Bug 14817

Noticed here too on a computer with card reader on DVD 32.

CC: (none) => eeeemail
Whiteboard: (none) => 5beta2

Can your attach your /root/drakx/report.bug.xz or run the "bug" command on tty2 once the installer displays that error?

Keywords: (none) => NEEDINFO

> Can your attach your /root/drakx/report.bug.xz 

  How does one obtain that, Thierry?

> or run the "bug" command on tty2 once the installer displays that error?

  I have no idea how to do that!
(N.B. 
  The "No medium found" errors appear for only a few seconds (before install files start to be loaded from DVD etc), then are gone.)

In reply to Maurice Batey from comment #3)
> > or run the "bug" command on tty2 once the installer displays that error?
> 
>   I have no idea how to do that!

Just type "alt +ctrl+f2" then "bug" (after having plugged a USB key)

> (N.B. 
>   The "No medium found" errors appear for only a few seconds (before install
> files start to be loaded from DVD etc), then are gone.)

You means in text mode?
Then those are warnings generated by udev(

Priority: release_blocker => Normal
CC: (none) => mageia
Assignee: thierry.vignaud => bugsquad

> Just type "alt +ctrl+f2" then "bug" (after having plugged a USB key)

OK - started install of 64-nit Cauldron fromm DVD.
 See "Loading Linux kernel", then as soon as "Detecting USB devices" appears  the "No medium found" messages are then superimposed on the screen.

Having already plugged in a flash drive I then hit Ctl+Alt+F2, whereupon the screen clears and I see just "sh-4.3#", so I enter the word "bug" and it says "Command not found" and carries on with the install...
  (Nothing written to flash drive.)

Created attachment 5937 [details]
Picture showing the problem

Here's a picture to make it clear. It looks untidy, but doesn't affect the install in any way. The head of /root/drakx/ddebug.log shows:

* getFile install/stage2/VERSION on disk://sdf1/x86_64
* second stage install running (DrakX v16.55)
* running: /usr/lib/systemd/systemd-udevd --daemon --resolve-names=never
* running: udevadm trigger --type=subsystems --action=add
* running: udevadm trigger --type=devices --action=add
* running: modprobe microcode
* getFile install/stage2/VERSION on disk://sdf1/x86_64
error: /dev/sdb: No medium found
error: /dev/sdc: No medium found
error: /dev/sde: No medium found
error: /dev/sdd: No medium found

This is caused by a multi-format card reader that gets mapped into the device tree like this:

<5>[    3.513106] scsi 4:0:0:0: Direct-Access     Generic- Compact Flash    1.00 PQ: 0 ANSI: 0 CCS
<5>[    3.520716] scsi 4:0:0:1: Direct-Access     Generic- SM/xD-Picture    1.00 PQ: 0 ANSI: 0 CCS
<5>[    3.521572] sd 4:0:0:0: [sdb] Attached SCSI removable disk
<5>[    3.524145] sd 4:0:0:1: [sdc] Attached SCSI removable disk
<5>[    3.544070] scsi 4:0:0:2: Direct-Access     Generic- SD/MMC           1.00 PQ: 0 ANSI: 0 CCS
<5>[    3.547496] sd 4:0:0:2: [sdd] Attached SCSI removable disk
<5>[    3.569179] scsi 4:0:0:3: Direct-Access     Generic- MS/MS-Pro        1.00 PQ: 0 ANSI: 0 CCS
<5>[    3.572698] sd 4:0:0:3: [sde] Attached SCSI removable disk

CC: (none) => mageia

Please attach the whole file.

CC: (none) => thierry.vignaud

aka attach a USB key then run the "bug" command on tty2 once the installer displays that error?

I think this message comes from systemd-udevd.
See also https://bugs.freedesktop.org/show_bug.cgi?id=86414

And Lennart basically says that's not your issue, fix kernel drivers.
But in order to know what driver is used, we still need the whole log file.

CC: (none) => tmb
Source RPM: (none) => kernel, udev

See attachment 5791 [details]

Keywords: NEEDINFO => (none)

sorry, attachment 5971 [details] which is attached to bug 15253

Looks like a C-media card reader which would use generic drivers.
Could you add "lsusb -vvt" output too?

lsusb is not available in the installer but this is taken from the installed mga5.

# lsusb -vvt
/:  Bus 02.Port 1: Dev 1, Class=root_hub, Driver=ohci-pci/8p, 12M
    |__ Port 1: Dev 2, If 0, Class=Audio, Driver=snd-usb-audio, 12M
    |__ Port 1: Dev 2, If 1, Class=Audio, Driver=snd-usb-audio, 12M
    |__ Port 1: Dev 2, If 2, Class=Human Interface Device, Driver=usbhid, 12M
    |__ Port 3: Dev 3, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
    |__ Port 4: Dev 4, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
    |__ Port 4: Dev 4, If 1, Class=Human Interface Device, Driver=usbhid, 1.5M
/:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/8p, 480M
    |__ Port 7: Dev 6, If 0, Class=Mass Storage, Driver=usb-storage, 480M


The only external usb devices attached are usb keyboard & mouse. The card reader is an internal one.

sorry, and some usb speakers.

Created attachment 5982 [details]
Output from bug command

Here's the output from the bug command on my machine.

Although it would be better if these error messages weren't emitted at all, they would be far less noticeable if something could be done to convert the line feeds into CRLF pairs when the messages are displayed.

Colin see:
https://www.mail-archive.com/systemd-devel@lists.freedesktop.org/msg28235.html

Assignee: bugsquad => mageia

Same problem still there with 64-bit Mageia-5-RC of 15/3/2015.

(In reply to Thierry Vignaud from comment #17)
> Colin see:
> https://www.mail-archive.com/systemd-devel@lists.freedesktop.org/msg28235.html

I've tested this patch in an installed copy of M5, and it appears to work without any obvious ill effects.

I'm sorry to report that exactly the same problem still occurs with the classic iso for 64-bit Mageia-5-RC of 23/3/2015, on my desktop (real h/w).

(As Martin said above it would be far less noticeable if something could be done to convert the line feeds into CRLF pairs when the messages are displayed.)

Problem still there in 23/3/15 classical iso for 64-bit Mageia-5RC (KDE) on real h/w.

Seeing this in the Classical i586 DVD too, after installing a generic internal multi-format card reader.

FWIW, if I were to use the F12 option to bypass the normal boot preferences, each of the card reader slots appears as a potential boot device, whether it contains media or not. Could this then be a BIOS-related error?

CC: (none) => andrewsfarm

(In reply to Thomas Andrews from comment #22)
> FWIW, if I were to use the F12 option to bypass the normal boot preferences,
> each of the card reader slots appears as a potential boot device, whether it
> contains media or not. Could this then be a BIOS-related error?

Not really. The device exists because the card reader is installed in the system. It is quite reasonable that it should report "no medium found" when it is probed to see if a card is present. We just don't want to see those messages splashed across the screen!

As noted above, the patch Thierry found seems to fix the problem quite nicely.

Regret problem persists in latest (1/4/2015) 64-bit Mageia-5RC iso, booted from DVD on real h/w.

Still there with the latest Classical i586 RC, too. Also when booting from DVD on real hardware.

Also seeing it on i586 when booting from a USB device.

Regret problem persists in latest (9/4/2015) 64-bit Mageia-5RC iso, booted from DVD on real h/w, Legacy BIOS.

Valid April 15th ISOs

As Colin is MIA at the moment, is there anyone else qualified to decide whether to apply the patch that fixes this? As I reported earlier, it works on my system with no obvious ill effects.

I know this is largely a cosmetic issue, but it is a rather noticeable one!

Yeah, I've not really got much time at present :(

But these patches were not committed upstream so I'd rather see it progressed there first. Unless I've read this bug incorrectly, this is just a cosmetic issue, so I wouldn't be too keen on adding non-upstreamed patches for that.

(FWIW, I've pinged the upstream thread).

FWIW, this should also address the issue sufficiently. It's upstream so I'll add it in if I get a moment (not sure how likely that is tho!)

http://cgit.freedesktop.org/systemd/systemd/commit/?id=8d8ce9e2cd066e90c17e2d1eb1882defabb1fa63

Unfortunately that relies on

http://cgit.freedesktop.org/systemd/systemd/commit/?id=086891e5c119abb9854237fc32e736fe2d67234c

which is a lot more invasive. But I've back-ported the patch, and am about to test it now...

Created attachment 6395 [details]
Backported patch from upstream

Here's the back-ported patch. I've tested it on an installed system and in stage 2 of the classic installer, and it does the job. I don't know a way to modify the stage 1 initrd, so can't test it there.

Just rebuild the drakx-installer-images package.
Then use the boot.iso it contains

Thanks Thierry. I can confirm the patch fixes the problem in stage 1 of the installer (which is what this bug is really about).

N.B. There looks to be something wrong with the build requires of the drakx-installer-images package on a 64-bit system - 'urpmi --buildrequires' attempts (and fails) to install the 32-bit kernel package, and when you subsequently run rpmbuild, it reports a missing dependency on the grub2-efi package.

Ping. This bug primarily affects the installer, so can't be fixed after release.

IIRC udev should be patched using the patch in attachment 6395 [details], and that does the trick as soon as stage1 will be rebuilt?

If so I can add the patch now, it seems harmless.

(In reply to Rémi Verschelde from comment #38)
> IIRC udev should be patched using the patch in attachment 6395 [details],
> and that does the trick as soon as stage1 will be rebuilt?
> 
> If so I can add the patch now, it seems harmless.

Yeah that looks fine to me, please go ahead if you think it's important enough (personally it seems like a very minor issue to me, but hey ho).

That would make one less bug report open, and add to the visual smoothness of the installation process, so if the patch is indeed harmless, +1

I also think attachment 5937 [details] is ugly enough to warrant a cosmetic patch :)

You'll still see such message when booting on fake raid such as IMSM...

Patch added to systemd-217-9.mga5.

Closing then as drakx-installer-images has been rebuild with it

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

Very happy to report that the problem has indeed gone away!
(23/5 64-bit Mageia-5RC installed from DVD).