I tried installing from the minimal install-cd from the experimental autobuilds (mirror/gentoo/experimental/x86/autobuilds/20081224/install-x86-minimal-20081224.iso) and the boot process hung at "scanning for wd7000..." Reproducible: Always Steps to Reproduce: 1. Attempt to boot off a 20081224 minimal CD on certain hardware 2. 3. Actual Results: Hang (waited for > 30min before killing) at scanning for wd7000... Expected Results: A successful boot The media should be clean, as I burned two disks to confirm, and checked the md5 sum. The CDs boot successfully on a computer containing a Western Digital HD. Reproduced on both an Inspiron 600m and Inspiron 5100. Both have from the factory hard drives, the 600m a Hitachi, and the 5100 a Toshiba. Standard 2008.0 release CD works just fine, so it must be a rather recent regression.
If you boot with 'noload=wd7000', does it work? What about booting with 'noscsi'?
No, it doesn't boot with either noload=wd7000 or noscsi. Luckily, further inspection reveals that wd7000 isn't the first sign of trouble. I looked more closely at the text output, and the pata_qdi scan segfaults instead of loading properly. About half the time with the various flags above, it will get past pata_qdi with a segfault, and continue, hanging at ehci-hcd. The other half will freeze at the pata_qdi step. Booting with noload=pata_qdi results in a successful boot.
Confirmed on Dell Latitude D600
Confirmed on Dell Inspiron 8600
Has anyone tried a newer autobuild installcd?
20090422 (http://mirrors.kernel.org/gentoo/releases/x86/autobuilds/20090422/install-x86-minimal-20090422.iso to be specific) still fails on an Inspiron 600m with identical behavior (sometimes hanging at scanning for pata_qdi, others hanging at wd7000 with a segfault at pata_qdi). noload=pata_qdi results in a successful boot. I'll try again once there is an iso which has the 2.6.28 kernel in it. I won't hold my breath, as the upstream changelog suggests nothing has changed in drivers/ata/pata_qdi.c itself.
Kernel team, Any insights on this?
Not enough info here really.. I presume there is some oops message being hidden somewhere. Maybe passing the "debug" parameter to the kernel would help gain more verbose output.
another idea..boot with noload=pata_qdi and then run: dmesg -n 7 modprobe pata_qdi Will hopefully give a nice oops message which could be captured on camera.
(In reply to comment #9) > another idea..boot with noload=pata_qdi and then run: > > dmesg -n 7 > modprobe pata_qdi > > Will hopefully give a nice oops message which could be captured on camera. > Sadly, that does not yield an oops. Only a single error (transcribed by hand): FATAL: Error inserting pata_qdi (/lib/modules/2.6.27-gentoo-r10/kernel/drivers/ata/pata_qdi.ko): no such device. At this point, the machine becomes unresponsive to anything (Ctrl-C, Ctrl-Alt-F2 and the like). Booting with debug as the argument might be mildly more helpful; when it gets all the way to wd7000, the pata_qdi line reads: :: Checking for pata_qdi... :: Scanning for pata_qdi...Floating point exception The rest of the screen is filled with the expected (for me at least) checking, scanning, loaded sequence of outputs for each module.
I meant to state explicitly /lib/modules/2.6.27-gentoo-r10/kernel/drivers/ata/pata_qdi.ko does in fact exist. Also, upon further trials with debug as an argument, the output intermittently switches between segmentation fault and Floating point exception when attempting to load pata_qdi.
You definitely ran "dmesg -n 7" first? Any chance of hooking up a serial console or netconsole? I'm pretty convinced there should be *some* message...
Created attachment 190972 [details] screenshot (In reply to comment #12) > You definitely ran "dmesg -n 7" first? > Any chance of hooking up a serial console or netconsole? I'm pretty convinced > there should be *some* message... > I couldn't be more certain. Proof attached. As for a serial or net console, only with a fair bit of google work that I don't have the time for currently, but perhaps in a week.
Confirmed on an IBM Thinkpad 570
Maybe I should add that this is with "install-x86-minimal-20090520.iso" Additionally, the "noload=pata_qdi" boot option works for me.
Tried to boot a Dell Dimension 4100 with the minimal cd, and it hung when loading pata_qdi. Used install-x86-minimal-20090526.iso I did get output when I demsg -n 7 and modprobe pata_qdi. It said it was a divide by zero error. I've attached two jpegs with most of the screen. Let me know if the part I cut off to the right is required.
Created attachment 192775 [details] screen shot
Created attachment 192777 [details] screen shot bottom half
Confirmed on Dell PowerEdge server, install-x86-minimal-20090602.iso Confirmed workaround, noload=pata_qdi skips pata_qdi.
Confirmed on IBM ThinkPad A21m
`noload=pata_qdi skips pata_qdi` works
Comfirmed on IBM Thinkpad T20
Confirmed on Compaq Armada M700 on 14 Aug 2009 using minimal install CD 20090623. Using noload=pata_qdi works.
Confirmed on Compaq Deskpro ENS/C466/6A/64 today using minimal install CD 20090901. Using noload=pata_qdi works.
append="noload=pata_qdi" works fine!
still there on 2009-09-01
Confirmed on Thinkpad R51, using install-x86-minimal-20090623.iso.
Confirmed on Dell Latitude D600 using install-x86-minimal-20090915.iso.
The same here on my very old desktop PC (00:04.1 IDE interface: Intel Corporation 82371AB/EB/MB PIIX4 IDE (rev 01)) to which I've connected pretty new laptop IDE HDD (WD3200BEVE 320GB) in order to install Gentoo on it. Installation process is hanging on wd7000 message, the noload=pata_qdi helped. I've booted install-x86-minimal-20090915.iso
Similar problem with Compaq Evo N620c and 10.0 liveDVD - also fixed by skipping pata_qdi.
Confirmed on IBM Thinkpad T23
confirmed on Acer Travelmate 6004 Wlmi (6000 series) with Live CD 10.0 hanging at ehci, giving up because I need USB 2.0 ;-)
So, anyone that can hook a serial or a netconsole and see if he gets any extra messages? This bug has been here for quite some time...
Confirmed on an (ancient) PIII Compaq Presario 7598, using install-x86-minimal-20091013.iso.
Computer: PIII Compaq Presario 7598, using install-x86-minimal-20091013.iso. When I start it with "gentoo noload=pata_qdi" and then run: # dmesg -n 7 # modprobe pata_qdi I get the following result: <see attachment '7598.jpg'> After this is shown, the computer totally freezes - no key combination does anything and I must power the computer off. The underscore cursor does keep flashing though, but the computer responds to nothing (even after leaving it still for some time). Also, I ran the memtest86 and it failed before finishing (on check #8 I think it was). It was a similar result as well, the computer stopped responding to any input (ESC wouldn't reboot, etc). I know it froze because the timer also stopped updating and the percentage didn't change after a long time. Interestingly, the little red plus sign that flashes did keep flashing though...Does this pata_qdi get loaded for the memtest86? Perhaps this freeze is related? My 2 RAM sticks are identical (2x256MB) and I'm pretty certain they are good. The CD is good (md5sum was good and it started fine on my laptop). Right now I'm trying to install a fresh gentoo over the Xubuntu that is currently on the computer. The computer has been doing these random freezes for a long time (on Xubuntu). Sometimes it would run for days without a problem, other times it would run for a minute or two before freezing. The freezes were always really random (which suggests RAM to me, but now this pata_qdi is giving me the exact same problem). The result was the same; the mouse pointer would disappear and the computer wouldn't response to anything. Perhaps this is also related to pata_dqi? I can't seem to find much information on pata_qdi and I am not familiar with it myself.
Created attachment 207381 [details] Error Message for 'modprobe pata_qdi' on Compaq Presario 7598 File '7598.jpg' for my comment.
This morning I tried running it with "gentoo debug noload=pata_qdi" When I do this, I get the following output: # dmesg -n 7 # modprobe pata_qdi do_IRQ: 0.40 No irq handler for vector (irq -1) # _ The last line is to indicate the cursor...which keeps blinking, but again the computer has frozen (or at least doesn't accept any input).
(In reply to comment #35) > Does this pata_qdi get loaded for the memtest86? memtest86 is not linux-based. It's a stand-alone "kernel" and runs completely under its own power.
Dell D600 Same problem today with both minimal and 10.2 DVD Solved with noload pata_qdi
Confirming, 'noload=pata_qdi' allows successful boot of install-x86-minimal-20091103.iso on an ancient Dell Dimension XPS T500 (IDE interface: Intel Corporation 82371AB/EB/MB PIIX4 IDE), on which boot was previously hanging at wd7000.
It's not just Dell laptops! I have this problem w/ an Acer Aspire laptop. I suggest altering the title (called "summary" here) since - most probably - users having other laptops won't find this report...
Confirmed on an old PIII DELL OptiPlex GX110, using install-x86-minimal-20091113.iso. "noload=pata_qdi" works.
The last cd which works for me was "install-x86-minimal-2008.0.iso". I have been using it since then.
Confirmed on DELL OptiPlex GX150, GX240, GX260, and GX270, using install-x86-minimal-20091113.iso. It works on DELL PowerEdge T100.
Confirmed as problem on HP Omnibook 4150, noload=pata_qdi works.
(In reply to comment #2) > No, it doesn't boot with either noload=wd7000 or noscsi. Luckily, further > inspection reveals that wd7000 isn't the first sign of trouble. I looked more > closely at the text output, and the pata_qdi scan segfaults instead of loading > properly. About half the time with the various flags above, it will get past > pata_qdi with a segfault, and continue, hanging at ehci-hcd. The other half > will freeze at the pata_qdi step. Booting with noload=pata_qdi results in a > successful boot. > Exact Same problem with install-x86-minimal-20091103 on Compaq Evo n620c...
Comfirmed on Dell Inspiron 1100 With the 2010 liveDVD: I once during about a dozen bots got a freeze at Scanning at pata_qdi with a noload IT got past that After a reboot it no longer frezed at pata_qdi but instead on ata_piix after noload=ata_piix it freezes at scanning wd7000 After a few more reboots It's a hit and miss it can freeze on pata_qdi or ata_piix more frequently at ata_piix
Created attachment 212841 [details] 'Divide error' during modprobe pata_qdi I could now attach a serial console to my N620C and booted 10.0 with "gentoo console=ttyS0,38400 verbose debug" - neither screen nor serial console show any message related to pata_qdi or an Oops, it just stops somewhere inside the ide-secion as mentioned in the first posts. Also the system does not respond to SysRq so i cant dump anything on the system after the module tries to load. However i could get an 'divide error' with "gentoo verbose debug noload=pata_qdi" when i tried to load pata_qdi via modprobe. The message does not show up every time but maybe it can help - picture of the screen is attached. Also have a look at http://patchwork.ozlabs.org/patch/39342/ - after this post "QDI65x0 controllers are fully supported by pata_legacy driver". Also the header inside pata_qdi.c says its mostly proof of concept so maybe pata_qdi should just be removed.
Confirmed on an IBM Thinkpad T40. This bug has been confirmed by an awful lot of people and was opened nearly a year ago. Is this going to get fixed, or should we include "noload=pata_qdi" in the install instructions?
Confirmed on Dell PowerEdge 1600SC bios rev A12. Boots fine with "gentoo noload=pata_qdi"
Confirmed on IBM ThinkPad T42 and Acer Travelmate 4003LCi. Boot up resolved by "gentoo noload=pata_qdi". Using install-x86-minimal-20091113.iso.
Confirmed on a Toshiba Satellite L15-S104 using Gentoo minimal install 2009-11-03. noload=pata_qdi fixed it and allowed normal boot from the live cd and the installed system.
Confirmed on Micron (MPC) Transport T2200
Confirmed on IBM PC 300PL
Problem is confirmed on Panasonic CF-18 laptop with iso "install-x86-minimal-20100126.iso" Using "noload=pata_qdi" workaround corrects this.
*** Bug 306399 has been marked as a duplicate of this bug. ***
Confirmed on a Dell Dimension V350. Appending noload=pata_qdi works.
(In reply to comment #57) > Confirmed on a Dell Dimension V350. Appending noload=pata_qdi works. > I forgot to mention that I was using the Gentoo 10.1 LiveDVD.
Confirmed on acer TravelMate 290 with latest install-x86-minimal-20100216.is. Workaround confirmed, too.
Confirmed on an IBM Thinkpad X31 with 512 MB RAM, Intel Pentium M, when booting from Gentoo minimal 20091103 livecd. Booting with gentoo noload=pata_qdi results in a working livecd environment. However, my 40GB IDE harddrive is recognized as /dev/hda rather than /dev/sda. So after partitioning I am forced to mount /dev/hda3 /mnt/gentoo mount /dev/hda1 /mnt/gentoo/boot instead of sda1 and sda3 resp. I then keep following the Handbook "A. Installing Gentoo" with default options until Chapter "10.d Rebooting the system". Then, when booting from harddrive, boot stops with * Checking root filesystem ... /dev/sda3: UNEXPECTED INCONSISTENCY: RUN fsck MANUALLY. (i.e., without -a or -p options) * Filesystem couldn't be fixed :( Give root password for maintenance (or type Control-D to continue): It seems to me that the inconsistency results from my partitioning as hda while booting as sda. Don't know how to fix this. Any hints?
(In reply to comment #60) > Confirmed on an IBM Thinkpad X31 with 512 MB RAM, Intel Pentium M, when booting > from Gentoo minimal 20091103 livecd. Booting with gentoo noload=pata_qdi > results in a working livecd environment. However, my 40GB IDE harddrive is > recognized as /dev/hda rather than /dev/sda. So after partitioning I am forced > to > mount /dev/hda3 /mnt/gentoo > mount /dev/hda1 /mnt/gentoo/boot > instead of sda1 and sda3 resp. > I then keep following the Handbook "A. Installing Gentoo" with default options > until Chapter "10.d Rebooting the system". > Then, when booting from harddrive, boot stops with > > * Checking root filesystem ... > /dev/sda3: UNEXPECTED INCONSISTENCY: RUN fsck MANUALLY. > (i.e., without -a or -p options) > > * Filesystem couldn't be fixed :( > Give root password for maintenance > (or type Control-D to continue): > > It seems to me that the inconsistency results from my partitioning as hda while > booting as sda. Don't know how to fix this. Any hints? > You might want to post on the Gentoo forums about that. It sounds like a separate issue to me.
Problem exists on a HP Vectra VL400 D using the gentoo minimal install CD from January 26th, 2010, x86 version. The system hangs upon reaching "looking for wd7000.." and is unresponsive to keyboard input. Using noload=pata_qdi leads to a successful boot.
I too have this problem on a Dell Optiplex GX110 with autobuild 2010-02-16. Using noload=pata_qdi fixed the problem for me as well.
Is bug going to be fixed? Comment #48 From Florian Knodt seems very reasonable to me.
(In reply to comment #60) > Confirmed on an IBM Thinkpad X31 with 512 MB RAM, Intel Pentium M, when booting > from Gentoo minimal 20091103 livecd. Booting with gentoo noload=pata_qdi > results in a working livecd environment. However, my 40GB IDE harddrive is > recognized as /dev/hda rather than /dev/sda. Same here: workaround of noload pata_qdi works but I now have /dev/hda instead of /dev/sda . install-x86-minimal-20100216.iso on a DELL OptiPlex GX115
Running into the same issue install 2010 livedvd on a thinkpad x32 on a new drive I hang on the following: ata_piix wd7000 ehci_hcd...usbcore (this is weird because the format is non-standard) I get around the first 2 using noload=whatever the last one sticks The hda/sda stuff may or may not be related. I've noticed this on other distros and other machines. It seems since sata inception the drivers have changed on how they report using sda instead of hda. This is has cause me grief and confusion more then once. Hda info from thinkwiki is not confirmed by me but is the most straight forward information on this I have seen. * ata_piix: the disk shows as /dev/sda and DMA is enabled. * Generic IDE driver (ide-disk): the disk shows as /dev/hda and DMA is disabled. The simplest way to enable DMA is to force the IDE driver to ignore the system hard disk by passing the hda=noprobe and hda=none kernel argument. I'll see if I can't get a debug on this before I try the workaround.
I got no debug info, machine locks up hard after each module gets loaded (ata_piix,wd7000,ehci_hcd...usbcore. noload=pata_qdi works. everything loads cleanly using that option
After loading all the modules skipping pata_qdi, I get the following: Making tmpfs for /newroot Looking for the cdrom Attempting to mount media:- /dev/hdc1 Attempting to mount media:- /dev/hdc2 Attempting to mount media:- /dev/hdc3 Media not found No bootable medium found. Waiting for new devices... Looking for the cdrom Attempting to mount media:- /dev/hdc1 Attempting to mount media:- /dev/hdc2 Attempting to mount media:- /dev/hdc3 Media not found Could not find CD to boot, something else needed! Determining root device... Could not find the root block device in .
I can confirm this new behaviour on a D815EEA motherboard, the noload=pata_qdi workaround is no longer sufficient. The system still hangs on boot at the previously described locations but attempts to work around by preventing pata_qdi to load prevents the cdrom from being detected.
(In reply to comment #69) > I can confirm this new behaviour on a D815EEA motherboard, the noload=pata_qdi > workaround is no longer sufficient. The system still hangs on boot at the > previously described locations but attempts to work around by preventing > pata_qdi to load prevents the cdrom from being detected. > Same here on a DELL PowerEdge 1600SC - noload=pata_qdi on a mini-boot-cd from 02/2010 hangs while trying to mount the cdrom - I am currently using a cd from 03/2009, which works.
How is this new? Is there an update out there somewhere?
(In reply to comment #71) > How is this new? > Is there an update out there somewhere? > Can you reproduce this with System Rescue CD? http://www.sysresccd.org/
I can't find this option in the latest stable kernel specs. Was it renamed?
No answer, so closing.
