Bugzilla – Bug 144652
System to slow
Last modified: 2006-01-26 11:52:46 UTC
After the last kernel update, the system loads to slow. I mean the proble is with the serial ata controller. The disk is Hitachi UDMA 133, but in boot.msg the system sets as UDMA 100. I do'nt know if the problem becomes by the irq sharing from the SATA controller with other controllers such as usb. I will attach hwinfo and boot.msj later. Thanks
Created attachment 64413 [details] boot.msg Section of the setting UDMA 100: <5>SCSI subsystem initialized <7>libata version 1.20 loaded. <7>sata_via 0000:00:0f.0: version 1.1 <4>ACPI: PCI Interrupt Link [LNKF] BIOS reported IRQ 0, using IRQ 9 <4>ACPI: PCI Interrupt Link [LNKF] enabled at IRQ 9 <4>PCI: setting IRQ 9 as level-triggered <6>ACPI: PCI Interrupt 0000:00:0f.0[B] -> Link [LNKF] -> GSI 9 (level, low) -> IRQ 9 <6>PCI: Via IRQ fixup for 0000:00:0f.0, from 0 to 9 <6>sata_via 0000:00:0f.0: routed to hard irq line 9 <6>ata1: SATA max UDMA/133 cmd 0xB800 ctl 0xB402 bmdma 0xA400 irq 9 <6>ata2: SATA max UDMA/133 cmd 0xB000 ctl 0xA802 bmdma 0xA408 irq 9 <6>input: ImPS/2 Generic Wheel Mouse as /class/input/input2 <7>ata1: dev 0 cfg 49:2f00 82:746b 83:7fe8 84:4023 85:7469 86:3e00 87:4023 88:203f <6>ata1: dev 0 ATA-6, max UDMA/100, 156250000 sectors: LBA48 <6>ata1: dev 0 configured for UDMA/100 <6>scsi0 : sata_via <6>ata2: no device found (phy stat 00000000) <6>scsi1 : sata_via <5> Vendor: ATA Model: HDS722580VLSA80 Rev: V32O <5> Type: Direct-Access ANSI SCSI revision: 05
Created attachment 64414 [details] hwinfo
The irq 9 is shared by acpi, sata, and usb. With the command lsdev, I got: lsdev Device DMA IRQ I/O Ports ------------------------------------------------ 0000:00:09.0 d800-d8ff 0000:00:0e.0 d400-d41f 0000:00:0e.1 d000-d007 0000:00:0f.0 a000-a0ff a400-a40f a800-a803 b000-b007 b400-b403 b800-b807 0000:00:0f.1 9800-980f 0000:00:10.0 9400-941f 0000:00:10.1 9000-901f 0000:00:10.2 8800-881f 0000:00:10.3 8400-841f acpi 9 cascade 4 2 dma 0080-008f dma1 0000-001f dma2 00c0-00df ehci_hcd:usb5 11 EMU10K1 10 d400-d41f emu10k1-gp d000-d007 fpu 00f0-00ff GPE0_BLK e420-e423 i8042 1 12 ide0 14 01f0-01f7 03f6-03f6 9800-9807 it87-isa 0290-0297 keyboard 0060-006f libata 9 motherboard e400-e47f e800-e81f PCI 0cf8-0cff pic1 0020-0021 pic2 00a0-00a1 PM1a_CNT_BLK e404-e405 PM1a_EVT_BLK e400-e403 PM_TMR e408-e40b pnp 0290-0297 0370-0375 e800-e81f rtc 8 0070-0077 sata_via a000-a0ff a400-a40f a800-a803 b000-b007 b400-b403 b800-b807 serial 03f8-03ff SysKonnect d800-d8ff SysKonnect SK-98xx 3 timer 0 timer0 0040-0043 timer1 0050-0053 uhci_hcd 8400-841f 8800-881f 9000-901f 9400-941f uhci_hcd:usb1 9 uhci_hcd:usb2 9 uhci_hcd:usb3 9 uhci_hcd:usb4 9 vesafb 03c0-03df vt596_smbus e800-e807
Why is this bug critical? The test system around here are not slower than before.
(In reply to comment #4) > Why is this bug critical? The test system around here are not slower > than before. > Sorry. The problem is with time for loading and unloading KDE and all of the graphical applications, and for example konsole takes 20 secs to load. Yesterday I probe with the kernel of tha day (mantel), and the problem continue. I probe to disable preload from init, and the problem continue. The original install was alpha2, but I have upgraded via apt to alfa 3 and 4, and the last update to Beta1, via smart. The other strange factor I see, is in the memory info (with kde loaded), appears more than 260 mb used as "cache disk" of the total of 512 mb, when I have a swap of 1 gb. Tell em if You need more info.
It seems to be the problem with preload. I try to reinstall preload, and got: smart reinstall preload Loading cache... Updating cache... ###################################################################################### [100%] Computing transaction... Installed packages (1): preload-0.2-7@i586 128.9kB of package files are needed. 1.4MB will be used. Confirm changes? (Y/n): Y Fetching packages... -> http://ftp.opensuse.org/pub/opensuse/distribution/SL-OSS-factory/inst-source/suse/i586/preload-0.2-7.i586.rpm preload-0.2-7.i586.rpm ###################################################################################### [100%] Committing transaction... Preparing... ###################################################################################### [ 0%] 1:Installing preload ###################################################################################### [100%] Output from preload-0.2-7@i586: insserv: Service boot.coldplug has to be enabled for service boot.preload insserv: exiting now! error: %post(preload-0.2-7.i586) scriptlet failed, exit status 1 I do'nt find boot.coldplug, in /etc/init.d
I guess we have different issues here. The preload issue is fixed now.
The configured speed of the drive itself isn't an issue, it's done because of problems there. In practice this means exactly nothing to the drive speed (you can compare both kernels with hdparm -t /dev/sda and see what you get, you probably want to do a few to make sure it's stable).
(In reply to comment #9) > The configured speed of the drive itself isn't an issue, it's done because of > problems there. In practice this means exactly nothing to the drive speed (you > can compare both kernels with hdparm -t /dev/sda and see what you get, you > probably want to do a few to make sure it's stable). > What info do You need? Only the result of hdparm -t /dev/sda? I will resolve this problem, because my system (Athlon XP 2600 - 512 mb ram), is slower than my old AMD k6-2-500. The first install of alfa 2 the system was more fastest than 9.3, but it has problems with hardware detection. When resolved the hardware detection, the system turned slowly.
I'm not sure what you expect of me, I have about zero details to go from here. Are you sure it's even a slowdown causes by io, it could be a ton of things that cause the system to feel slower. Please describe exactly what you think is slower! So yes, a hdparm -Tt /dev/sda would be a good start.
If only the desktop is slow and you have updated the machine instead of reinstalling, this might be a duplicate of bug #143715. See https://bugzilla.novell.com/show_bug.cgi?id=143715#c15 for my workaround.
(In reply to comment #12) > If only the desktop is slow and you have updated the machine instead of > reinstalling, this might be a duplicate of bug #143715. See > https://bugzilla.novell.com/show_bug.cgi?id=143715#c15 for my workaround. > Yes, fontconfig is part of the problem, sure; when I made the update, and when runs fontconfig, it takes some minutes. Yesterday, I has no time to probe hdparm -Tt /dev/sda, but other problem I see, (I do'nt know why the name of the network card is eth1, when I have only one card). In the attached hwinfo and boot.msg: <4> [<c01010a3>] cpu_idle+0x38/0x4d <4> [<c030862f>] start_kernel+0x17c/0x17e <4>printk: 50 messages suppressed. <3>dsl0: hw csum failure. <4> [<c0220d12>] __skb_checksum_complete+0x3c/0x4d <4> [<c0250fd6>] tcp_v4_rcv+0x98/0x73a <4> [<c0239423>] ip_route_input+0x39/0x14d <4> [<c023b35e>] ip_local_deliver+0x126/0x1b2 <4> [<c023b8c0>] ip_rcv+0x387/0x3e3 <4> [<c02231b2>] netif_receive_skb+0x259/0x290 <4> [<c0223256>] process_backlog+0x6d/0xd2 <4> [<c0223325>] net_rx_action+0x6a/0x104 <4> [<c011b715>] __do_softirq+0x35/0x7f <4> [<c011b781>] do_softirq+0x22/0x26 <4> [<c010507f>] do_IRQ+0x4b/0x53 <4> [<c0103b4a>] common_interrupt+0x1a/0x20 <4> [<c0101043>] default_idle+0x2b/0x53 <4> [<c01010a3>] cpu_idle+0x38/0x4d <4> [<c030862f>] start_kernel+0x17c/0x17e <4>printk: 49 messages suppressed. <3>dsl0: hw csum failure. <4> [<c0220d12>] __skb_checksum_complete+0x3c/0x4d <4> [<c0250fd6>] tcp_v4_rcv+0x98/0x73a <4> [<c0239423>] ip_route_input+0x39/0x14d <4> [<c023b35e>] ip_local_deliver+0x126/0x1b2 <4> [<c023b8c0>] ip_rcv+0x387/0x3e3 <4> [<c02231b2>] netif_receive_skb+0x259/0x290 <4> [<c0223256>] process_backlog+0x6d/0xd2 <4> [<c0223325>] net_rx_action+0x6a/0x104 <4> [<c011b715>] __do_softirq+0x35/0x7f <4> [<c011b781>] do_softirq+0x22/0x26 <4> [<c010507f>] do_IRQ+0x4b/0x53 <4> [<c0103b4a>] common_interrupt+0x1a/0x20 <4> [<c0101043>] default_idle+0x2b/0x53 <4> [<c01010a3>] cpu_idle+0x38/0x4d <4> [<c030862f>] start_kernel+0x17c/0x17e <4>printk: 47 messages suppressed. section boot.msg: 14 <notice>checkproc: /opt/kde3/bin/kdm 3185 <notice>killproc: kill(3178,15) <notice>pidofproc: dhcpcd 3291 1 eth1 device: 3Com Corporation 3c940 10/100/1000Base-T [Marvell] (rev 12) eth1 configuration: eth-id-00:0b:6a:f0:57:fa eth1 DHCP client (dhcpcd) is running eth1 IP address: 192.168.1.2/24 done dsl0 done modem0 modem0 Startmode is 'manual' skippedSetting up service network . . . . . . . . . . . . . . . .done
Marking as dupe of the other font bug. The Marvell network bug can be worked around with: ethtool -K eth0 rx off (or eth1 of course, if that is your device). In the future, please open a bug for each problem you see, these big "here's everything that's wrong for me" bugs are a bad idea. *** This bug has been marked as a duplicate of 143715 ***
(In reply to comment #11) > I'm not sure what you expect of me, I have about zero details to go from here. > Are you sure it's even a slowdown causes by io, it could be a ton of things > that cause the system to feel slower. Please describe exactly what you think is > slower! > > So yes, a hdparm -Tt /dev/sda would be a good start. > The result of this command was: /dev/sda: Timing cached reads: 1352 MB in 2.01 seconds = 674.02 MB/sec Timing buffered disk reads: 140 MB in 3.03 seconds = 46.27 MB/sec It's good?
Yes that looks fine, certainly the raw drive io speed isn't showing a slowdown.
(In reply to comment #16) > Yes that looks fine, certainly the raw drive io speed isn't showing a slowdown. > Thanks. About fontconfig, I installed the version 2.3.93.20060120, with the command rpm -U, and deleted all the font caches, and the system becomes normally, but if I run SuSEconfig, and when runs the fontconfig, it various minutes. I need to delete the old scripts by hand? (I have'nt made rpm -e with the old fontconfig). Thanks again.