Bug 5667 - Installer fail finding SATA drive - regression
Summary: Installer fail finding SATA drive - regression
Status: RESOLVED INVALID
Alias: None
Product: Mageia
Classification: Unclassified
Component: Installer (show other bugs)
Version: Cauldron
Hardware: All Linux
Priority: Normal critical
Target Milestone: ---
Assignee: Mageia Bug Squad
QA Contact:
URL:
Whiteboard:
Keywords: NEEDINFO
Depends on:
Blocks:
 
Reported: 2012-04-29 02:10 CEST by Morgan Leijström
Modified: 2012-05-08 23:33 CEST (History)
4 users (show)

See Also:
Source RPM:
CVE:
Status comment:


Attachments
lspcidrake -v using earlier installed cauldron, working great (3.64 KB, text/plain)
2012-04-29 22:40 CEST, Morgan Leijström
Details
report.bug (116.77 KB, text/plain)
2012-04-29 23:14 CEST, Morgan Leijström
Details

Description Morgan Leijström 2012-04-29 02:10:42 CEST
A month or so i did sucessful network install on my machine "metall"

Now, i was to have a go on it again so i downloaded boot.iso and booted it.

The installer fail to find any disk!   It say it found 
"nVidia Corporation|CK804 Serial ATA Controller", "nVidia Corporation|CK804 Serial ATA Controller", "nVidia Corporation|CK804 IDE", "usb_storage"

 and asks if i have more.  I do not, but i do have a drive attached to SATA...

I tried both 64 and 32 bit boot.iso, same result.

The installed system is running happily.

What info can i provide?
Comment 1 Morgan Leijström 2012-04-29 02:32:49 CEST
some libkmod error messages were output before stage2, bug 5669
Manuel Hiebel 2012-04-29 11:15:23 CEST

CC: (none) => pterjan, thierry.vignaud

Comment 2 Thierry Vignaud 2012-04-29 11:34:27 CEST
Please attach (not paste) output of the "lspcidrake -v" command (or "lspci -nnk" if you don't have any mga installed on this machine (or any livecd))?
Also could you insert a floppy or plug a usb key while installing, then when the bug happens, go to the second console (tty2) and type "bug".
Then attach the report file added to the USB key here.

Keywords: (none) => NEEDINFO

Comment 3 Morgan Leijström 2012-04-29 14:16:41 CEST
Tried on another machine, there it found the disk OK:
Working: Lenovo Thinkpad T61, with Intel SSD SATA, current 64 bit mga2
boot.iso.

Not working: the machine first post is an about 7 year old main board with
nvidia nforce2 chipset IIRC, two dualcore opteron, and a pretty recent 2,5"
SATA disk.

I have not changed anything but the graphics card and mouse since i installed
cauldron a month ago.  It is my old workstation under resurrection for casual
use.

Is there some way to provide a log of what it have tried and what it detected?
I can also try to see what the working system is using.  Tell me how!

-MID AIR COLLISION-

OK, thanks Thierry, will do today.
Comment 4 Morgan Leijström 2012-04-29 22:37:56 CEST
This time i see last 5 rows:

Entering step 'Hard drive detection"
cannot get info for device (0:0:0:0) at /usr/lib/libDrakX/detect_devices.pm line 222.
cannot get info for device (1:0:0:0) at /usr/lib/libDrakX/detect_devices.pm line 222.
cannot get info for device (0:0:0:0) at /usr/lib/libDrakX/detect_devices.pm line 222.
cannot get info for device (1:0:0:0) at /usr/lib/libDrakX/detect_devices.pm line 222.


Hm "bug" failed to store on USB, posted Bug 5685
Fortunately that machine have floppy and it looked like it worked.
So now we just wait for Bug 4642 to be resolved so i can read back the result.
Doh!

Maybe i can assemble some junk and DOS or old MSwindows... but i really have no time.

Other idea to try?
Comment 5 Morgan Leijström 2012-04-29 22:40:18 CEST
Created attachment 2138 [details]
lspcidrake -v using earlier installed cauldron, working great
Comment 6 Morgan Leijström 2012-04-29 23:14:06 CEST
Created attachment 2139 [details]
report.bug

OK, booted on sysresccd.org live rescue system, mounted floppy - *just works* - copied to USB stick... and back on my normal workstation i see mga1 seem to have problem mounting USB stick with several partitions or just ext3?  Arrgh format it all ext4, and a next round... Bingo.  - I won this battle.

BTW also sysresccd have no problem finding my SATA.
Comment 7 Pascal Terjan 2012-04-29 23:33:30 CEST
ide_pci_generic : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (subv:10de subd:cb84)
ide_pci_generic : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (subv:1462 subd:6200)
ide_pci_generic : nVidia Corporation|CK804 IDE [STORAGE_IDE] (subv:1462 subd:6200)


Actually trying on my machine, lspcidrake on cauldron reports:

ide_pci_generic : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0055 subv:1043 subd:815a) (rev: f3)
ide_pci_generic : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0054 subv:1043 subd:815a) (rev: f3)
ide_pci_generic : nVidia Corporation|CK804 IDE [STORAGE_IDE] (vendor:10de device:0053 subv:1043 subd:815a) (rev: f2)

While 1 reports:

sata_nv         : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0055 subv:1043 subd:815a) (rev: f3)
sata_nv         : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0054 subv:1043 subd:815a) (rev: f3)
pata_amd        : nVidia Corporation|CK804 IDE [STORAGE_IDE] (vendor:10de device:0053 subv:1043 subd:815a) (rev: f2)

So, something seems to have been broken in ldetect or ldetect-lst
Comment 8 Pascal Terjan 2012-04-29 23:35:26 CEST
Actually I had not updated my cauldron for a few days and it seems better now :)

sata_nv         : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0055 subv:1043 subd:815a) (rev: f3)
sata_nv         : nVidia Corporation|CK804 Serial ATA Controller [STORAGE_IDE] (vendor:10de device:0054 subv:1043 subd:815a) (rev: f3)
amd74xx         : nVidia Corporation|CK804 IDE [STORAGE_IDE] (vendor:10de device:0053 subv:1043 subd:815a) (rev: f2)
Comment 9 Thierry Vignaud 2012-04-30 10:21:42 CEST
But nothing has changed regarding this in either ldetect or ldetect-lst.
Maybe a transient kernel issue? or an issue with files generated by kmod?

CC: (none) => tmb

Comment 10 Morgan Leijström 2012-04-30 10:31:43 CEST
I tried at least twice 32 bit boot.iso, and twice the 64-bit, always using http network install mode and my urpmi-proxy server.

Only thing i have changed since successful install is the PCIe graphics card and USB mouse.  Maybe i should switch back to the old ones an try again?

I have upgraded the running system and it still can boot :)

Maybe you are interesting in some boot log if it say how disk was found.
How to provide?
Comment 11 Pascal Terjan 2012-04-30 12:41:10 CEST
(In reply to comment #9)
> But nothing has changed regarding this in either ldetect or ldetect-lst.
> Maybe a transient kernel issue? or an issue with files generated by kmod?

Can be, there were both new kernel and new kmod in my update. I'll try to do more tests.
Comment 12 Pascal Terjan 2012-04-30 14:14:00 CEST
Reverting to ldetect-lst-0.1.300 gives ide_pci_generic again
Comment 13 Thierry Vignaud 2012-04-30 14:21:05 CEST
uh? The only change was for the Xorg driver for Poulsbo...
Comment 14 Thierry Vignaud 2012-04-30 14:21:39 CEST
But a build time we do copy some kernel files.
New ldetect-lst was build with a newer kernel, thus with a fixed modules.pcimap maybe
Comment 15 Pascal Terjan 2012-04-30 14:25:47 CEST
The ids actually don't appear in the list in both cases...
Comment 16 Pascal Terjan 2012-04-30 14:28:52 CEST
Ah it is listed in /lib/module-init-tools/ldetect-lst-modules.alias but the same way for both versions:

alias pci:v000010DEd00000055sv*sd*bc*i* sata_nv
Comment 17 Dave Hodgins 2012-04-30 21:26:08 CEST
Could it be that the fix for ide controllers is causing the ide module to
be loaded first, and it's grabbing the controller, so that when sata_nv
loads, it can't access it.

I'm thinking it might be the order the modules are loaded, similar to
the problems caused when pata_acpi loaded before pata_via, etc.

On an installed system, rdloaddriver=sata_nv would probably fix it,
but what can be done with the installer, to force that module to load
first?

CC: (none) => davidwhodgins

Comment 18 Morgan Leijström 2012-05-02 10:38:05 CEST
The installed system still works

Just by chance i tried at installer boot "linux rdloaddriver=sata_nv" but it panicked shortly after loading kernel: 
  RAMDISK: gzip image foound at block 0
  List of all partitions:
  No filesystem could mount root, tried: ext2 cramfs
  Kernel Panic - not syncing: VFS: Unable to mount root fs on unknown-block(252,0)

Can i give some command in terminal during install and see if it works?
Manually load correct module or unlioad ide before it tries itself? (How?)

I wish to make a reinstall/dualboot on that machine tomorrow but i can also make base install on that drive using a another machine.
Comment 19 Thierry Vignaud 2012-05-04 01:36:00 CEST
Pascal: since you've access to both packages, can you attach the full diff between two rpm2cpios?
Comment 20 Morgan Leijström 2012-05-08 00:28:20 CEST
Now i got similar result on a PATA on a different machine. Bug 5788
Comment 21 Morgan Leijström 2012-05-08 23:33:45 CEST
STOP!

When i found out more computers had same problem i purged stage2 from urpmi-proxy cache, and then it worked.

Either urpmi-proxy sent old cached version instead of a new available, or it coincided with a new stage 2 becoming available?  -> Bug 5797 

Sorry for the noise here.

Status: NEW => RESOLVED
Resolution: (none) => INVALID


Note You need to log in before you can comment on or make changes to this bug.