Subject: 10x kernel slowdown in less than a week?
To: None <port-sparc@NetBSD.ORG>
From: David Gilbert <dgilbert@jaywon.pci.on.ca>
List: port-sparc
Date: 02/28/1996 20:19:07
I don't know what people have been doing, but now, using the
same config file as a week ago, I get a kernel that is nearly 10x
slower for disk operations.
My Sun4/260 usually takes about 5 minutes to fsck it's 5 disks
(5 partitions, 2 of the disks are one ccd). With Monday's Sup, it's
taking nearly 45 minutes. The whole thing 'in use' is slow, too --- X
and launching applications is noticably slower.
The new kernel also failed to run the news expire in 12 hours
(I had to kill it). It normally takes 2 or 3. It fell seriously
behind on unbatching (batches came in over the modem faster than they
were processed). The system normally runs 200-300 batches/hour. It
was doing less than 10.
I'm going to include my config file at the end of this post.
I noticed also that the kernel ballooned in size by about 100K in one
week. Is this the nfs v3 code, or something more evil?
Whatever is slowing the whole process down SOOOO much is going
to stall my device driver work (I can't run that slow a kernel).
--begin src/sys/arch/sparc/conf/REPEAT--
# $NetBSD: GENERIC,v 1.13 1995/09/11 19:20:12 pk Exp $
machine sparc
options "SUN4"
#options MMU_3L
#options DDB,DEBUG,DIAGNOSTIC
#options SCSIDEBUG
# obsolete timezone spec
options TIMEZONE=0, DST=0
# Standard system options
options SWAPPAGER, VNODEPAGER, DEVPAGER # paging
#options DEBUG, DIAGNOSTIC # extra kernel debugging
options KTRACE # system call tracing support
#options KGDB # support for kernel gdb
#options KGDBDEV=0xc01, KGDBRATE=38400 # device & baud rate
options RCONSOLE # fast rasterop console
options SYSVMSG,SYSVSEM,SYSVSHM
options "COMPAT_09", "COMPAT_10"
# Filesystem options
options FFS
options NFSSERVER # Sun NFS-compatible filesystem
options NFSCLIENT # Sun NFS-compatible filesystem
options KERNFS # kernel data-structure filesystem
options FIFO # POSIX fifo support (in all filesystems)
#options QUOTA # fast filesystem with user and group quotas
options MFS # memory-based filesystem
options LOFS # Loop-back filesystem
options FDESC # user file descriptor filesystem
options UMAPFS # uid/gid remapping filesystem
#options LFS # Log-based filesystem (still experimental)
#options PORTAL # portal filesystem (still experimental)
options PROCFS # /proc
#options CD9660 # ISO 9660 + Rock Ridge file system
options UNION # union file system
# Networking options
options INET
options TCP_COMPAT_42 # compatibility with 4.2BSD TCP/IP
#options GATEWAY # IP packet forwarding
#options ISO # OSI networking
#options TPIP
#options EON
options COMPAT_43
options LKM
# Options for SPARCstation hardware
options COMPAT_SUNOS # compatibility with SunOS binaries
options COMPAT_SVR4 # compatibility with SVR4 binaries
config netbsd root on sd1a swap on sd1b and sd2b and sd3b and sd5b dumps on sd3b
mainbus0 at root
cpu0 at mainbus0
sbus0 at mainbus0
obio0 at mainbus0
vmes0 at mainbus0
vmel0 at mainbus0
#audio0 at mainbus0
auxreg0 at mainbus0
clock0 at mainbus0
clock0 at obio0 addr 0xf2000000
oclock0 at obio0 addr 0xf3000000
oclock0 at obio0 addr 0x03000000 # 4/100
eeprom0 at obio0 addr 0xf2000000
memreg0 at mainbus0
memreg0 at obio0 addr 0xf4000000
memreg0 at obio0 addr 0x04000000 # 4/100
timer0 at mainbus0
timer0 at obio0 addr 0xef000000
zs0 at mainbus0
zs0 at obio0 addr 0xf1000000 level 12 flags 0x103
zs0 at obio0 addr 0x01000000 level 12 flags 0x103 # 4/100
zs1 at mainbus0
zs1 at obio0 addr 0xf0000000 level 12 flags 0x103
zs1 at obio0 addr 0x00000000 level 12 flags 0x103 # 4/100
zs2 at obio0 addr 0xe0000000 level 12 flags 0x103
# FSBE/S or sun4m
#espdma0 at sbus0 slot ? offset ?
#espdma* at sbus0 slot ? offset ?
#esp0 at espdma? slot ? offset ?
#esp* at espdma? slot ? offset ?
# old sun4c SCSI
#dma0 at sbus0 slot ? offset ?
#dma* at sbus0 slot ? offset ?
#dma0 at obio0 addr 0xfa001000 level 4
#esp0 at sbus0 slot ? offset ?
#esp* at sbus0 slot ? offset ?
#esp0 at obio0 addr 0xfa000000 level 4
# old sun4c
#le0 at sbus? slot ? offset ?
#le* at sbus? slot ? offset ?
#le0 at obio0 addr 0xf9000000 level 6
ie0 at obio0 addr 0xf6000000 level 6
ie0 at obio0 addr 0x06000000 level 6 # 4/100
#ie1 at vmes0 addr 0xffe88000 level 5 vect 0x75
#ie2 at vmes0 addr 0xff31ff02 level 5 vect 0x76
#ie3 at vmes0 addr 0xff35ff02 level 5 vect 0x77
#ie4 at vmes0 addr 0xff2dff02 level 5 vect 0x7c
#xdc0 at vmel0 addr 0xffffee80 level 3 vect 0x44
#xdc1 at vmel0 addr 0xffffee90 level 3 vect 0x45
#xdc2 at vmel0 addr 0xffffeea0 level 3 vect 0x46
#xdc3 at vmel0 addr 0xffffeeb0 level 3 vect 0x47
#xd* at xdc? drive ?
#si0 at vmes0 addr 0xff200000 level 3 vect 0x40 flags 0x03
si0 at vmes0 addr 0xff200000 level 3 vect 0x40 flags 0x03
#sw0 at obio0 addr 0x0a000000 level 3
#bwtwo0 at sbus? slot ? offset ?
#bwtwo* at sbus? slot ? offset ?
#bwtwo0 at obio0 addr 0xfd000000 level 4 # 4/260
#bwtwo0 at obio0 addr 0xfb300000 level 4 # 4/{300/400} ??
# Add experimental cgtwo driver to try it out.
#cgtwo0 at vmes0 addr 0xff400000 level 4
cgtwo0 at vmes0 addr 0xff400000 level 4 vect 0xa8
#cgthree0 at sbus? slot ? offset ?
#cgthree* at sbus? slot ? offset ?
#cgsix0 at sbus? slot ? offset ?
#cgsix* at sbus? slot ? offset ?
#scsibus* at esp?
scsibus0 at si0
scsibus* at si?
#scsibus* at sw?
# Add in some definate devices of our own.
#sd0 at scsibus0 target 0 lun 0
sd1 at scsibus0 target 0 lun 1
sd2 at scsibus0 target 1 lun 0
sd3 at scsibus0 target 2 lun 0
sd4 at scsibus0 target 3 lun 0
sd5 at scsibus0 target 5 lun 0
sd* at scsibus? target ? lun ?
st0 at scsibus0 target 4 lun 0
#st* at scsibus? target ? lun ?
#cd* at scsibus? target ? lun ?
#fdc0 at mainbus0
#fd* at fdc0
pseudo-device loop
pseudo-device pty 32
pseudo-device ccd 2
pseudo-device sl 2
pseudo-device kbd
pseudo-device ppp 2
pseudo-device tun 4
pseudo-device vnd 3
pseudo-device bpfilter 16
--end src/sys/arch/sparc/conf/REPEAT--
Dave.
--
----------------------------------------------------------------------------
|David Gilbert, PCI, Richmond Hill, Ontario. | Two things can only be |
|Mail: dgilbert@jaywon.pci.on.ca | equal if and only if they |
|http://www.pci.on.ca/~dgilbert | are precisely opposite. |
---------------------------------------------------------GLO----------------