neutro Posted February 1, 2004 Report Share Posted February 1, 2004 My Mandrake 9.1 install boots quite fast. That is, once init is running. When I boot up, I'd say there's a ~20 sec. delay between the moment where I press "enter" in LILO and the moment where I see init starting. During this delay, the disk access LED is flashing. I have a reasonably fast computer (XP 1800+, ATA-100 disk, 512 MB DDR), but I am running Mandrake 9.1's stock kernel. I was simply wondering if this delay between LILO and init loading (I guess this is where the kernel image is loaded and bootstrapped) is normal. Quote Link to comment Share on other sites More sharing options...
bvc Posted February 1, 2004 Report Share Posted February 1, 2004 9.1 was quirky with startup. It really depends on too many things to say. It's not really kernel related though, but more likely initscripts. Even when I would get updates and the boot would be fast, the next set of updates would make it slow again. 9.2-->10 however has consistantly been fast. Quote Link to comment Share on other sites More sharing options...
Michel Posted February 1, 2004 Report Share Posted February 1, 2004 I wouldn't say it is normal ..I have a,d amd thunderbird 900 Mhz and 128 MB ram and you can say it is quite instantaneous with me .. ~3 seconds or so. Can be some mroe or less Quote Link to comment Share on other sites More sharing options...
neutro Posted February 2, 2004 Author Report Share Posted February 2, 2004 (edited) I wouldn't say it is normal That's what I thought ;) It's not really kernel related though, but more likely initscripts. Probably not: the delay is *before* init starts. My understanding is that after the kernel being loaded in memory, init (pid = 1) is the first thing to run, and is resbonsible to start all initscripts. Beside, most of these scripts I believe produce a visual output (which I can see, later). I would get updates and the boot would be fast, the next set of updates would make it slow again. Interesting... Edited February 2, 2004 by neutro Quote Link to comment Share on other sites More sharing options...
vincentv Posted February 4, 2004 Report Share Posted February 4, 2004 20 seconds! Is that all? Mine takes far longer than that and I have a 2.5GHz Athlon! However, the reason it does this is simple. In between starting lilo and Init appearing the system (among other things) checks the hard drives and devices reported by the machine's BIOS to find out what they are etc. (look in the /var/log/messages - you need to be root) My mother board, an ASUS A7N8X - Delux reports 6 harddrives: hda - my first hard drive hdb - my second hard drive hdc - my cd burner hdd - my dvd player these are not a problem since they exist. It also reports two other drives hde, hdg which are part of the serial ATA RAID system (if you choose to use it). Now, there are no drives on this RAID system, but because the BIOS has reported their existance the operating system keeps tying to communicate with each of them and it tries for ages until it eventually gives up and moves on to continue with the boot process Here's an extract from my messages. I have marked the lines >>>> ------------------------------------ Feb 2 15:35:07 vince kernel: NFORCE2: IDE controller at PCI slot 00:09.0 Feb 2 15:35:07 vince kernel: NFORCE2: chipset revision 162 Feb 2 15:35:07 vince kernel: NFORCE2: not 100%% native mode: will probe irqs later Feb 2 15:35:07 vince kernel: AMD_IDE: Bios didn't set cable bits corectly. Enabling workaround. Feb 2 15:35:07 vince kernel: AMD_IDE: nVidia Corporation nForce2 IDE UDMA133 (rev a2) UDMA133 controller on pci00:09.0 Feb 2 15:35:07 vince kernel: ide0: BM-DMA at 0xf000-0xf007, BIOS settings: hda:DMA, hdb:DMA Feb 2 15:35:07 vince kernel: ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: hdc:DMA, hdd:DMA Feb 2 15:35:07 vince kernel: SiI3112 Serial ATA: IDE controller at PCI slot 01:0b.0 Feb 2 15:35:07 vince kernel: SiI3112 Serial ATA: chipset revision 2 Feb 2 15:35:07 vince kernel: SiI3112 Serial ATA: not 100%% native mode: will probe irqs later Feb 2 15:35:07 vince kernel: ide2: MMIO-DMA , BIOS settings: hde:pio, hdf:pio Feb 2 15:35:07 vince kernel: ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio Feb 2 15:35:07 vince kernel: hda: Maxtor 6Y080P0, ATA DISK drive Feb 2 15:35:07 vince kernel: hdb: Maxtor 6Y080P0, ATA DISK drive Feb 2 15:35:07 vince kernel: blk: queue c0181bc0, I/O limit 4095Mb (mask 0xffffffff) Feb 2 15:35:07 vince kernel: blk: queue c0181cfc, I/O limit 4095Mb (mask 0xffffffff) Feb 2 15:35:07 vince kernel: hdc: ARTEC WRR-52Z 1.15 20030107, ATAPI CD/DVD-ROM drive Feb 2 15:35:07 vince kernel: hdd: AOPEN DVD1648/LKY, ATAPI CD/DVD-ROM drive >>>>Feb 2 15:35:07 vince kernel: hde: no response (status = 0xfe) >>>>Feb 2 15:35:07 vince kernel: hdg: no response (status = 0xfe) Feb 2 15:35:07 vince kernel: ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 Feb 2 15:35:07 vince kernel: ide1 at 0x170-0x177,0x376 on irq 15 Feb 2 15:35:07 vince kernel: hda: attached ide-disk driver. Feb 2 15:35:07 vince kernel: hda: host protected area => 1 ------------------------------------------ Unfortunately, I cannot disable these drives in the BIOS (even though it says that they are disabled) so the boot process always checks them out. Other devices that might cause boot delays are unused on-board lan, you might want to disable them in the BIOS if you don't use them. Quote Link to comment Share on other sites More sharing options...
Qchem Posted February 4, 2004 Report Share Posted February 4, 2004 Is it not possible to switch off the RAID (done by a jumper on my mobo) and just use the channel as standard SATA? If you have lots of time to spare you could attempt to recompile the kernel without RAID support - could stop it, could also be a waste of time. :D Quote Link to comment Share on other sites More sharing options...
neutro Posted February 8, 2004 Author Report Share Posted February 8, 2004 vincentv, you probably have a good point. I use an Asus A7V266 mobo with two IDE controllers (8 possible IDE devices). My configuration is quite simple now (it's been much more crowded in the past!): First controller (Promise controller, not configurable through BIOS): ------------------------- hda - primary HD hdb - secondary HD hdc - unused hdd - unused Second controller (configurable through BIOS): -------------------------- hde - dvd player hdf - cd burner hdg - unused (set to 'None' instead of 'Auto' in BIOS) hdh - unused (set to 'None' instead of 'Auto' in BIOS) I set hdg and hdg (that is, secondary master and slave) to 'None' instead of 'Auto' in BIOS in an attempt to reduce boot time, without any effect. Yet I managed to find how to increase verobsity during boot-up (removing the "quiet" argument in the appropriate lilo.conf entry) and I can see that the choking point is here: hdg: IRQ probe failed (0xfffffff8)hdg: IRQ probe failed (0xfffffff8) hdg: no response (status = 0x0e), resetting drive hdg: IRQ probe failed (0xfffffff8) hdg: no response (status = 0x0e) hdh: IRQ probe failed (0xfffffff8) hdh: IRQ probe failed (0xfffffff8) hdh: no response (status = 0x0e), resetting drive hdh: IRQ probe failed (0xfffffff8) hdh: no response (status = 0x0e) This happens regardless of the BIOS settings for hdg or hdh (Auto or None). Note also that there is nothing about IRQ probing of hdc and hdd in /var/log/dmesg. Well if anybody know how to speed this up (bypassing IRQ probing on hdg and hdh?), tell me... Quote Link to comment Share on other sites More sharing options...
neutro Posted February 16, 2004 Author Report Share Posted February 16, 2004 Sorry to ressuscitate that thread B) However, for the record, I got a workaround. In lilo.conf, I pass the options hdg=none hdh=none in the append line. IRQ probing is not attempted anymore and now Mandrake boots *really* fast. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.