (that appears to be stuck in *Giant) the system immediately starts to improve in performance. So it seems that this process (usually a daemon) is still running in bg and eating up more cycles than necessary. Even then performance is extremely slow compared to 4BSD. However in single user mode (with no bg processes), performance seems normal. When I mean slow.. it means being able to read line by line as ipfw rules are being added by a script, or an ls -l output on a large directory. I'm going to compile a UP kernel, and report if it makes a difference. -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 12:55AM up 5:36, 1 user, load averages: 0.45, 0.39, 0.34 signature.asc Description: This is a digitally signed message part
background processes stuck in locks with ULE
Trying out ULE scheduling on an SMP machine causes background processes to be stuck in locks. One or two processes will always get stuck in *Giant (and takes up like 60% of lock) and if these are killed, then other processes are always taking up a total of around 10% in lock, which makes the system crawl. Anybody else seeing this on an SMP machine with ULE scheduler? -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 1:08PM up 40 mins, 2 users, load averages: 1.53, 1.62, 1.33 signature.asc Description: This is a digitally signed message part
Re: build error building cvs doc?
On Sun, 2003-07-13 at 11:55, John Reynolds wrote: > Anybody seeing this same error during "buildworld"? > ===> gnu/usr.bin/cvs/doc > makeinfo --no-split -I /usr/src/gnu/usr.bin/cvs/doc -I > /usr/src/gnu/usr.bin/cvs/doc/../../../../contrib/cvs/doc > /usr/src/gnu/usr.bin/cvs/doc/../../../../contrib/cvs/doc/cvs.texinfo -o cvs.info > makeinfo --no-split -I /usr/src/gnu/usr.bin/cvs/doc -I > /usr/src/gnu/usr.bin/cvs/doc/../../../../contrib/cvs/doc > /usr/src/gnu/usr.bin/cvs/doc/../../../../contrib/cvs/doc/cvsclient.texi -o > cvsclient.info > gzip -cn cvsclient.info > cvsclient.info.gz > gzip -cn cvs.info > cvs.info.gz These is actually a segfault in sh. It will appear at other times too.. in other places for other /bin stuff like ln. This happens with -jN but it occured for me without it also. buildworld NOCLEAN=1 multiple times will get you through. I'm going to try a buildworld again later today, and see if this still occurs with new world. Does buildworld use current sh? -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 1:45PM up 14 mins, 2 users, load averages: 1.17, 0.62, 0.33 signature.asc Description: This is a digitally signed message part
Re: Panic in scheduler code with SCHED_ULE during boot tomulti-user.
On Fri, 2003-07-04 at 16:55, Jeff Roberson wrote: > Excellent debugging. Can you tell me whats in kseq_cpu[0] and > kseq_cpu[1]? I think I may know what the problem is. Do you have some > negative niced processes or some positive nice processes? Just tried the fix, and tried out ULE with KSE (universal map). In single user mode, it works great (build world -j8) but it has performance problems once I get into multiuser mode where there are background processes. Basically everything just slows down and becomes jerky, and some idle processes are pegging idle cycles on one CPU. It is still jerky even when I kill them and load drops (ie. 0.1% user). Can you tell me what info you need, for me to help find out what is causing background processes to have such an adverse affect on performance with KSE? ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
libkse now running quite well on smp
First of all thanks to Julian Elisher who gave hints on what might be wrong (userland/kernel), even though I was a blockhead for not enabling debugging options in the kernel before posting. It help me track down broken stuff in userland first. I hope I can give proper helpful debug info next time I get problems. Tested everything else throughly, before enabling kse one app at a time. Current uptime of 23hrs with these apps running with libkse.so.1, basically a usable Gnome 2.2 desktop environment: jdk1.4.1 (tomcat/netbeans) openoffice 1.03 evolution 1.4 mozilla-firebird 0.6 metacity xmms gnome-terminal nautilus vte dia gnome-panel gkrellm mysql40 (not heavilly tested yet) Wow.. with recent updates, you can now see the threads with top(8): Mozilla-Firebird 826 kaeru 96 0 40508K 32172K select 1 0:07 0.00% 0.00% MozillaFire 826 kaeru 96 0 40508K 32172K select 1 0:07 0.00% 0.00% MozillaFire 826 kaeru 20 0 40508K 32172K kserel 1 0:07 0.00% 0.00% MozillaFire 826 kaeru 20 0 40508K 32172K kserel 1 0:07 0.00% 0.00% MozillaFire Great work by the KSE team! Looks like we might get libpthreads as default on 5.2. ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
LOR: dev/sound/isa/mss.c and dev/sound/pcm/channel.c
I was trying to get core or some sort of trace for what I think are kse related, but I get this instead. Randomly after this during various operations, my system totally locks up (I don't know enough, whether it's related or not). lock order reversal 1st 0xc40bb540 pcm0 (sound softc) @ /usr/src/sys/dev/sound/isa/mss.c:179 2nd 0xc40bb180 pcm0:play:0 (pcm channel) @ /usr/src/sys/dev/sound/pcm/channel.c :440 Stack backtrace: backtrace(c03e3e7c,c40bb180,c405ced4,c03d57a5,c03d587f) at backtrace+0x17 witness_lock(c40bb180,8,c03d587f,1b8,a) at witness_lock+0x697 _mtx_lock_flags(c40bb180,0,c03d587f,1b8,c40b9700) at _mtx_lock_flags+0xb2 chn_intr(c405ce80,18,0,1010b400,c40bb400) at chn_intr+0x2f mss_intr(c40b9700,0,c03deda1,216,c4091d3c) at mss_intr+0x8b ithread_loop(c4099080,d68e0d48,c03dec52,30c,0) at ithread_loop+0x182 fork_exit(c022cc70,c4099080,d68e0d48) at fork_exit+0xc0 fork_trampoline() at fork_trampoline+0x1a --- trap 0x1, eip = 0, esp = 0xd68e0d7c, ebp = 0 --- -- "Optimized, readable, on time; Pick any two." FreeBSD 5.1-CURRENT i386 3:17PM up 37 mins, 1 user, load averages: 0.56, 0.54, 0.46 signature.asc Description: This is a digitally signed message part
Re: -O2 broke ppp NAT
On Fri, 2003-03-07 at 01:24, Bruce Cran wrote: > On Thu, Mar 06, 2003 at 09:22:06PM +0800, Khairil Yusof wrote: > > On Thu, 2003-03-06 at 06:00, Nuno Teixeira wrote: > > > > > For the first time I compile current-p3 -> current-p4 with > > > -march=pentium2 -O2 -mmmx -pipe and aparently everything works ok > > > except ppp -nat. NAT just don't work on my network. All machines are > > > able to ping except ftp, http, etc. > > > > I can confirm this. nat fails to work with -O2 for usr.sbin/ppp. It > > compiles cleanly though, but I don't know enough about gcc optimizations > > to find out how O2 might break it. Upon further testing (recompiling usr.sbin/ppp) with no optimizations except -O -pipe. -nat still fails to work with anything other than ICMP traffic. I'm gonna try rebuild world without -O2, but that's not gonna help trace where the problem is. :( -- Khairil Yusof <[EMAIL PROTECTED]> signature.asc Description: This is a digitally signed message part
Re: -O2 broke ppp NAT
On Thu, 2003-03-06 at 06:00, Nuno Teixeira wrote: > For the first time I compile current-p3 -> current-p4 with > -march=pentium2 -O2 -mmmx -pipe and aparently everything works ok > except ppp -nat. NAT just don't work on my network. All machines are > able to ping except ftp, http, etc. I can confirm this. nat fails to work with -O2 for usr.sbin/ppp. It compiles cleanly though, but I don't know enough about gcc optimizations to find out how O2 might break it. -- Khairil Yusof <[EMAIL PROTECTED]> signature.asc Description: This is a digitally signed message part