Re: UP kernel performance and Matt Dillon's patches
:With this commit, my Kernel seems looping in doreti_ast :(doreti_ast+0x15) (as seen by trace after Ctrl+Alt+Esc) after :displaying "mounting root device" :) : :My kernel config is on http://www.hsc.fr/~thivillo/YOKO50 : :i have rerun cvsup to be sure that all things are in sync, ipl.s shows: : :$FreeBSD: src/sys/i386/isa/ipl.s,v 1.34 2000/03/29 06:15:43 dillon Exp $ : :-- :Alain Thivillon -+- [EMAIL PROTECTED] -+- Hervé Schauer Consultants Ah ha! I've been trying to track that down with two other people but I wasn't getting enough debugging information out of them. This is weird, it isn't happening to me at all. But I think you've given me enough info to locate the problem. -Matt Matthew Dillon [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
UP kernel performance and Matt Dillon's patches
I'm running a UP machine with Matt's latest changes. I was just compiling a new kernel and noticed the my PS/2 mouse under X was very sluggish. I never noticed this sort of behavior before the changes, even while doing ``make -j8 buildworld''. The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra SCSI adapter on my motherboard, in case it matters. Looks like something has been verschlimmbessert (a wonderful German word which means "made worse through improvement" :) Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
Alfred Perlstein writes: * Gary Jennejohn [EMAIL PROTECTED] [000328 14:04] wrote: I'm running a UP machine with Matt's latest changes. I was just compiling a new kernel and noticed the my PS/2 mouse under X was very sluggish. I never noticed this sort of behavior before the changes, even while doing ``make -j8 buildworld''. The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra SCSI adapter on my motherboard, in case it matters. Looks like something has been verschlimmbessert (a wonderful German word which means "made worse through improvement" :) This is unlikely as a UP kernel doesn't seem to compile after Matt's changes (no offence Matt, I know you're getting to it), when was the last time you didn't see this sluggish behavior, how are you compiling your kernel? What's your Id line for sys/i386/i386/mplock.s ? Mine is: * $FreeBSD: src/sys/i386/i386/mplock.s,v 1.30 2000/03/28 07:16:15 dillon Exp $ I have * $FreeBSD: src/sys/i386/i386/mplock.s,v 1.31 2000/03/28 18:06:37 dillon Exp $ Matt fixed the bug which was preventing compilng a UP kernel. So I guess it is likely, after all. --- Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
I havn't noticed this behavior... or any other performance hits and I'm running a kernel that was cvsupped about 5 10 minutes ago.. and recompiled about 5 minutes ago... = | Kenneth Culver | FreeBSD: The best OS around.| | Unix Systems Administrator | ICQ #: 24767726 | | and student at The | AIM: muythaibxr | | The University of Maryland, | Website: (Under Construction) | | College Park. | http://www.wam.umd.edu/~culverk/| = On Wed, 29 Mar 2000, Gary Jennejohn wrote: Alfred Perlstein writes: * Gary Jennejohn [EMAIL PROTECTED] [000328 14:04] wrote: I'm running a UP machine with Matt's latest changes. I was just compiling a new kernel and noticed the my PS/2 mouse under X was very sluggish. I never noticed this sort of behavior before the changes, even while doing ``make -j8 buildworld''. The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra SCSI adapter on my motherboard, in case it matters. Looks like something has been verschlimmbessert (a wonderful German word which means "made worse through improvement" :) This is unlikely as a UP kernel doesn't seem to compile after Matt's changes (no offence Matt, I know you're getting to it), when was the last time you didn't see this sluggish behavior, how are you compiling your kernel? What's your Id line for sys/i386/i386/mplock.s ? Mine is: * $FreeBSD: src/sys/i386/i386/mplock.s,v 1.30 2000/03/28 07:16:15 dillon Exp $ I have * $FreeBSD: src/sys/i386/i386/mplock.s,v 1.31 2000/03/28 18:06:37 dillon Exp $ Matt fixed the bug which was preventing compilng a UP kernel. So I guess it is likely, after all. --- Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
:I'm running a UP machine with Matt's latest changes. I was just :compiling a new kernel and noticed the my PS/2 mouse under X was very :sluggish. I never noticed this sort of behavior before the changes, :even while doing ``make -j8 buildworld''. : :The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra :SCSI adapter on my motherboard, in case it matters. : :Looks like something has been verschlimmbessert (a wonderful German :word which means "made worse through improvement" :) : : :Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] Ok, I'll take a look at it... the most likely cause is that I somehow broke need_resched. Not impossible, I'll check it out. -Matt Matthew Dillon [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
On Tue, Mar 28, 2000 at 11:37:55PM +0200, Gary Jennejohn wrote: I'm running a UP machine with Matt's latest changes. I was just compiling a new kernel and noticed the my PS/2 mouse under X was very sluggish. I never noticed this sort of behavior before the changes, even while doing ``make -j8 buildworld''. The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra SCSI adapter on my motherboard, in case it matters. Looks like something has been verschlimmbessert (a wonderful German word which means "made worse through improvement" :) I have noticed this too. Except on an SMP kernel with IDE drives. My keyboard is lagging in the same manner with high cpu loads. Abit bp6 + 2xceleron 500 IBM DJNA 13.5gig drive on the HPT366 -Chris -- [EMAIL PROTECTED] [EMAIL PROTECTED] Abbotsford, BC, Canada To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Very weird assembly failure (was Re: UP kernel performance and Matt Dillon's patches)
I found a couple of minor nits, but only one real bug. In i386/swtch.s I forgot to change out a WANT_RESCHED for AST_RESCHED: sw1a: call_chooseproc /* trash ecx, edx, ret eax*/ testl %eax,%eax CROSSJUMP(je, _idle, jne) /* if no proc, idle */ movl%eax,%ecx xorl%eax,%eax andl$~WANT_RESCHED,_astpending The problem is that a kernel build is not reporting any errors! WANT_RESCHED does not exist at all, anywhere. If I change it to a garbage name the kernel still builds. I don't get it. In anycase, please try changing WANT_RESCHED to AST_RESCHED in i386/i386/swtch.s and see if that fixes the reported performance problems. If WANT_RESCHED defaults to 0 by being undefined, then the reschedule flag is never cleared when a context switch is made and this could certainly lead to problems. I also found a movb that had to be a movl, but since only the low bits were being used anyway this would not have caused any problems. -Matt To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
Matthew Dillon writes : :I'm running a UP machine with Matt's latest changes. I was just :compiling a new kernel and noticed the my PS/2 mouse under X was very :sluggish. I never noticed this sort of behavior before the changes, :even while doing ``make -j8 buildworld''. : :The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra :SCSI adapter on my motherboard, in case it matters. : :Looks like something has been verschlimmbessert (a wonderful German :word which means "made worse through improvement" :) : : :Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] Ok, I'll take a look at it... the most likely cause is that I somehow broke need_resched. Not impossible, I'll check it out. I'm seeing the same symptoms while doing a make - my console performance is very jumpy and sluggish. Geoff. -- Geoff Rehmet, The Internet Solution [EMAIL PROTECTED]; [EMAIL PROTECTED]; [EMAIL PROTECTED] tel: +27-83-292-5800 To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Very weird assembly failure (was Re: UP kernel performance and Matt Dillon's patches)
On Tue, Mar 28, 2000 at 07:48:00PM -0800, Matthew Dillon wrote: I found a couple of minor nits, but only one real bug. In i386/swtch.s I forgot to change out a WANT_RESCHED for AST_RESCHED: The problem is that a kernel build is not reporting any errors! WANT_RESCHED does not exist at all, anywhere. If I change it to a garbage name the kernel still builds. I don't get it. In anycase, please try changing WANT_RESCHED to AST_RESCHED in i386/i386/swtch.s and see if that fixes the reported performance problems. If WANT_RESCHED defaults to 0 by being undefined, then the reschedule flag is never cleared when a context switch is made and this could certainly lead to problems. Changing it to AST_RESCHED did not fix the problem for me. -Chris -- [EMAIL PROTECTED] [EMAIL PROTECTED] Abbotsford, BC, Canada To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: Very weird assembly failure (was Re: UP kernel performance and Matt Dillon's patches)
: problems. If WANT_RESCHED defaults to 0 by being undefined, then : the reschedule flag is never cleared when a context switch is made : and this could certainly lead to problems. : :Changing it to AST_RESCHED did not fix the problem for me. : :-Chris Ok. I'm seeing the same behavior here too over an ssh link. I'm pretty sure I broke need_resched and the processes are incorrectly getting too much cpu in the face of a wakeup. I just have to find out where. -Matt Matthew Dillon [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
Yeah, I was wrong before.. I just had to really hit the system hard before I noticed this behavior... it get's pretty bad... the mouse get's jumpy, and the keyboard input is really slow... = | Kenneth Culver | FreeBSD: The best OS around.| | Unix Systems Administrator | ICQ #: 24767726 | | and student at The | AIM: muythaibxr | | The University of Maryland, | Website: (Under Construction) | | College Park. | http://www.wam.umd.edu/~culverk/| = On Wed, 29 Mar 2000, Geoff Rehmet wrote: Matthew Dillon writes : :I'm running a UP machine with Matt's latest changes. I was just :compiling a new kernel and noticed the my PS/2 mouse under X was very :sluggish. I never noticed this sort of behavior before the changes, :even while doing ``make -j8 buildworld''. : :The compile was running on a UW SCSI disk on an Adaptec aic7880 Ultra :SCSI adapter on my motherboard, in case it matters. : :Looks like something has been verschlimmbessert (a wonderful German :word which means "made worse through improvement" :) : : :Gary Jennejohn / [EMAIL PROTECTED] [EMAIL PROTECTED] [EMAIL PROTECTED] Ok, I'll take a look at it... the most likely cause is that I somehow broke need_resched. Not impossible, I'll check it out. I'm seeing the same symptoms while doing a make - my console performance is very jumpy and sluggish. Geoff. -- Geoff Rehmet, The Internet Solution [EMAIL PROTECTED]; [EMAIL PROTECTED]; [EMAIL PROTECTED] tel: +27-83-292-5800 To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: UP kernel performance and Matt Dillon's patches
: :Yeah, I was wrong before.. I just had to really hit the system hard before :I noticed this behavior... it get's pretty bad... the mouse get's jumpy, :and the keyboard input is really slow... : := :| Kenneth Culver | FreeBSD: The best OS around.| I definitely blew the need_resched stuff. I found it. I was clearing the 'astpending' variable in doreti, which used to be just fine, but now that we have two flags in it instead of one it was clearing the need-resched flag bit as well as the astpending flag bit. If you don't want to wait, here it is (it's really the last bit that fixes the problem. The first two bits are incidental). I've committed the fix. I would appreciate others testing it as well. It seems to solve the problem I was able to reproduce on my systems. Index: isa/ipl.s === RCS file: /home/ncvs/src/sys/i386/isa/ipl.s,v retrieving revision 1.33 diff -u -r1.33 ipl.s --- isa/ipl.s 2000/03/28 07:16:23 1.33 +++ isa/ipl.s 2000/03/29 06:00:29 @@ -123,10 +123,10 @@ andl_ipending,%ecx /* set bit = unmasked pending INT */ jne doreti_unpend movl%eax,_cpl - MPLOCKED decb _intr_nesting_level + decb_intr_nesting_level /* Check for ASTs that can be handled now. */ - cmpb$0,_astpending + cmpl$0,_astpending je doreti_exit testb $SEL_RPL_MASK,TF_CS(%esp) jne doreti_ast @@ -271,7 +271,7 @@ ALIGN_TEXT doreti_ast: - movl$0,_astpending + andl$~AST_PENDING,_astpending sti movl$T_ASTFLT,TF_TRAPNO(%esp) call_trap To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
