Bugzilla – Bug 346478
Long hangs during boot while initializing harddrives.
Last modified: 2008-05-08 16:00:20 UTC
My machine hangs here: [ 8.942017] Freeing unused kernel memory: 236k freed [ 9.010449] ACPI: CPU0 (power states: C1[C1] C2[C2] C3[C3]) [ 9.023144] ACPI: Processor [CPU] (supports 8 throttling states) [ 9.039996] ACPI: Thermal Zone [THM0] (64 C) [ 9.061726] SCSI subsystem initialized [ 9.081321] libata version 3.00 loaded. [ 9.088476] ata_piix 0000:00:1f.1: version 2.12 [ 9.088493] PCI: Enabling device 0000:00:1f.1 (0005 -> 0007) [ 9.101538] ACPI: PCI Interrupt Link [LNKC] enabled at IRQ 11 [ 9.114046] ACPI: PCI Interrupt 0000:00:1f.1[A] -> Link [LNKC] -> GSI 11 (level, low) -> IRQ 11 [ 9.126905] PCI: Setting latency timer of device 0000:00:1f.1 to 64 [ 9.126980] scsi0 : ata_piix [ 9.139724] scsi1 : ata_piix [ 9.152652] ata1: PATA max UDMA/100 cmd 0x1f0 ctl 0x3f6 bmdma 0x1860 irq 14 [ 9.165667] ata2: PATA max UDMA/100 cmd 0x170 ctl 0x376 bmdma 0x1868 irq 15 === HERE ==== [ 9.340969] ata1.00: ATA-6: ST960812A, 3.04, max UDMA/100 [ 9.353647] ata1.00: 117210240 sectors, multi 16: LBA48 [ 9.380821] ata1.00: configured for UDMA/100 [ 9.484361] Clocksource tsc unstable (delta = -4175703340 ns) [ 9.497008] Time: acpi_pm clocksource has been installed. [ 9.537450] scsi 0:0:0:0: Direct-Access ATA ST960812A 3.04 PQ: 0 ANSI: 5 [ 9.566471] BIOS EDD facility v0.16 2004-Jun-25, 6 devices found [ 9.679891] usbcore: registered new interface driver usbfs for ~10 seconds during boot (even though the timestamps are not increasing). Will attach lspci -nnvvv, the machine is a Thinkpad X32.
Created attachment 186156 [details] lspci -nnvvv
Does 'clock=pit' help?
To explain a bit. I think what's wrong here is the timer not the ata_piix driver. libata probing contains a bunch of msleep()'s and it seems the pulse stops during one of those sleeps. Later clock code sees that tsc timer doesn't work too well and switch to acpi_pm. Ah... the parameter should be "clocksource=acpi_pm" or "clocksource=hpet".
*** Bug 345442 has been marked as a duplicate of this bug. ***
>Does 'clock=pit' help? yes, confirmed! >why doesn`t timing/profiling information show the delay ? >i would have expected something like this ><6>[ 24.730725] ata2: PATA max UDMA/100 cmd 0x170 ctl 0x376 bmdma 0x1c28 irq 15 >( >10s wait here) ><6>[ 34.891676] ata1.00: ATA-6: ST92011A, 3.04, max UDMA/100 >maybe i`m misinterpreting the timing output? now i understand! :)
clocksource=acpi_pm helped for me, on the ThinkPad X32, too. Sorry for the latency in anwering.
Libata won't even work for me at all w/ OS 11 alpha 0. I have to select a different driver and still get long hangs at boot time. https://bugzilla.novell.com/show_bug.cgi?id=300685 Been having various problems since I got solid state media. Being an early adopter sucks. :-(
Andi, any idea how to get around the early timer stuff?
Can we get full boot logs please?
Created attachment 190379 [details] dmesg output
here we go
Note that i cannot reproduce that any longer with current STABLE kernel 2.6.24-rc7-git5-2-default
*** Bug 355811 has been marked as a duplicate of this bug. ***
*** Bug 357665 has been marked as a duplicate of this bug. ***
I still can reproduce it with 2.6.24 (final) from Factory. In my case it hangs for ~5 minutes!
I can no longer reproduce it with Alpha 2 final. Can anyone confirm this too so we can close this bug report?
2.6.24 still showed the long delay when I don't have clocksource=acpi_pm in the kernel parameters. I'm just upgrading to 2.6.24.1 (current factory kernel) and give it a reboot without the clocksource again... to see if this helps. Otherwise, I actually don't mind having the clocksource parameter there, but it should be added automatically during the setup in this case.
Created attachment 195752 [details] dmesg output (dimstar) My notebook hangs at line 291 for about 1 minute... with kernel 2.6.24.1-3-default if Iadd clocksource=acpi_pm to the kernel parameters, the hang is not happening.
when I last tried with kernel 2.6.25-rc2, the delay was no longer there even without the clocksource parameter...
Great, closing as fixed.
Sorry, but using 11.0 Beta2, I have still such a 10 second delay on boot. But neither clock=pit does not help nor clocksource=acpi_pm. I own a ThinkPad R61.
Jan, please open a new bug, I think this is a different one, or at the least, it is the same one as bug 371776
Bug 388295 submitted.