Bugzilla – Bug 117541
Install get stuck during Initialization, lost interrupts
Last modified: 2006-09-12 13:08:28 UTC
hdc CRN-8241B ATAPI CD/DVD-ROM drive ide1 at 0x170-0x177, 0x376 on irq 15 hdc ATAPI 24x CD-ROM drive, 128 kB cache, DMA Uniform CD-ROM driver Revision 3.20 - - - hdc: task_in_intr status=0x51 {DriveReady, SeekComplete, Error} error=0xb4 {AbortedCommand, LastFailedSense)=0x0b} failed opcode was 0xa1 - - - after this it keeps spitting out lost interrupt...
10.0 beta 3 was possible to install on the Thinkpad, but install of first CD was very slow (bug #113283), can it be related? When rebooting into 10.0 beta 3, I can read the 10.0 RC1 CD without getting the errors nor lost interrupts. I will retry RC1 with safe settings.
Install with safe settings seems to be a working workaround.
It is not working perfectly even in "safe settings", I run into an IO-error when reading unzip-5.52-2.i586.rpm. But retrying seems to work - still slow.
It seems to be a hardware or CD media problem. Attaching file /var/log/messages would help us.
Created attachment 50350 [details] The requested messages Or is it the messages from a failing (non "safe settings") install attempt you want?
Now tried to install on my main system - same problem! Quite different hardware (AMD Athlon, SiS chipset) Will attach some logs. CD Media is verified, once during CD-burn, once with the tool.
Created attachment 50442 [details] Boor messages from AMD system /var/log/messages was empthy!
Created attachment 50443 [details] /proc/interrupts from AMD system
Created attachment 50444 [details] /proc/devices for AMD system
Created attachment 50445 [details] /proc/modules for AMD system
Created attachment 50446 [details] /proc/ide/sis from AMD system
Created attachment 50447 [details] lsmod output from AMD system
Created attachment 50448 [details] lspci from SuSE 9.2 booted system
Verified the MD5SUM of the RC1 CD1, it matches with http://ftp.opensuse.org/pub/opensuse/distribution/SL-10.0-OSS-RC1/iso/MD5SUMS # md5sum /dev/dvdrecorder e479f35810ead9238f0cca363ace87e6 /dev/dvdrecorder CD is correct but does not work on two completely different systems... Suggestions? (Did not see the checkbox below earlier)
Created attachment 50460 [details] boot.msg from 10.0 beta3 Since beta3 worked a lot better on both systems. And that should be very similar - I retried it on the AMD. Notice differences in APIC (IOAPIC v. PIC), ACPI, irq assignments, and last but not least IDE <7>Probing IDE interface ide0... -<4>hda: HDS722512VLAT80, ATA DISK drive -<4>hdb: ST3200822A, ATA DISK drive +<4>hda: V33OA63AHDS722512VLAT80, ATA DISK drive +<4>hdb: 3.01 ST3200822A, ATA DISK drive <4>ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 -<6>hda: max request size: 1024KiB -<6>hda: 241254720 sectors (123522 MB) w/7938KiB Cache, CHS=16383/255/63, UDMA(1 00) +<6>hda: max request size: 128KiB +<4>hda: cannot use LBA48 - full capacity 802213000 sectors (410733 MB) +<6>hda: 268435456 sectors (137438 MB) w/8614KiB Cache, CHS=47189/85/200 +<4>hda: set_multmode: status=0x51 { DriveReady SeekComplete Error } +<4>hda: set_multmode: error=0x04 { DriveStatusError } +<4>ide: failed opcode was: 0xef <6>hda: cache flushes supported -<6> hda: hda1 hda2 -<6>hdb: max request size: 1024KiB -<6>hdb: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63, UDMA(1 00) +<3>hda: INVALID GEOMETRY: 85 PHYSICAL HEADS? +<6> hda: unknown partition table +<6>hdb: max request size: 128KiB +<6>hdb: 0 sectors (0 MB) w/9496KiB Cache, CHS=0/165/55 +<4>hdb: set_multmode: status=0x51 { DriveReady SeekComplete Error } +<4>hdb: set_multmode: error=0x04 { DriveStatusError } +<4>ide: failed opcode was: 0xef <6>hdb: cache flushes supported -<6> hdb: hdb1 hdb2 hdb3 hdb4 +<3>hdb: INVALID GEOMETRY: 165 PHYSICAL HEADS?
Created attachment 50461 [details] acpidmp from beta3
Created attachment 50462 [details] lsmod for beta3
Created attachment 50463 [details] lspci from beta3
Created attachment 50464 [details] lsusb from beta3
Created attachment 50465 [details] /proc/devices from beta3
Created attachment 50466 [details] /proc/ide/sis from beta3
Created attachment 50467 [details] /proc/interrupts from beta3
Created attachment 50468 [details] /proc/modules from beta3
Created attachment 50469 [details] y2log from beta3
Summary of AMD system: motherboard: ASRock K7S8X R3 (not that uncommon) disks: hda: HDS722512VLAT80 hdb: ST3200822A hdc: _NEC DVD_RW ND-2500A video: GeForce FX 5200
This all looks like (the usual :-() broken BIOS problem. Roger, you did try "failsafe" booting, didn't you ?!
I have not tried failsafe on the AMD. Failsafe on the Thinkpad did work. But failure to install RC1 without failsafe on two completely different systems when Beta3 did work - I would count this as a blocker! I say it again: beta3 worked for both this systems! I installed it on the Thinkpad, did everything but the final command on the AMD.
Created attachment 50515 [details] boot.msg from 10.0 RC1 in safe settings This is a true blocker, it does not even start with safe settings!!! (I am home with my 1 year old son, he requires on-demand play RIGHT NOW...)
Have already sent the needed info, but have to check that checkbox...
Created attachment 50535 [details] boot.msg from 10.0 RC1 with insmod=ide-generic With ide-generic you come a bit further.
Created attachment 50536 [details] dmesg output from 10.0 RC1 with insmod=ide-generic But you will get lots of task_in_intr, extremely slow, not practically installable (I gave up)
Created attachment 50564 [details] boot.msg from 2.6.13.2 So I downloaded and recompiled 2.6.13.2 It boots nicely on the AMD system (but lacks modules to become fully operational)
Jens, any idea? This looks like a mix of several problems to me
Is there any newer RC that I could test? (only first CD is needed)
So I guess 10.0 will be of no use for me then... Nobody is working on this bug, it is NEW not ASSIGNED. No reaction to questions or other input...
Tried the new GM release on the Thinkpad - did not work! (not even with Failsafe) I have not tried on the AMD yet...
Tried on the AMD now - lost interrupts etc. during udev startup. Did not start properly. I have not tried Failsafe nor ide-generic on the AMD yet.
Created attachment 54503 [details] dmesg output from 10.0 GM
Created attachment 54504 [details] boot.msg from 10.0 GM with insmod=ide-generic
Created attachment 54506 [details] dmesg output from 10.0 GM with insmod=ide-generic
Created attachment 54508 [details] y2log from 10.0 GM with insmod=ide-generic
As you can see I have now tried to run with failsafe - did not work any better insmod=ide-generic - works "best" but see the dmesg log... FULL of stuff like this hda: task_in_intr: status=0x59 { DriveReady SeekComplete DataRequest Error } hda: task_in_intr: error=0x10 { SectorIdNotFound }, CHS=15790/10/129, sector=268435328 ide: failed opcode was: unknown What is happening here? It is only hda in the list... Could it be some incompability with that? BTW Knoppix 4.0.2 DVD from Linux Magazine works!
I'v got a very similar problem - install failed due to all these ide opcodes/errors/lost interrupts on a quite reliable desktop machine with 2 CDROMs and VIA chipset. It never had any such problems with a bunch of distoros I had tried on it over time from slackware 9 to some bleeding edge live cd's (with 2.6.13 kernel). Booting in safemode was even worse - it froze up completely, whereas the normal or noacpi case just left the ide light on, but with otherwise responsible system. Disconencting the other CD-drive helped... so I could install and than, later on reconnect that drive, update CD-integration and it all worked as it should. Is the install cd kernel different from the default? If so - there might be something gooten bad with ide or mounting.
Created attachment 55980 [details] boot.msg from 10.1 alpha 2 Tried 10.1 alpha 2 - same problem...
Created attachment 55981 [details] dmesg output from 10.1 alpha 2
BTW since this affects SiS systems with both Intel and AMD CPU my guess is that the problem is in the south bridge driver/hw: My south bridge is: SiS963L
Created attachment 56736 [details] boot.msg from 10.1 alpha 2 with irqpoll pci=usepirqmask Found a possible workaround in bug #128931 Did not work. But notice that a common theme when things do not work is that disk geometry is completely wrong!
Created attachment 56737 [details] dmesg output from 10.1 alpha 2 with irqpoll pci=usepirqmask
Severe system breakdown! Suddenly the AMD failed to boot, nothing on screen. And since I really needed the system this weekend I went out shopping... Bought a new motherboard (ASRock 939Dual-SATA2, ULi M1695+M1567 chipset) and a matching processor AMD64 3500+ and put them in a new case. I will try to find out the cause of the failure of my old system. But will also try installation on the new one (same disks, DRAM, DVD, etc.)
Created attachment 57337 [details] boot.msg from 10.1 alpha 2 on new ULi based system There are no problems with "10.1 alpha 2" on the ULi based system. * Same DISKs (other but identical IDE cable - 80pin) * Same DVD * Same Floppy * Same VGA board - Nvidia AGP * Same RAM (but right now only one module - 512 MB) * Same Power supply * New case, with case cooler * New Motherboard - ASRock 939Dual-SATA2 * New CPU * New CPU cooler Cooling is better (but also louder) on the new system
Created attachment 57338 [details] dmesg output from 10.1 alpha on ULi based system
Roger, I can't read any of your gzip'ed attachments. They come out garbled here, are you sure you attached the right ones?
Created attachment 57417 [details] one zip of boot.msg from different starts Same for me... strange... Trying to zip everything in one file (had it on a floppy)
I have verified that I can download and read the new attachment.
Reopen for 10.1 if the bug still exists.