Delay in booting 2.6.14 on SATA laptop

smurf · 2005-12-10 14:47:52

This is my first post to the board so I like to say hello to the community.

I face a problem when booting my laptop with the kernel 2.6.14. During bootup the process gets to a rest for about 90 seconds. After that time, the boot process finishes successfully.

The laptop is a FujitsuSiemens Lifbebook s7020 with the follwing hardware:
Centrino "Sonoma" P760 (2 Ghz)
Intel i915 PCI Express
SXGA+ at 14" (1400x1050)
DD 80Go SATA
Ethernet Broadcom 1 Gb/s tigon 3
Intel Pro Wireless 2200 BG
Bluetooth / FireWire ieee 1394

messages.log (part of)
Dec 6 16:21:22 (none) sym53c416.c: Version 1.0.0-ac
Dec 6 16:21:22 (none) scsi: <fdomain> Detection failed (no card)
Dec 6 16:21:22 (none) GDT-HA: Storage RAID Controller Driver. Version: 3.04
Dec 6 16:21:22 (none) GDT-HA: Found 0 PCI Storage RAID Controllers
Dec 6 16:21:22 (none) 3ware Storage Controller device driver for Linux v1.26.02.001.
Dec 6 16:21:22 (none) imm: Version 2.05 (for Linux 2.4.0)
Dec 6 16:21:22 (none) ipr: IBM Power RAID SCSI Device Driver version: 2.0.14 (May 2, 2005)
Dec 6 16:21:22 (none) Adaptec aacraid driver (1.1-4 Nov 11 2005 19:47:11)
Dec 6 16:21:22 (none) nsp32: loading...
Dec 6 16:21:22 (none) PCI: Found IRQ 11 for device 0000:00:1f.2
Dec 6 16:21:22 (none) PCI: Sharing IRQ 11 with 0000:00:1d.1
Dec 6 16:21:22 (none) PCI: Sharing IRQ 11 with 0000:00:1f.3
Dec 6 16:21:22 (none) input: PS/2 Generic Mouse on synaptics-pt/serio0

booting stops here for about 90 seconds

Dec 6 16:21:22 (none) ahci(0000:00:1f.2) AHCI 0001.0000 32 slots 4 ports 1.5 Gbps 0x5 impl SATA mode
Dec 6 16:21:22 (none) ahci(0000:00:1f.2) flags: 64bit ncq pm led slum part
Dec 6 16:21:22 (none) ata1: SATA max UDMA/133 cmd 0xF8B04500 ctl 0x0 bmdma 0x0 irq 11
Dec 6 16:21:22 (none) ata2: SATA max UDMA/133 cmd 0xF8B04580 ctl 0x0 bmdma 0x0 irq 11
Dec 6 16:21:22 (none) ata3: SATA max UDMA/133 cmd 0xF8B04600 ctl 0x0 bmdma 0x0 irq 11
Dec 6 16:21:22 (none) ata4: SATA max UDMA/133 cmd 0xF8B04680 ctl 0x0 bmdma 0x0 irq 11
Dec 6 16:21:22 (none) ata1: dev 0 ATA, max UDMA/100, 156301488 sectors: lba48
Dec 6 16:21:22 (none) ata1: dev 0 configured for UDMA/100
Dec 6 16:21:22 (none) scsi0 : ahci

Although the stop occurs after the touchpad driver message, I do not know, if it is really related to that driver. Probably more related to the messages following the stop (ahci, ata, sata?). Is there a chance to overcome the delay?

Please tell me, if you need more informations than provided in that message.
Many thanks for your help.

smurf · 2005-12-14 07:12:22

Anybody any idea...?

tpowa · 2005-12-14 08:25:27

if you run kernel from testing with initrd modify your /etc/mkinitrd.conf to not use scsi and rebuild it, then the scsi modules are not included anymore and boot should be faster.
greetings
tpowa

smurf · 2005-12-14 12:23:00

If I switch off SCSI, the kernel will not boot at all. I learned, that SATA drives are seen as SCSI drives (and therefore, the SCSI should not switched off) , but I might be wrong with that...(?)

phrakture · 2005-12-14 16:26:17

smurf wrote:

If I switch off SCSI, the kernel will not boot at all. I learned, that SATA drives are seen as SCSI drives (and therefore, the SCSI should not switched off) , but I might be wrong with that...(?)

Hmm, then keep scsi on - try to configure /etc/mkinitrd.conf for your specific hardware. You can use hwdetect to dump the modules for your hardware, and then add them to the config - details should be in the wiki.

With a custom initrd, see if the boot speed changes. If it does, that we can track down what causes the hang in the full image.

smurf · 2005-12-14 16:45:22

hwdetect says:

AGP : agpgart intel-agp
IDE : ide-cd ide-core generic piix
SCSI : ahci sd_mod
SATA : ata_piix libata
USB : usb-storage usblp usbhid usbcore ehci-hcd uhci-hcd
FW : eth1394 ieee1394 ohci1394 sbp2
NET : ppp_generic slhc tg3 ieee80211 ieee80211_crypt ipw2200
PCMCIA : pcmcia_core rsrc_nonstatic yenta_socket
SOUND : snd-mixer-oss snd-pcm-oss snd-page-alloc snd-pcm snd-timer snd snd-hda-codec snd-hda-intel soundcore
OTHER : cdrom hw_random lp ppdev rtc tpm tpm_atmel tpm_infineon tpm_nsc i2c-i801 i2c-core evdev pcspkr serio_raw parport parport_pc pci_hotplug shpchp

so changed mkinitrd.conf to:

# Disable whole subsystems by setting to "1"
REMOVE_IDE=
REMOVE_SCSI=
REMOVE_SATA=
REMOVE_CDROM=
REMOVE_USB=1
REMOVE_FW=1
REMOVE_RAID=1
REMOVE_DM=1
REMOVE_FS=
# Enable auto-detection of HOSTCONTROLlER and FILESYSTEMS (set to "1")
# (NOTE: This is still experimental, so it is disabled by default)
AUTODETECT=1
# Define which modules are needed by adding "moduleX moduleY"
# If left empty, all modules are included if they are not disabled above
HOSTCONTROLLER_IDE="ide-cd ide-core generic piix"
HOSTCONTROLLER_SCSI="ahci sd_mod"
HOSTCONTROLLER_SATA="ata_piix libata"
HOSTCONTROLLER_USB="usb-storage usblp usbhid usbcore ehci-hcd uhci-hcd"
FILESYSTEMS="ext3 reiserfs"

but system still waits at the point shown above for about 90 seconds like as it waits for a timeout to happen in ahci...?

paranoos · 2005-12-14 17:41:02

turn off the autodetection and see if that's the culprit

smurf · 2005-12-15 09:07:27

Switched off autodetection and made mkinitrd auto but no success.

In kernel.log I found this message:

Dec 15 09:50:40 (none) ACPI: PCI Interrupt 0000:00:1f.2[B] -> GSI 19 (level, low) -> IRQ 19
Dec 15 09:50:40 (none) input: PS/2 Generic Mouse on synaptics-pt/serio0
>> Dec 15 09:50:40 (none) PCI: Setting latency timer of device 0000:00:1f.2 to 64
Dec 15 09:50:40 (none) ahci(0000:00:1f.2) AHCI 0001.0000 32 slots 4 ports 1.5 Gbps 0x5 impl SATA mode
Dec 15 09:50:40 (none) ahci(0000:00:1f.2) flags: 64bit ncq pm led slum part
Dec 15 09:50:40 (none) ata1: SATA max UDMA/133 cmd 0xF8822500 ctl 0x0 bmdma 0x0 irq 19
Dec 15 09:50:40 (none) ata2: SATA max UDMA/133 cmd 0xF8822580 ctl 0x0 bmdma 0x0 irq 19
Dec 15 09:50:40 (none) ata3: SATA max UDMA/133 cmd 0xF8822600 ctl 0x0 bmdma 0x0 irq 19
Dec 15 09:50:40 (none) ata4: SATA max UDMA/133 cmd 0xF8822680 ctl 0x0 bmdma 0x0 irq 19
Dec 15 09:50:40 (none) ata1: dev 0 cfg 49:2f00 82:346b 83:7f09 84:6063 85:3469 86:3f09 87:6063 88:203f
Dec 15 09:50:40 (none) ata1: dev 0 ATA, max UDMA/100, 156301488 sectors: lba48
Dec 15 09:50:40 (none) ata1: dev 0 configured for UDMA/100

This sequence looks similar than the one in my first post, which was copied out of messages.log. In addition there is the message 'PCI: Setting latency timer of device 0000:00:1f.2 to 64' ( >> inserted to emphasize).
I then more careful timed the delay with a stopwatch and found out, that the delay is around these 64 seconds mentioned in the message.
The message itself appears on other places before the sequenze shown above:

Dec 15 09:50:40 (none) PCI: Setting latency timer of device 0000:00:1c.0 to 64
Dec 15 09:50:40 (none) PCI: Device 0000:00:1c.1 not available because of resource collisions
Dec 15 09:50:40 (none) PCI: Setting latency timer of device 0000:00:1c.1 to 64
Dec 15 09:50:40 (none) PCI: Setting latency timer of device 0000:00:1e.0 to 64

the process however stops only once for these 64 seconds. Can this latency timer be the cause of the delay? If yes, can it be influenced (set lower) somewhere?

phrakture · 2005-12-15 16:08:15

smurf wrote:

HOSTCONTROLLER_IDE="ide-cd ide-core generic piix"
HOSTCONTROLLER_SCSI="ahci sd_mod"
HOSTCONTROLLER_SATA="ata_piix libata"
HOSTCONTROLLER_USB="usb-storage usblp usbhid usbcore ehci-hcd uhci-hcd"
FILESYSTEMS="ext3 reiserfs"

I need to point out something here, and this may fix your problem. In the initrd you only need the modules to mount your root device. That would be whatever you have in the "root=/dev/blah" line in menu.lst/lilo.conf - you need nothing else.

Remove everything else, and try again - I have a feeling the 90s delay may be from usb stuff, which you don't need.

smurf · 2005-12-16 10:06:26

Thank you, phrakture for the advice. As soon as I had set REMOVE_IDE= 1, the problem was gone. I then also tried to switch off SCSI and this time I was able to boot. Maybe something in the kernel has changed. I am one of the early adopter of ARCH 0.7.1 (right when the version was put on the net, I owned a new laptop so I thought, I would take the challenge and set it up with 0.7.1) I am sure, that the laptop then failed to boot when either IDE or SCSI was set to 1 so I was afraid to switch it off again...

My mkinitrd.conf now looks like:

REMOVE_IDE=1
REMOVE_SCSI=1
REMOVE_SATA=
REMOVE_CDROM=1
REMOVE_USB=1
REMOVE_FW=1
REMOVE_RAID=1
REMOVE_DM=1
REMOVE_FS=

AUTODETECT=1

HOSTCONTROLLER_IDE=
HOSTCONTROLLER_SCSI=
HOSTCONTROLLER_SATA="ata_piix libata"
HOSTCONTROLLER_USB=
FILESYSTEMS="ext3 reiserfs"

Many thanks for all the help and have a nice day
smurf

lanrat · 2005-12-16 11:25:55

I'm pretty sure you don't need libata in HOSTCONTROLLER_SATA. It'll be loaded automatically. At least that's how it works on my i915 based notebook. Also hwdetect has special options for scanning modules to be used in mkinitrd. Check your lsmod output for more modules to be eliminated.

phrakture · 2005-12-16 16:16:30

Hmm, it looks like it's a problem with the ide modules themselves... like the *almost* detect your drive, then decide to give up.

PS You don't need 2 file systems in the initrd unless you have two possible "/" root partitions - you only need the filesystem for the partition listed on the "root=/dev/*" kernel parameter.

edit PPS Can you please bug report this - it might be something we can take care of somehow (I promise nothing, heh). Please provide all information for your system - that is 'lspci' info, 'hwdetect --show-modules' and anything else related to the hardware you can think of. Thanks.

kozaki · 2005-12-17 00:27:54

As smurf, i tried to remove every module in /etc/mkinitrd.conf but SATA module for my boxes.

But as soon as I add "1" in 'REMOVE_CDROM=' I find neither cd* neither dvd in /dev
Therefore I can't mount or use the CD/DVD optical drives.

That's on a 100% IDE Dell PIII as well on my asrock Dual sata2 boxes with IDE NEC drive.
How comes i have to add CDrom module in /etc/mkinitrd.conf so the system can see the optical drives ?

paranoos · 2005-12-17 09:52:31

kozaki wrote:

How comes i have to add CDrom module in /etc/mkinitrd.conf so the system can see the optical drives ?

you don't have to put them in initrd, you can simply load ide-cd or whatever modules you need in rc.conf. the reason you didn't have to do this before is that the cd modules aren't built into the kernel anymore, they need to be loaded.

remember, initrd only needs to include the modules required to mount root. everything else belongs in rc.conf

kozaki · 2005-12-17 18:24:37

paranoos wrote:

remember, initrd only needs to include the modules required to mount root. everything else belongs in rc.conf

That's the clearest string (your 2 phrases put together) I've read on this subject

I'm presently in the middle of downloading a very big torrent file I'd like not to interrupt (File hashing takes forever on x GB) but I edited both files so that they look like this :

# /etc/rc.conf - Main Configuration for Arch Linux
#
# hwdetect :
MOD_AUTOLOAD="yes"
MOD_BLACKLIST=()
# initrd only needs to include the modules required to mount root.
# everything else belongs in rc.conf :
MODULES=(!usbserial !ide-scsi sata_uli ide_cd uli526x)

&

# /etc/mkinitrd.conf - load those modules that are necessary to boot the "/" partition
#
REMOVE_IDE=1
REMOVE_SCSI=1
REMOVE_SATA=
REMOVE_CDROM=1
REMOVE_USB=1
REMOVE_FW=1
REMOVE_RAID=1
REMOVE_DM=1
REMOVE_FS=
#
AUTODETECT=0
#
# Define which modules are needed by adding "moduleX moduleY"
# If left empty, all modules are included if they are not disabled above
HOSTCONTROLLER_IDE=
HOSTCONTROLLER_SCSI=
HOSTCONTROLLER_SATA=sata_uli
HOSTCONTROLLER_USB=
FILESYSTEMS=ext3

Hope it will do the job nicelly

lanrat · 2005-12-18 19:38:07

There are some mainstream problems using dvd/cdrw with sata/ide drivers at the same time.
A good explanation is available at these webpages:
1. thinkwiki
2. redhat bugzilla
3. kerneltrap

Basically my notebook dvd/cdrw should work with sata drivers. And it's being recognized by the kernel. But the standard drivers do not have PATA ATAPI support enabled by default and there is no device link under /dev so you can't mount it. And if the ide driver is loaded after sata driver it will not be able to take ownership of dvd device producing a long delay on kernel boot (kernel will boot fine though).

There are a few workarounds described on the above webpages. You can try to force ide_generic driver for dvd/cdrw using HOSTCONTROLLER_IDE="ide_generic piix" and HOSTCONTROLLER_SATA="ata_piix" in /etc/mkinird.conf. This will load the ide_generic driver first but your /dev/sda will become /dev/hda and boot will fail. It should be fixable by adding some options to /etc/modprobe.conf and regenerating initrd (details in the links) but even if it works you'll not be able to use DMA for dvd/cdrw. I didn't go that far (no time) so this is only a theory. You can also enable the experimental PATA ATAPI option for sata driver and have a /dev link + DMA enabled but it's... well experimental :-)

Meanwhile there is a new kernel in the upgrade que. We'll see how it works without any modifications.

smurf · 2005-12-19 08:00:22

kozaki wrote: But as soon as I add "1" in 'REMOVE_CDROM=' I find neither cd* neither dvd in /dev
Therefore I can't mount or use the CD/DVD optical drives.

This in fact happened with me as well, when I had the delay in booting, but I did not check it right now. I do not need the CDROM very often and so I did not pay attention if the driver was loaded or not. I will check that in a few hours.

phrakture wrote: PS You don't need 2 file systems in the initrd unless you have two possible "/" root partitions - you only need the filesystem for the partition listed on the "root=/dev/*" kernel parameter.

You are right. I was not consequent enough with removing everything, which is not needed to mount the root device. I will do that.

edit PPS Can you please bug report this - it might be something we can take care of somehow

'Bug report' sounds to me rather malignant... but if it helps to further improve Arch, I will try to file my first bug report...

kozaki · 2005-12-19 21:22:12

Everything seems allright so far, with mkinitrd.conf & rc.conf I last quoted in this post

$ lsmod | grep ide
ide_cd 38788 0
cdrom 38304 1 ide_cd
ide_core 117584 4 usb_storage,generic,alim15x3,ide_cd

Thank you lanrat for the links

smurf · 2005-12-20 07:26:00

The CDROM disappeared after my changes in mkinitrd.conf. After enabeling IDE (removed the "1" from REMOVE_IDE=), the CD was here again and could be mounted. I did not check however, if DMA can be influenced. Looks like initrd loads the correct driver for the CD although it is not used for booting. To my embarassement, the delay in booting was gone, although it seemed, that the IDE part was the reason for the delay. What finally is different in mkinitrd.conf to my first post is:

REMOVE_IDE=
REMOVE_SCSI=1 (switched off)
REMOVE_SATA=
REMOVE_CDROM=1 (switched off)
REMOVE_USB=1
REMOVE_FW=1
REMOVE_RAID=1
REMOVE_DM=1
REMOVE_FS=

AUTODETECT=0 (set from 1 to 0)

HOSTCONTROLLER_IDE=
HOSTCONTROLLER_SCSI=
HOSTCONTROLLER_SATA="ata_piix" (added ata_piix)
HOSTCONTROLLER_USB=
FILESYSTEMS="reiserfs" (removed ext3 where /boot is)

I am happy with that setup for the FujitsuSiemens s7020. The CD is usable but maybe not tweakable (did not check yet) so lanrats' advice or kozakis' setup might be better for certain laptops or systems.

Arch Linux

#1 2005-12-10 14:47:52

Delay in booting 2.6.14 on SATA laptop

#2 2005-12-14 07:12:22

Re: Delay in booting 2.6.14 on SATA laptop

#3 2005-12-14 08:25:27

Re: Delay in booting 2.6.14 on SATA laptop

#4 2005-12-14 12:23:00

Re: Delay in booting 2.6.14 on SATA laptop

#5 2005-12-14 16:26:17

Re: Delay in booting 2.6.14 on SATA laptop

#6 2005-12-14 16:45:22

Re: Delay in booting 2.6.14 on SATA laptop

#7 2005-12-14 17:41:02

Re: Delay in booting 2.6.14 on SATA laptop

#8 2005-12-15 09:07:27

Re: Delay in booting 2.6.14 on SATA laptop

#9 2005-12-15 16:08:15

Re: Delay in booting 2.6.14 on SATA laptop

#10 2005-12-16 10:06:26

Re: Delay in booting 2.6.14 on SATA laptop

#11 2005-12-16 11:25:55

Re: Delay in booting 2.6.14 on SATA laptop

#12 2005-12-16 16:16:30

Re: Delay in booting 2.6.14 on SATA laptop

#13 2005-12-17 00:27:54

Re: Delay in booting 2.6.14 on SATA laptop

#14 2005-12-17 09:52:31

Re: Delay in booting 2.6.14 on SATA laptop

#15 2005-12-17 18:24:37

Re: Delay in booting 2.6.14 on SATA laptop

#16 2005-12-18 19:38:07

Re: Delay in booting 2.6.14 on SATA laptop

#17 2005-12-19 08:00:22

Re: Delay in booting 2.6.14 on SATA laptop

#18 2005-12-19 21:22:12

Re: Delay in booting 2.6.14 on SATA laptop

#19 2005-12-20 07:26:00

Re: Delay in booting 2.6.14 on SATA laptop

Board footer