Re: Possibly erroneous "device not present" message during boot

[Brian]

On Tue 25 Apr 2017 at 14:09:15 +0100, Brian wrote:

> On Mon 24 Apr 2017 at 15:10:16 -0500, David Wright wrote:
> 
> > On Mon 24 Apr 2017 at 19:37:22 (+0100), Brian wrote:
> > > 
> > > A modicum of reassurance and help is never wasted, particularly for
> > > those users who come to this thread in the future.
> > > 
> > > My laptop has a single slotted reader on the PCI bus.
> > 
> > Before booting, can you see the SD card's device in the CMOS screens?
> 
> No. Hard disk and CD are the only offerings. That's with the card in the
> slot when the machine is started.
> 
> > > The installer
> > > boots and shows that mmc_core has been loaded. When it gets to the
> > > partitioning stage the SD card is not offered as an option. The
> > > module mmc_block is absent from 'lsmod' and does not appear when the
> > > card is taken out and reinserted. It is not detected.
> > 
> > If there's no driver, would you expect the kernel to react?
> 
> I would not. But your remark caused *me* to react by taking a closer
> look at the modules loaded on Jessie.
> 
>   brian@laptop:~$ lsmod | grep mmc
>   mmc_block  30466  0 
>   mmc_core   91803  4 mmc_block,sdhci,tifm_sd,sdhci_pci
>   brian@laptop:~$ lsmod | grep tifm
>   tifm_sd17060  0 
>   tifm_7xx1  12769  0 
>   tifm_core  13113  2 tifm_7xx1,tifm_sd
>   mmc_core   91803  4 mmc_block,sdhci,tifm_sd,sdhci_pci
> 
> 'rmmod tifm_7xx1' causes /dev/mmcblk01p to disappear. Booting without it
> present leads to no device file. Stating the obvious, the card will not
> work without the tifm_7xx1 module.
> 
> In the installer tifm_7xx1 should be in /lib/modules/./drivers/misc/.
> It is not present.
> 
> > Is there an mmc driver in that installation moduoles screen?
> 
> No.
>  
> > It might be hard to spot, but is there a /proc/interrupts file,
> > and does the number of interrupts increase on the appropriate line
> > when you insert and remove the card?
> > 
> > How do you find the line. On my laptop with that sort of SD card,
> >  18:   2344  0   IO-APIC-fasteoi   mmc0
> > the 2344 increases by 1 when I take the card out and by many when
> > I reinsert it. I don't know if mmc_core can provide that line in
> > the absense of mmc_block. (Obviously my kernel has both loaded now.)
> 
> In the installer environment that number does not change when the card
> is taken out and pushed back in. In the Jessie environment it changes
> all the time, card in or out.
>  
> > > What is unusual is that the card is detected by a Jessie OS on being
> > > inserted. Is this an installer problem with different hardware (my
> > > laptop's is described in another post) or with the card? Basically,
> > > is an installation to an SD card on a PCI bus a case of hit or miss?
> 
> It very much looks like an installer bug. It is present on Stretch and
> Jessie.
> 
> The hardware:
> 
>   brian@laptop:~$ lspci | grep Texas  
> >   00:10.0 CardBus bridge: Texas 
> Instruments PCIxx21/x515 Cardbus Controller   
> 
>   00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 
> Host Controller>   00:10.3 Mass storage 
> controller: Texas Instruments PCIxx21 Integrated FlashMedia Controller
>   00:10.4 SD Host controller: Texas Instruments 
> PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller

We have a happy ending:

 
http://metadata.ftp-master.debian.org/changelogs/main/l/linux/linux_4.9.25-1_changelog

  linux (4.9.25-1) unstable; urgency=medium

  * udeb: Add tifm_7xx1 to mmc-modules (Closes: #861195)

The modicum of help wasn't wasted. Kudos to everyone who participated in
this thread. :)

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 22 Apr 2017 at 09:43:25 -0500, Richard Owlett wrote:

> On 04/22/2017 08:45 AM, Brian wrote:
> >
> >Any better?
> >
> >  brian@laptop:~$ lspci | grep Texas
> >  00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller
> >  00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 
> > Host Controller
> >  00:10.3 Mass storage controller: Texas Instruments PCIxx21 Integrated 
> > FlashMedia Controller
> >  00:10.4 SD Host controller: Texas Instruments 
> > PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller
> >
> 
> On the Lenovo ThinkPad T510 which started this thread I ran lspci as root. I
> believe the relevant lines are:
> 0d:00.0 SD Host controller: Ricoh Co Ltd MMC/SD Host Controller (rev 01)
> 0d:00.1 System peripheral: Ricoh Co Ltd R5U2xx (R5U230 / R5U231 / R5U241)
> [Memory Stick Host Controller] (rev 01)

I think only the sdhci, sdhci_pci and mmc* modules are required to make
this card visible to the installer and available for partitioning. (The
output of 'lsmod' would be the place to look for confirmation). The
installer's initrd contains these modules so the card would be seen.

You were doubly fortunate at the store. Having the right kind of card
reader made it possible to install Debian on the card. Having the
kernels on card and hard disk identical made it possible for the OS on
the card to boot.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Mon 24 Apr 2017 at 15:10:16 -0500, David Wright wrote:

> On Mon 24 Apr 2017 at 19:37:22 (+0100), Brian wrote:
> > 
> > A modicum of reassurance and help is never wasted, particularly for
> > those users who come to this thread in the future.
> > 
> > My laptop has a single slotted reader on the PCI bus.
> 
> Before booting, can you see the SD card's device in the CMOS screens?

No. Hard disk and CD are the only offerings. That's with the card in the
slot when the machine is started.

> > The installer
> > boots and shows that mmc_core has been loaded. When it gets to the
> > partitioning stage the SD card is not offered as an option. The
> > module mmc_block is absent from 'lsmod' and does not appear when the
> > card is taken out and reinserted. It is not detected.
> 
> If there's no driver, would you expect the kernel to react?

I would not. But your remark caused *me* to react by taking a closer
look at the modules loaded on Jessie.

  brian@laptop:~$ lsmod | grep mmc
  mmc_block  30466  0 
  mmc_core   91803  4 mmc_block,sdhci,tifm_sd,sdhci_pci
  brian@laptop:~$ lsmod | grep tifm
  tifm_sd17060  0 
  tifm_7xx1  12769  0 
  tifm_core  13113  2 tifm_7xx1,tifm_sd
  mmc_core   91803  4 mmc_block,sdhci,tifm_sd,sdhci_pci

'rmmod tifm_7xx1' causes /dev/mmcblk01p to disappear. Booting without it
present leads to no device file. Stating the obvious, the card will not
work without the tifm_7xx1 module.

In the installer tifm_7xx1 should be in /lib/modules/./drivers/misc/.
It is not present.

> Is there an mmc driver in that installation moduoles screen?

No.
 
> It might be hard to spot, but is there a /proc/interrupts file,
> and does the number of interrupts increase on the appropriate line
> when you insert and remove the card?
> 
> How do you find the line. On my laptop with that sort of SD card,
>  18:   2344  0   IO-APIC-fasteoi   mmc0
> the 2344 increases by 1 when I take the card out and by many when
> I reinsert it. I don't know if mmc_core can provide that line in
> the absense of mmc_block. (Obviously my kernel has both loaded now.)

In the installer environment that number does not change when the card
is taken out and pushed back in. In the Jessie environment it changes
all the time, card in or out.
 
> > What is unusual is that the card is detected by a Jessie OS on being
> > inserted. Is this an installer problem with different hardware (my
> > laptop's is described in another post) or with the card? Basically,
> > is an installation to an SD card on a PCI bus a case of hit or miss?

It very much looks like an installer bug. It is present on Stretch and
Jessie.

The hardware:

  brian@laptop:~$ lspci | grep Texas

  00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller 

  00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 Host 
Controller  
  00:10.3 Mass storage controller: Texas Instruments PCIxx21 Integrated 
FlashMedia Controller
  00:10.4 SD Host controller: Texas Instruments 
PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Mon 24 Apr 2017 at 19:37:22 (+0100), Brian wrote:
> On Sun 23 Apr 2017 at 20:46:39 +0100, Brian wrote:
> 
> > On Wed 12 Apr 2017 at 06:18:51 -0500, Richard Owlett wrote:
> > 
> > > Due to circumstances, will note rule out any of hardware, software, or
> > > confused operator as underlying cause.
> > > 
> > > BACKGROUND:
> > > The hardware is a used Lenovo T510 laptop purchased a couple of months 
> > > ago.
> > 
> > This machine has an internal card reader.
> > 
> > > Jessie with MATE desktop has been running fine.
> > > Yesterday I purchased a Kingston Technology 512GB SD, primarily for 
> > > backups.
> > > 
> > > As this individual unit had never had an SD card installed, the agreement
> > > with the store was if the card did not work in this machine I would not 
> > > have
> > > to purchase it. My acceptance test was to install Jessie using a flash 
> > > drive
> > > with DVD 1 of 13 of Debian 8.6.0.
> > 
> > We suppose the flash drive was a USB stick.
> >  
> > > It was my standard install which does NOT include either a swap partition
> > > being activated during install NOR installing a boot loader.
> > 
> > First (and the most important question): was the SD card inserted in a
> > slot of the *machine's* internal card reader when installing Debian to
> > it?
> > 
> > > I booted into my primary OS and ran update-grub with no problems.
> > > I rebooted, choosing the SD card from the Grub2 menu. No apparent problem.
> > > Also verified my other Debian installs could read/write the SD card.
> > > 
> > > These tests were performed before leaving the store.
> > > 
> > > THE SYMPTOM:
> > > 
> > > When I got home and booted to the SD card I noticed a message appeared for
> > > ~10 seconds reporting that a device was not found giving the UUID of the 
> > > SD
> > > card.
> > 
> > If I had carried out the procedure you describe (something that had not
> > been done before) my eyes would have been glued to the screen when the
> > OS on the SD card had been booted for the first time. Yet, at the store,
> > there was no "device not found" message observed. Did your attention
> > wander for the 10 seconds such a message would have been on the screen? :)
> 
> A modicum of reassurance and help is never wasted, particularly for
> those users who come to this thread in the future.
> 
> My laptop has a single slotted reader on the PCI bus.

Before booting, can you see the SD card's device in the CMOS screens?

> The installer
> boots and shows that mmc_core has been loaded. When it gets to the
> partitioning stage the SD card is not offered as an option. The
> module mmc_block is absent from 'lsmod' and does not appear when the
> card is taken out and reinserted. It is not detected.

If there's no driver, would you expect the kernel to react?
Is there an mmc driver in that installation moduoles screen?

It might be hard to spot, but is there a /proc/interrupts file,
and does the number of interrupts increase on the appropriate line
when you insert and remove the card?

How do you find the line. On my laptop with that sort of SD card,
 18:   2344  0   IO-APIC-fasteoi   mmc0
the 2344 increases by 1 when I take the card out and by many when
I reinsert it. I don't know if mmc_core can provide that line in
the absense of mmc_block. (Obviously my kernel has both loaded now.)

> What is unusual is that the card is detected by a Jessie OS on being
> inserted. Is this an installer problem with different hardware (my
> laptop's is described in another post) or with the card? Basically,
> is an installation to an SD card on a PCI bus a case of hit or miss?
> 
> ALT-F2 at the installer's first screen gives a console.
> 
>  lsmod | grep mmc
> 
> indicates whether both modules are loaded and the card is detected.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sun 23 Apr 2017 at 20:46:39 +0100, Brian wrote:

> On Wed 12 Apr 2017 at 06:18:51 -0500, Richard Owlett wrote:
> 
> > Due to circumstances, will note rule out any of hardware, software, or
> > confused operator as underlying cause.
> > 
> > BACKGROUND:
> > The hardware is a used Lenovo T510 laptop purchased a couple of months ago.
> 
> This machine has an internal card reader.
> 
> > Jessie with MATE desktop has been running fine.
> > Yesterday I purchased a Kingston Technology 512GB SD, primarily for backups.
> > 
> > As this individual unit had never had an SD card installed, the agreement
> > with the store was if the card did not work in this machine I would not have
> > to purchase it. My acceptance test was to install Jessie using a flash drive
> > with DVD 1 of 13 of Debian 8.6.0.
> 
> We suppose the flash drive was a USB stick.
>  
> > It was my standard install which does NOT include either a swap partition
> > being activated during install NOR installing a boot loader.
> 
> First (and the most important question): was the SD card inserted in a
> slot of the *machine's* internal card reader when installing Debian to
> it?
> 
> > I booted into my primary OS and ran update-grub with no problems.
> > I rebooted, choosing the SD card from the Grub2 menu. No apparent problem.
> > Also verified my other Debian installs could read/write the SD card.
> > 
> > These tests were performed before leaving the store.
> > 
> > THE SYMPTOM:
> > 
> > When I got home and booted to the SD card I noticed a message appeared for
> > ~10 seconds reporting that a device was not found giving the UUID of the SD
> > card.
> 
> If I had carried out the procedure you describe (something that had not
> been done before) my eyes would have been glued to the screen when the
> OS on the SD card had been booted for the first time. Yet, at the store,
> there was no "device not found" message observed. Did your attention
> wander for the 10 seconds such a message would have been on the screen? :)

A modicum of reassurance and help is never wasted, particularly for
those users who come to this thread in the future.

My laptop has a single slotted reader on the PCI bus. The installer
boots and shows that mmc_core has been loaded. When it gets to the
partitioning stage the SD card is not offered as an option. The
module mmc_block is absent from 'lsmod' and does not appear when the
card is taken out and reinserted. It is not detected.

What is unusual is that the card is detected by a Jessie OS on being
inserted. Is this an installer problem with different hardware (my
laptop's is described in another post) or with the card? Basically,
is an installation to an SD card on a PCI bus a case of hit or miss?

ALT-F2 at the installer's first screen gives a console.

 lsmod | grep mmc

indicates whether both modules are loaded and the card is detected.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 12 Apr 2017 at 06:18:51 -0500, Richard Owlett wrote:

> Due to circumstances, will note rule out any of hardware, software, or
> confused operator as underlying cause.
> 
> BACKGROUND:
> The hardware is a used Lenovo T510 laptop purchased a couple of months ago.

This machine has an internal card reader.

> Jessie with MATE desktop has been running fine.
> Yesterday I purchased a Kingston Technology 512GB SD, primarily for backups.
> 
> As this individual unit had never had an SD card installed, the agreement
> with the store was if the card did not work in this machine I would not have
> to purchase it. My acceptance test was to install Jessie using a flash drive
> with DVD 1 of 13 of Debian 8.6.0.

We suppose the flash drive was a USB stick.
 
> It was my standard install which does NOT include either a swap partition
> being activated during install NOR installing a boot loader.

First (and the most important question): was the SD card inserted in a
slot of the *machine's* internal card reader when installing Debian to
it?

> I booted into my primary OS and ran update-grub with no problems.
> I rebooted, choosing the SD card from the Grub2 menu. No apparent problem.
> Also verified my other Debian installs could read/write the SD card.
> 
> These tests were performed before leaving the store.
> 
> THE SYMPTOM:
> 
> When I got home and booted to the SD card I noticed a message appeared for
> ~10 seconds reporting that a device was not found giving the UUID of the SD
> card.

If I had carried out the procedure you describe (something that had not
been done before) my eyes would have been glued to the screen when the
OS on the SD card had been booted for the first time. Yet, at the store,
there was no "device not found" message observed. Did your attention
wander for the 10 seconds such a message would have been on the screen? :)

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 22 Apr 2017 at 13:34:41 (-0500), Richard Owlett wrote:
> On 04/22/2017 01:00 PM, Brian wrote:
> >On Sat 22 Apr 2017 at 11:56:27 -0500, Richard Owlett wrote:
> >
> >>On 04/22/2017 06:47 AM, Brian wrote:
> >>>On Wed 19 Apr 2017 at 15:16:52 -0500, Richard Owlett wrote:
> >>>
> On 04/19/2017 02:31 PM, Brian wrote:
> >
> >As I have detailed elsewhere,
> >
> >https://lists.debian.org/debian-user/2017/04/msg00578.html
> >
> >the card becomes visible to GRUB with its nativedisk command. The OP has
> >a reader which does not show an SD disk in GRUB and does not boot from
> >it. Perhaps he could take some time to see whether this command does
> >anything for him.
> 
> I just received my USB SD card reader this afternoon. Attempting to
> reproduce the results of your ...msg00578 is on my TODO list. I'll not 
> have
> time for Debian until Monday. I did have the opportunity to do a quickie
> >>>
> >>>Thanks for considering testing. However, it is likely not to work for
> >>>you because your 5-in-1 appparently is a Ricoh reader on the PCI bus.
> >>>You can check this with 'lspci'. Failure is also indicated by the
> >>>presence of 'root=/dev/mmcblk0p1' in your grub.cfg. AFAICT, GRUB does
> >>>not have modules for SD/MMC devices, so nativedisk would have nothing
> >>>to load to drive the device.
> >>>
> >>
> >>Monday came early - IOW the project I was working on collapsed and I needed
> >>a break.
> >>
> >>I attempted to run the tests you described in Paragraph 4 of
> >>.../msg00578.html.
> >>My notes are:
> >>>
> >>
> >>>with no usb devices present, I get
> >>
> >>>(ahci) (ahci0,msdosXX) (ahci1)
> >>
> >>>where XX maches the partions on my "hard disk" {it is actually a
> >>>solid state drive which was manufactured to be a plugin replacement
> >>>for original mechanical dive (aka 'spinning rust'}.
> >
> >I have no experience with SSD but am with you up to here.
> >
> >>>I don't know what (ahci1) is.
> >
> >Neither do I.
> >
> >>>Repeating above with empty card reader I get:
> >>
> >>>(ahci) (ahci0,msdosXX) (usb0a) (usb0b) (ahci1)
> >
> >I assume 'nativedisk' has been run prior to getting this output. GRUB
> >now sees USB devices. They can only have come from the card reader.
> >
> >>>I repeated the above with the SD card plugged into a USB card reader
> >>>getting:
> >>
> >>>(ahci) (ahci0,msdosXX) (usb0a) (usb0a,msdos1) (usb0b) (ahci1)
> >>
> >>>When not using "nativedisk" in the second case I get:
> >>
> >>>(hd0) (hd0,msdosXX) (hd1) (hd1,msdos1)
> >
> >I'd take that to be the SSD disk and whatever is in the reader. (hd1)
> >has a partition. In principle, it could be booted from.
> >
> >>I'll not attempt "Edit /etc/grub.d/40_custom to have ..."
> >
> >When I spoke of "5-in-1 reader" I was using your description in
> >
> > https://lists.debian.org/debian-user/2017/04/msg00505.html
> >
> >IOW, I was referring to the reader which is inbuilt into the Lenovo. Are
> >these data for when your recently purchased USB reader has been used? If
> >not, there is some thinking to be done.
> >
> 
> Any line containing "(ahci)" was after having used "nativedisk".
> Any line containing "(hd0)"  was after *NOT* having used "nativedisk".
> The only time the SD card was seen was when the USB reader was used.
> All lines followed a power off/on cycle to force identical starting
> conditions.
> 
> My reference to "5-in-1 reader" was reference to
> https://www.cnet.com/products/lenovo-thinkpad-t510/specs/ .
> I have no means to verify CNET's statement.
> 
> I'd place more trust in my statement in
> https://lists.debian.org/debian-user/2017/04/msg00729.html saying:
> On the Lenovo ThinkPad T510 which started this thread I ran lspci as root.
> I believe the relevant lines are:
> 0d:00.0 SD Host controller: Ricoh Co Ltd MMC/SD Host Controller (rev 01)
> 0d:00.1 System peripheral: Ricoh Co Ltd R5U2xx (R5U230 / R5U231 /
> R5U241) [Memory Stick Host Controller] (rev 01)

That looks about right. I'd be interested to know what the BIOS can
see, ie before Grub has loaded. I would predict you can see the USB
but not the MMC/SD card slot. (Note that you will probably need to
insert something in them to stand _any_ chance of seeing any of them.)

BTW did you notice that the webpage gave the battery life as 7.8 sec!

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 22 Apr 2017 at 13:34:41 -0500, Richard Owlett wrote:

> On 04/22/2017 01:00 PM, Brian wrote:
> >
> >When I spoke of "5-in-1 reader" I was using your description in
> >
> > https://lists.debian.org/debian-user/2017/04/msg00505.html
> >
> >IOW, I was referring to the reader which is inbuilt into the Lenovo. Are
> >these data for when your recently purchased USB reader has been used? If
> >not, there is some thinking to be done.
> 
> Any line containing "(ahci)" was after having used "nativedisk".
> Any line containing "(hd0)"  was after *NOT* having used "nativedisk".

That makes things clearer.

> The only time the SD card was seen was when the USB reader was used.

That is exactly the information wanted. When the card is on the PCI bus
GRUB does not see it; on the USB bus it does. It fits the lspci info you
give below.

> All lines followed a power off/on cycle to force identical starting
> conditions.
> 
> My reference to "5-in-1 reader" was reference to
> https://www.cnet.com/products/lenovo-thinkpad-t510/specs/ .
> I have no means to verify CNET's statement.
> 
> I'd place more trust in my statement in
> https://lists.debian.org/debian-user/2017/04/msg00729.html saying:
> On the Lenovo ThinkPad T510 which started this thread I ran lspci as root.
> I believe the relevant lines are:
> 0d:00.0 SD Host controller: Ricoh Co Ltd MMC/SD Host Controller (rev 01)
> 0d:00.1 System peripheral: Ricoh Co Ltd R5U2xx (R5U230 / R5U231 / R5U241)
> [Memory Stick Host Controller] (rev 01)

> Side question
> When interrupting the grub menu by typing C, is there any functional
> equivalent of using the "script" command in a terminal window?

I'm afraid not.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/22/2017 01:00 PM, Brian wrote:

On Sat 22 Apr 2017 at 11:56:27 -0500, Richard Owlett wrote:


On 04/22/2017 06:47 AM, Brian wrote:

On Wed 19 Apr 2017 at 15:16:52 -0500, Richard Owlett wrote:


On 04/19/2017 02:31 PM, Brian wrote:


As I have detailed elsewhere,

https://lists.debian.org/debian-user/2017/04/msg00578.html

the card becomes visible to GRUB with its nativedisk command. The OP has
a reader which does not show an SD disk in GRUB and does not boot from
it. Perhaps he could take some time to see whether this command does
anything for him.


I just received my USB SD card reader this afternoon. Attempting to
reproduce the results of your ...msg00578 is on my TODO list. I'll not have
time for Debian until Monday. I did have the opportunity to do a quickie


Thanks for considering testing. However, it is likely not to work for
you because your 5-in-1 appparently is a Ricoh reader on the PCI bus.
You can check this with 'lspci'. Failure is also indicated by the
presence of 'root=/dev/mmcblk0p1' in your grub.cfg. AFAICT, GRUB does
not have modules for SD/MMC devices, so nativedisk would have nothing
to load to drive the device.



Monday came early - IOW the project I was working on collapsed and I needed
a break.

I attempted to run the tests you described in Paragraph 4 of
.../msg00578.html.
My notes are:





with no usb devices present, I get



(ahci) (ahci0,msdosXX) (ahci1)



where XX maches the partions on my "hard disk" {it is actually a
solid state drive which was manufactured to be a plugin replacement
for original mechanical dive (aka 'spinning rust'}.


I have no experience with SSD but am with you up to here.


I don't know what (ahci1) is.


Neither do I.


Repeating above with empty card reader I get:



(ahci) (ahci0,msdosXX) (usb0a) (usb0b) (ahci1)


I assume 'nativedisk' has been run prior to getting this output. GRUB
now sees USB devices. They can only have come from the card reader.


I repeated the above with the SD card plugged into a USB card reader
getting:



(ahci) (ahci0,msdosXX) (usb0a) (usb0a,msdos1) (usb0b) (ahci1)



When not using "nativedisk" in the second case I get:



(hd0) (hd0,msdosXX) (hd1) (hd1,msdos1)


I'd take that to be the SSD disk and whatever is in the reader. (hd1)
has a partition. In principle, it could be booted from.


I'll not attempt "Edit /etc/grub.d/40_custom to have ..."


When I spoke of "5-in-1 reader" I was using your description in

 https://lists.debian.org/debian-user/2017/04/msg00505.html

IOW, I was referring to the reader which is inbuilt into the Lenovo. Are
these data for when your recently purchased USB reader has been used? If
not, there is some thinking to be done.



Any line containing "(ahci)" was after having used "nativedisk".
Any line containing "(hd0)"  was after *NOT* having used "nativedisk".
The only time the SD card was seen was when the USB reader was used.
All lines followed a power off/on cycle to force identical starting 
conditions.


My reference to "5-in-1 reader" was reference to
https://www.cnet.com/products/lenovo-thinkpad-t510/specs/ .
I have no means to verify CNET's statement.

I'd place more trust in my statement in
https://lists.debian.org/debian-user/2017/04/msg00729.html saying:
On the Lenovo ThinkPad T510 which started this thread I ran lspci as root.
I believe the relevant lines are:
0d:00.0 SD Host controller: Ricoh Co Ltd MMC/SD Host Controller (rev 01)
0d:00.1 System peripheral: Ricoh Co Ltd R5U2xx (R5U230 / R5U231 / 
R5U241) [Memory Stick Host Controller] (rev 01)


Side question
When interrupting the grub menu by typing C, is there any functional 
equivalent of using the "script" command in a terminal window?

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 22 Apr 2017 at 19:00:50 +0100, Brian wrote:

> When I spoke of "5-in-1 reader" I was using your description in
> 
>  https://lists.debian.org/debian-user/2017/04/msg00505.html
> 
> IOW, I was referring to the reader which is inbuilt into the Lenovo. Are
> these data for when your recently purchased USB reader has been used? If
> not, there is some thinking to be done.

All you have to do is

1. Boot with the SD card in a slot of the Lenovo's reader.

2. Use the c key at the GRUB prompt and type 'ls'. Provide us with the
   output.

3. Issue the command 'nativedisk' and provide the output of 'ls' again.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 22 Apr 2017 at 11:56:27 -0500, Richard Owlett wrote:

> On 04/22/2017 06:47 AM, Brian wrote:
> >On Wed 19 Apr 2017 at 15:16:52 -0500, Richard Owlett wrote:
> >
> >>On 04/19/2017 02:31 PM, Brian wrote:
> >>>
> >>>As I have detailed elsewhere,
> >>>
> >>>https://lists.debian.org/debian-user/2017/04/msg00578.html
> >>>
> >>>the card becomes visible to GRUB with its nativedisk command. The OP has
> >>>a reader which does not show an SD disk in GRUB and does not boot from
> >>>it. Perhaps he could take some time to see whether this command does
> >>>anything for him.
> >>
> >>I just received my USB SD card reader this afternoon. Attempting to
> >>reproduce the results of your ...msg00578 is on my TODO list. I'll not have
> >>time for Debian until Monday. I did have the opportunity to do a quickie
> >
> >Thanks for considering testing. However, it is likely not to work for
> >you because your 5-in-1 appparently is a Ricoh reader on the PCI bus.
> >You can check this with 'lspci'. Failure is also indicated by the
> >presence of 'root=/dev/mmcblk0p1' in your grub.cfg. AFAICT, GRUB does
> >not have modules for SD/MMC devices, so nativedisk would have nothing
> >to load to drive the device.
> >
> 
> Monday came early - IOW the project I was working on collapsed and I needed
> a break.
> 
> I attempted to run the tests you described in Paragraph 4 of
> .../msg00578.html.
> My notes are:
> >
> 
> >with no usb devices present, I get
> 
> >(ahci) (ahci0,msdosXX) (ahci1)
> 
> >where XX maches the partions on my "hard disk" {it is actually a
> >solid state drive which was manufactured to be a plugin replacement
> >for original mechanical dive (aka 'spinning rust'}.

I have no experience with SSD but am with you up to here.

> >I don't know what (ahci1) is.

Neither do I.

> >Repeating above with empty card reader I get:
> 
> >(ahci) (ahci0,msdosXX) (usb0a) (usb0b) (ahci1)

I assume 'nativedisk' has been run prior to getting this output. GRUB
now sees USB devices. They can only have come from the card reader.
 
> >I repeated the above with the SD card plugged into a USB card reader
> >getting:
> 
> >(ahci) (ahci0,msdosXX) (usb0a) (usb0a,msdos1) (usb0b) (ahci1)
> 
> >When not using "nativedisk" in the second case I get:
> 
> >(hd0) (hd0,msdosXX) (hd1) (hd1,msdos1)

I'd take that to be the SSD disk and whatever is in the reader. (hd1)
has a partition. In principle, it could be booted from. 
 
> I'll not attempt "Edit /etc/grub.d/40_custom to have ..."

When I spoke of "5-in-1 reader" I was using your description in

 https://lists.debian.org/debian-user/2017/04/msg00505.html

IOW, I was referring to the reader which is inbuilt into the Lenovo. Are
these data for when your recently purchased USB reader has been used? If
not, there is some thinking to be done.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/22/2017 06:47 AM, Brian wrote:

On Wed 19 Apr 2017 at 15:16:52 -0500, Richard Owlett wrote:


On 04/19/2017 02:31 PM, Brian wrote:


As I have detailed elsewhere,

https://lists.debian.org/debian-user/2017/04/msg00578.html

the card becomes visible to GRUB with its nativedisk command. The OP has
a reader which does not show an SD disk in GRUB and does not boot from
it. Perhaps he could take some time to see whether this command does
anything for him.


I just received my USB SD card reader this afternoon. Attempting to
reproduce the results of your ...msg00578 is on my TODO list. I'll not have
time for Debian until Monday. I did have the opportunity to do a quickie


Thanks for considering testing. However, it is likely not to work for
you because your 5-in-1 appparently is a Ricoh reader on the PCI bus.
You can check this with 'lspci'. Failure is also indicated by the
presence of 'root=/dev/mmcblk0p1' in your grub.cfg. AFAICT, GRUB does
not have modules for SD/MMC devices, so nativedisk would have nothing
to load to drive the device.



Monday came early - IOW the project I was working on collapsed and I 
needed a break.


I attempted to run the tests you described in Paragraph 4 of 
.../msg00578.html.

My notes are:





with no usb devices present, I get



(ahci) (ahci0,msdosXX) (ahci1)



where XX maches the partions on my "hard disk" {it is actually a
solid state drive which was manufactured to be a plugin replacement
for original mechanical dive (aka 'spinning rust'}.



I don't know what (ahci1) is.







Repeating above with empty card reader I get:



(ahci) (ahci0,msdosXX) (usb0a) (usb0b) (ahci1)







I repeated the above with the SD card plugged into a USB card reader
getting:



(ahci) (ahci0,msdosXX) (usb0a) (usb0a,msdos1) (usb0b) (ahci1)







When not using "nativedisk" in the second case I get:



(hd0) (hd0,msdosXX) (hd1) (hd1,msdos1)



I'll not attempt "Edit /etc/grub.d/40_custom to have ..."

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/22/2017 08:45 AM, Brian wrote:

On Sat 22 Apr 2017 at 15:35:33 +0200, Pascal Hambourg wrote:


Le 22/04/2017 à 14:50, Brian a écrit :

On Wed 19 Apr 2017 at 21:10:36 -0500, David Wright wrote:


My guess from this is that the former SD card is connected to the
USB bus whereas the latter is connected to a different bus. The
likely candidate from what I've read is the PCIe bus.


Good guess.

brian@laptop:~$ lspci | grep CardBus
00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller


Err, what is the relationship between a CardBus (formerly PCMCIA) adapter
and a SD card reader ?

Aren't you confusing with something like this :


More than likely. Thanks.


0b:00.1 SD Host controller: Ricoh Co Ltd R5C822 SD/SDIO/MMC/MS/MSPro Host
Adapter (rev 13)
Kernel driver in use: sdhci-pci

(taken on my own ThinkPad, but what the heck, it does not have any SD card
slot !)


Any better?

  brian@laptop:~$ lspci | grep Texas
  00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller
  00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 Host 
Controller
  00:10.3 Mass storage controller: Texas Instruments PCIxx21 Integrated 
FlashMedia Controller
  00:10.4 SD Host controller: Texas Instruments 
PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller



On the Lenovo ThinkPad T510 which started this thread I ran lspci as 
root. I believe the relevant lines are:

0d:00.0 SD Host controller: Ricoh Co Ltd MMC/SD Host Controller (rev 01)
0d:00.1 System peripheral: Ricoh Co Ltd R5U2xx (R5U230 / R5U231 / 
R5U241) [Memory Stick Host Controller] (rev 01)

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 22/04/2017 à 15:45, Brian a écrit :

On Sat 22 Apr 2017 at 15:35:33 +0200, Pascal Hambourg wrote:


0b:00.1 SD Host controller: Ricoh Co Ltd R5C822 SD/SDIO/MMC/MS/MSPro Host
Adapter (rev 13)
Kernel driver in use: sdhci-pci

(taken on my own ThinkPad, but what the heck, it does not have any SD card
slot !)


Any better?

  brian@laptop:~$ lspci | grep Texas
  00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller
  00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 Host 
Controller
  00:10.3 Mass storage controller: Texas Instruments PCIxx21 Integrated 
FlashMedia Controller
  00:10.4 SD Host controller: Texas Instruments 
PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller


These 4 devices appear to be "functions" inside the same TI chip.

Looking again at my laptop, which has a PC-Card/CardBus slot :

0b:00.0 CardBus bridge: Ricoh Co Ltd RL5c476 II (rev 8d)
0b:00.1 SD Host controller: Ricoh Co Ltd R5C822 SD/SDIO/MMC/MS/MSPro 
Host Adapter (rev 13)


The SD card adapter is also in the same Ricoh chip as the CardBus 
adapter. It seems to be quite common.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 22 Apr 2017 at 15:35:33 +0200, Pascal Hambourg wrote:

> Le 22/04/2017 à 14:50, Brian a écrit :
> >On Wed 19 Apr 2017 at 21:10:36 -0500, David Wright wrote:
> >>
> >>My guess from this is that the former SD card is connected to the
> >>USB bus whereas the latter is connected to a different bus. The
> >>likely candidate from what I've read is the PCIe bus.
> >
> >Good guess.
> >
> > brian@laptop:~$ lspci | grep CardBus
> > 00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller
> 
> Err, what is the relationship between a CardBus (formerly PCMCIA) adapter
> and a SD card reader ?
> 
> Aren't you confusing with something like this :

More than likely. Thanks.

> 0b:00.1 SD Host controller: Ricoh Co Ltd R5C822 SD/SDIO/MMC/MS/MSPro Host
> Adapter (rev 13)
>   Kernel driver in use: sdhci-pci
> 
> (taken on my own ThinkPad, but what the heck, it does not have any SD card
> slot !)

Any better?

  brian@laptop:~$ lspci | grep Texas
  00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller
  00:10.2 FireWire (IEEE 1394): Texas Instruments OHCI Compliant IEEE 1394 Host 
Controller
  00:10.3 Mass storage controller: Texas Instruments PCIxx21 Integrated 
FlashMedia Controller
  00:10.4 SD Host controller: Texas Instruments 
PCI6411/6421/6611/6621/7411/7421/7611/7621 Secure Digital Controller

-- 
Brian.
 

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 22/04/2017 à 00:11, David Wright a écrit :

On Wed 19 Apr 2017 at 20:25:46 (+0200), Pascal Hambourg wrote:


The boot order in the BIOS has no effect on the device discovery and
naming in the Linux system. Also, most of the times USB devices lose
the discovery race against ATA devices due to extra delay. So hda
will probably always be the hard disk regardless of the boot device.


I've seen just one instance reported where an externally connected
USB drive got /dev/sda whereas the internal mSATA got /dev/sdb.


Yes, I just got a report of a case where the internal SATA drives was 
sdb and sdc, and sda was probably the internal all-in-one card reader. 
But I never observed such a case myself.



This was from someone who was about to dd copy the drive (which is
why they checked).


Hehe, we all know that sd* names are not reliable.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 22/04/2017 à 14:50, Brian a écrit :

On Wed 19 Apr 2017 at 21:10:36 -0500, David Wright wrote:


My guess from this is that the former SD card is connected to the
USB bus whereas the latter is connected to a different bus. The
likely candidate from what I've read is the PCIe bus.


Good guess.

 brian@laptop:~$ lspci | grep CardBus
 00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller


Err, what is the relationship between a CardBus (formerly PCMCIA) 
adapter and a SD card reader ?


Aren't you confusing with something like this :

0b:00.1 SD Host controller: Ricoh Co Ltd R5C822 SD/SDIO/MMC/MS/MSPro 
Host Adapter (rev 13)

Kernel driver in use: sdhci-pci

(taken on my own ThinkPad, but what the heck, it does not have any SD 
card slot !)

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 19 Apr 2017 at 21:10:36 -0500, David Wright wrote:

> On Sat 15 Apr 2017 at 23:21:33 (+0100), Brian wrote:
> > 
> > Nobody in this thread, apart from the OP, has a Lenovo and an SD card.
> > He is in the best position to test and report on this suggestion. Less
> > than twenty minutes work.
> 
> Frustratingly, I can't test this hypothesis, but I'm getting the
> impression that whether a PC can boot¹ directly from an SD card in
> the PC's SD slot depends² on how the card eventually appears on
> the bus.

I'll do the testing for you. :)

I have an oldish laptop (Turion based) with a full-sized SD slot. An SD
adapter with a micro SD card in it does not appear as

> $ ls -l /dev/disk/by-id/ [edited]
> usb-General_Combo_CardMMC_SD_12345678-0:0 -> ../../sdb
> usb-General_Combo_CardMMC_SD_12345678-0:0-part1 -> ../../sdb1
> usb-General_Combo_CardMMC_SD_12345678-0:0-part2 -> ../../sdb2
> $ cat /sys/bus/usb/devices/1-4:1.0/uevent
> DEVTYPE=usb_interface
> DRIVER=usb-storage
> PRODUCT=1025/5d/16b
> TYPE=0/0/0
> INTERFACE=8/6/80
> MODALIAS=usb:v1025p005Dd016Bdc00dsc00dp00ic08isc06ip50in00
> $ 

but does appear as

> $ ls -l /dev/disk/by-id/
> mmc-SE32G_0x0109e2ec -> ../../mmcblk0
> mmc-SE32G_0x0109e2ec-part1 -> ../../mmcblk0p1
> $ cat /sys/bus/mmc/devices/mmc0\:/uevent
> DRIVER=mmcblk
> MMC_TYPE=SD
> MMC_NAME=SE32G
> MODALIAS=mmc:block
> $ 
> 
> My guess from this is that the former SD card is connected to the
> USB bus whereas the latter is connected to a different bus. The
> likely candidate from what I've read is the PCIe bus.

Good guess.

 brian@laptop:~$ lspci | grep CardBus
 00:10.0 CardBus bridge: Texas Instruments PCIxx21/x515 Cardbus Controller

Also (without the card inserted):

 brian@laptop:~$ lsmod | grep mmc
 mmc_core   91803  3 sdhci,tifm_sd,sdhci_pci

With the card inserted:

 brian@laptop:~$ lsmod | grep mmc
 mmc_block  30466  2 
 mmc_core   91803  4 mmc_block,sdhci,tifm_sd,sdhci_pci

tifm_sd is the Texas Instruments FlashMedia SD driver.
 
> My next guess is that the USB bus (sorry about the tautology) can
> be run by the OS-less PC (obviously), whereas the other bus can't.
> The trade-off is that a PCIe bus would give a much greater speed.

This laptop only offers booting from CD and HD. GRUB's nativedisk module
will get it booting from USB but, because of the bus the SD controller
is on and GRUB having no modules for mmc, it will not boot from the SD
card.

There are quite a few complaints from users about machines not booting
from SD cards in internal readers. My guess is that, when it is possible,
the reader is always on the USB bus. 

My second guess is that aren't any machines which boot from a reader on
the PCI bus. Maybe it is too complex to implement; maybe it will come in
time. It took a fair amount of time after the introduction of USB for it
to become a common, reliable booting process.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 19 Apr 2017 at 15:16:52 -0500, Richard Owlett wrote:

> On 04/19/2017 02:31 PM, Brian wrote:
> >
> >As I have detailed elsewhere,
> >
> >https://lists.debian.org/debian-user/2017/04/msg00578.html
> >
> >the card becomes visible to GRUB with its nativedisk command. The OP has
> >a reader which does not show an SD disk in GRUB and does not boot from
> >it. Perhaps he could take some time to see whether this command does
> >anything for him.
> 
> I just received my USB SD card reader this afternoon. Attempting to
> reproduce the results of your ...msg00578 is on my TODO list. I'll not have
> time for Debian until Monday. I did have the opportunity to do a quickie

Thanks for considering testing. However, it is likely not to work for
you because your 5-in-1 appparently is a Ricoh reader on the PCI bus.
You can check this with 'lspci'. Failure is also indicated by the
presence of 'root=/dev/mmcblk0p1' in your grub.cfg. AFAICT, GRUB does
not have modules for SD/MMC devices, so nativedisk would have nothing
to load to drive the device.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Wed 19 Apr 2017 at 20:25:46 (+0200), Pascal Hambourg wrote:
> Le 19/04/2017 à 09:46, GiaThnYgeia a écrit :
> >
> >Question:  You have a system with everything on a HD /dev/sda, you use
> >/dev/sdb a live disk to boot and install debian on /dev/sdc including
> >the grub on /dev/sdc
> >You unplug the live disk and reboot, what will grub do by default, what
> >systems will it show, by default?
> 
> It depends which GRUB boots.
> 
> If it's the GRUB on the hard disk that was /dev/sda during the
> installation, it will show the same systems as before the
> installation, i.e. it won't show the new system until you run
> update-grub to update its menu.
> 
> If it's the GRUB on the device that was sdc during the installation,
> it will show all present systems that os-prober can detect.
> 
> >I did not specify whether there is grub installed on /dev/sda or not, so
> >there are two variations.
> 
> Indeed.
> 
> >I suspect based on Bios settings the system
> >will either go to hd or to usb to boot,
> 
> You did not mention previously that /dev/sdc was a USB device.
> 
> >and that will be sda
> 
> The boot order in the BIOS has no effect on the device discovery and
> naming in the Linux system. Also, most of the times USB devices lose
> the discovery race against ATA devices due to extra delay. So hda
> will probably always be the hard disk regardless of the boot device.

I've seen just one instance reported where an externally connected
USB drive got /dev/sda whereas the internal mSATA got /dev/sdb.
This was from someone who was about to dd copy the drive (which is
why they checked).

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Thu 20 Apr 2017 at 07:51:47 (+0200), Pascal Hambourg wrote:
> Le 20/04/2017 à 04:10, David Wright a écrit :
> >
> >AFAICT cards that can boot appear later as:
> >
> >$ ls -l /dev/disk/by-id/ [edited]
> >usb-General_Combo_CardMMC_SD_12345678-0:0 -> ../../sdb
> >usb-General_Combo_CardMMC_SD_12345678-0:0-part1 -> ../../sdb1
> >usb-General_Combo_CardMMC_SD_12345678-0:0-part2 -> ../../sdb2
> >$ cat /sys/bus/usb/devices/1-4:1.0/uevent
> >DEVTYPE=usb_interface
> >DRIVER=usb-storage
> >PRODUCT=1025/5d/16b
> >TYPE=0/0/0
> >INTERFACE=8/6/80
> >MODALIAS=usb:v1025p005Dd016Bdc00dsc00dp00ic08isc06ip50in00
> 
> I had the same impression, but it is not always true according to
> Brian's counterexample.

What Brian was demonstrating (not every card that appears as USB can
be seen in the CMOS and boot) is not what interests me. The example
I would be interested in seeing is the opposite, an SD card that
appears as mmcblk to linux but is also visible in the CMOS screens
and is therefore likely bootable. If so, how is it described in
the CMOS—you rarely see this sort of stuff well documented.

IOW how is a card like this one described in the CMOS?

$ ls -l /dev/disk/by-id/
mmc-SE32G_0x0109e2ec -> ../../mmcblk0
mmc-SE32G_0x0109e2ec-part1 -> ../../mmcblk0p1
$ cat /sys/bus/mmc/devices/mmc0\:/uevent
DRIVER=mmcblk
MMC_TYPE=SD
MMC_NAME=SE32G
MODALIAS=mmc:block
$

(To give myself the best chance of answering this, I now leave an
SD card in my laptop in case I ever manage to get into the CMOS again.)

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 20/04/2017 à 04:10, David Wright a écrit :


AFAICT cards that can boot appear later as:

$ ls -l /dev/disk/by-id/ [edited]
usb-General_Combo_CardMMC_SD_12345678-0:0 -> ../../sdb
usb-General_Combo_CardMMC_SD_12345678-0:0-part1 -> ../../sdb1
usb-General_Combo_CardMMC_SD_12345678-0:0-part2 -> ../../sdb2
$ cat /sys/bus/usb/devices/1-4:1.0/uevent
DEVTYPE=usb_interface
DRIVER=usb-storage
PRODUCT=1025/5d/16b
TYPE=0/0/0
INTERFACE=8/6/80
MODALIAS=usb:v1025p005Dd016Bdc00dsc00dp00ic08isc06ip50in00


I had the same impression, but it is not always true according to 
Brian's counterexample.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 23:21:33 (+0100), Brian wrote:
> On Sat 15 Apr 2017 at 16:30:29 -0500, David Wright wrote:
> 
> > On Sat 15 Apr 2017 at 19:57:32 (+0100), Brian wrote:
> > > On Sat 15 Apr 2017 at 13:08:30 -0500, Richard Owlett wrote:
> > > 
> > > > On 04/15/2017 12:24 PM, Brian wrote:
> > > > >
> > > > >Did you overlook this question? You have said your machine does not
> > > > >offer booting from an SD card. Your answer will be interesting.
> > > > 
> > > > Thought I'd answered it elsewhere.
> > > 
> > > Nope. There has been no mention of booting *directly* *from* the SD card
> > > until this subthread.

I think this was answered in 632231ab-5b9f-f22d-0c92-020728023...@cloud85.net
which is https://lists.debian.org/debian-user/2017/04/msg00468.html
in the other subthread.

> > > > It's on the menu that exists due to the grub on MBR of /dev/sda .
> > > 
> > > So - the card is in its slot on your machine. You do 'update-grub'.
> > > There is now an entry in GRUB's menu. That is fine. This is what you are
> > > booting from? Your grub.cfg looks similar to what you posted before?
> > > 
> > > If GRUB has been installed to the MBR of the SD card it has absolutely
> > > no bearing on the existence of the entry in GRUB's menu. It may as well
> > > not be there when GRUB on the MBR of /dev/sda constructs its grub.cfg.
> > > 
> > > > There is an fschk error of some sort that flies by too fast.
> > > > Otherwise, runs from SD card.
> > > 
> > > I like "simple"; I'm lost.
> > 
> > One can avoid all this messing about with Grub by just copying a
> > netinst ISO onto the SD card instead of a USB stick.
> > 
> > But then you need, as I've pointed out just now, to insert the
> > SD card into the slot _before_ booting, _and_ entering the CMOS
> > Setup Menu to make sure the device has highest booting priority.
> > (This is irrespective of how you wrote the SD card.)
> 
> Nobody in this thread, apart from the OP, has a Lenovo and an SD card.
> He is in the best position to test and report on this suggestion. Less
> than twenty minutes work.

Frustratingly, I can't test this hypothesis, but I'm getting the
impression that whether a PC can boot¹ directly from an SD card in
the PC's SD slot depends² on how the card eventually appears on
the bus.

AFAICT cards that can boot appear later as:

$ ls -l /dev/disk/by-id/ [edited]
usb-General_Combo_CardMMC_SD_12345678-0:0 -> ../../sdb
usb-General_Combo_CardMMC_SD_12345678-0:0-part1 -> ../../sdb1
usb-General_Combo_CardMMC_SD_12345678-0:0-part2 -> ../../sdb2
$ cat /sys/bus/usb/devices/1-4:1.0/uevent
DEVTYPE=usb_interface
DRIVER=usb-storage
PRODUCT=1025/5d/16b
TYPE=0/0/0
INTERFACE=8/6/80
MODALIAS=usb:v1025p005Dd016Bdc00dsc00dp00ic08isc06ip50in00
$ 

whereas those that can't appear as:

$ ls -l /dev/disk/by-id/
mmc-SE32G_0x0109e2ec -> ../../mmcblk0
mmc-SE32G_0x0109e2ec-part1 -> ../../mmcblk0p1
$ cat /sys/bus/mmc/devices/mmc0\:/uevent
DRIVER=mmcblk
MMC_TYPE=SD
MMC_NAME=SE32G
MODALIAS=mmc:block
$ 

My guess from this is that the former SD card is connected to the
USB bus whereas the latter is connected to a different bus. The
likely candidate from what I've read is the PCIe bus.

My next guess is that the USB bus (sorry about the tautology) can
be run by the OS-less PC (obviously), whereas the other bus can't.
The trade-off is that a PCIe bus would give a much greater speed.

¹ By boot, I mean read the boot sector from the MBR or PBR of a device.
² vice versa really.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/19/2017 02:31 PM, Brian wrote:

On Wed 19 Apr 2017 at 20:39:35 +0200, Pascal Hambourg wrote:


Le 19/04/2017 à 00:27, Brian a écrit :

On Wed 19 Apr 2017 at 00:12:15 +0200, Pascal Hambourg wrote:


Le 18/04/2017 à 21:14, Brian a écrit :


However, my main point was that the kernel
identifying a card as a /dev/sdX disk does not imply it will necessarily
be visible to GRUB and bootable by it.


If the card reader is idendified as /dev/sdX, then it probably behaves as a
generic USB mass storage device and most BIOS can see it.


A card in one of the slots of my card reader is identified as /dev/sdc1.
The reader itself is seen by lsusb. The BIOS doesn't see it and does not
have an option to boot from it. So I suppose my reader is not "most
BIOS". What is it behaving as?


How does lsusb identify the card reader ? What driver/module does it use ?
"lsusb -t" will tell both.


root@test:~# lsusb
Bus 006 Device 002: ID 0644:0200 TEAC Corp. All-In-One Multi-Card Reader 
CA200/B/S

root@test:~# lsusb -t
/:  Bus 07.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 06.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/6p, 480M
|__ Port 2: Dev 2, If 0, Class=Mass Storage, Driver=usb-storage, 480M
/:  Bus 05.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/4p, 480M
/:  Bus 04.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 03.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 02.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
|__ Port 1: Dev 2, If 0, Class=Hub, Driver=hub/3p, 12M
|__ Port 1: Dev 4, If 0, Class=Human Interface Device, Driver=usbhid, 
12M
|__ Port 1: Dev 4, If 1, Class=Human Interface Device, Driver=usbhid, 
12M
|__ Port 2: Dev 3, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M

All external USB devices were disconnected, apart from the keyboard.


If it uses usb-storage and the BIOS can boot from a USB device, then it
surprises me that the BIOS cannot see and boot it when it contains a card
with proper contents.


GRUB's 'ls' command shows

(hd0) (hd0,msdos5) (hd0,msdos1) (fd0)

when a CF card with Stretch on it is inserted before booting. (hd0) is a
SATA disk.

As I have detailed elsewhere,

https://lists.debian.org/debian-user/2017/04/msg00578.html

the card becomes visible to GRUB with its nativedisk command. The OP has
a reader which does not show an SD disk in GRUB and does not boot from
it. Perhaps he could take some time to see whether this command does
anything for him.


I just received my USB SD card reader this afternoon. Attempting to 
reproduce the results of your ...msg00578 is on my TODO list. I'll not 
have time for Debian until Monday. I did have the opportunity to do a 
quickie test. A few days ago I had re-installed Debian to the SD card 
along with installing GRUB2 to the MBR of the SD card. The first tings I 
did were:

  power off the laptop
  remove the SD card from the reader which BIOS does not see
  inserted into the new USB SD card reader
  powered on the laptop
  at the BIOS prompt selected to boot from an alternate device
  at the appropriate sub-menu chose the USB SD card reader (appeared
  as /dev/sdb1)
  booted with no apparent problem or error messages

Will run your tests Monday.




His feedback (and that of others) would be
appreciated, The opinion about the inability to boot from an internal
card reader appears in need of revision.

(As an aside; I *can* boot the CF card from the reader in my printer!).

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 19 Apr 2017 at 20:39:35 +0200, Pascal Hambourg wrote:

> Le 19/04/2017 à 00:27, Brian a écrit :
> >On Wed 19 Apr 2017 at 00:12:15 +0200, Pascal Hambourg wrote:
> >
> >>Le 18/04/2017 à 21:14, Brian a écrit :
> >>>
> >>>However, my main point was that the kernel
> >>>identifying a card as a /dev/sdX disk does not imply it will necessarily
> >>>be visible to GRUB and bootable by it.
> >>
> >>If the card reader is idendified as /dev/sdX, then it probably behaves as a
> >>generic USB mass storage device and most BIOS can see it.
> >
> >A card in one of the slots of my card reader is identified as /dev/sdc1.
> >The reader itself is seen by lsusb. The BIOS doesn't see it and does not
> >have an option to boot from it. So I suppose my reader is not "most
> >BIOS". What is it behaving as?
> 
> How does lsusb identify the card reader ? What driver/module does it use ?
> "lsusb -t" will tell both.

root@test:~# lsusb
Bus 006 Device 002: ID 0644:0200 TEAC Corp. All-In-One Multi-Card Reader 
CA200/B/S

root@test:~# lsusb -t
/:  Bus 07.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 06.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/6p, 480M
|__ Port 2: Dev 2, If 0, Class=Mass Storage, Driver=usb-storage, 480M
/:  Bus 05.Port 1: Dev 1, Class=root_hub, Driver=ehci-pci/4p, 480M
/:  Bus 04.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 03.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 02.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
/:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=uhci_hcd/2p, 12M
|__ Port 1: Dev 2, If 0, Class=Hub, Driver=hub/3p, 12M
|__ Port 1: Dev 4, If 0, Class=Human Interface Device, Driver=usbhid, 
12M
|__ Port 1: Dev 4, If 1, Class=Human Interface Device, Driver=usbhid, 
12M
|__ Port 2: Dev 3, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M

All external USB devices were disconnected, apart from the keyboard.

> If it uses usb-storage and the BIOS can boot from a USB device, then it
> surprises me that the BIOS cannot see and boot it when it contains a card
> with proper contents.

GRUB's 'ls' command shows

(hd0) (hd0,msdos5) (hd0,msdos1) (fd0)

when a CF card with Stretch on it is inserted before booting. (hd0) is a
SATA disk.

As I have detailed elsewhere,

https://lists.debian.org/debian-user/2017/04/msg00578.html

the card becomes visible to GRUB with its nativedisk command. The OP has
a reader which does not show an SD disk in GRUB and does not boot from
it. Perhaps he could take some time to see whether this command does
anything for him. His feedback (and that of others) would be
appreciated, The opinion about the inability to boot from an internal
card reader appears in need of revision.

(As an aside; I *can* boot the CF card from the reader in my printer!). 

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 19/04/2017 à 00:27, Brian a écrit :

On Wed 19 Apr 2017 at 00:12:15 +0200, Pascal Hambourg wrote:


Le 18/04/2017 à 21:14, Brian a écrit :


However, my main point was that the kernel
identifying a card as a /dev/sdX disk does not imply it will necessarily
be visible to GRUB and bootable by it.


If the card reader is idendified as /dev/sdX, then it probably behaves as a
generic USB mass storage device and most BIOS can see it.


A card in one of the slots of my card reader is identified as /dev/sdc1.
The reader itself is seen by lsusb. The BIOS doesn't see it and does not
have an option to boot from it. So I suppose my reader is not "most
BIOS". What is it behaving as?


How does lsusb identify the card reader ? What driver/module does it use 
? "lsusb -t" will tell both.


If it uses usb-storage and the BIOS can boot from a USB device, then it 
surprises me that the BIOS cannot see and boot it when it contains a 
card with proper contents.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 19/04/2017 à 09:46, GiaThnYgeia a écrit :


Question:  You have a system with everything on a HD /dev/sda, you use
/dev/sdb a live disk to boot and install debian on /dev/sdc including
the grub on /dev/sdc
You unplug the live disk and reboot, what will grub do by default, what
systems will it show, by default?


It depends which GRUB boots.

If it's the GRUB on the hard disk that was /dev/sda during the 
installation, it will show the same systems as before the installation, 
i.e. it won't show the new system until you run update-grub to update 
its menu.


If it's the GRUB on the device that was sdc during the installation, it 
will show all present systems that os-prober can detect.



I did not specify whether there is grub installed on /dev/sda or not, so
there are two variations.


Indeed.


I suspect based on Bios settings the system
will either go to hd or to usb to boot,


You did not mention previously that /dev/sdc was a USB device.


and that will be sda


The boot order in the BIOS has no effect on the device discovery and 
naming in the Linux system. Also, most of the times USB devices lose the 
discovery race against ATA devices due to extra delay. So hda will 
probably always be the hard disk regardless of the boot device.

Re: Possibly erroneous "device not present" message during boot

[GiaThnYgeia]

The Wanderer:
> On 2017-04-18 at 15:42, Brian wrote:
> 
>> On Tue 18 Apr 2017 at 20:55:43 +0200, Pascal Hambourg wrote:
>>
>>> Le 17/04/2017 à 22:46, GiaThnYgeia a écrit :

 I hope this makes some sense to someone.
>>>
>>> Not to me. TL;DR.

Question:  You have a system with everything on a HD /dev/sda, you use
/dev/sdb a live disk to boot and install debian on /dev/sdc including
the grub on /dev/sdc
You unplug the live disk and reboot, what will grub do by default, what
systems will it show, by default?
I did not specify whether there is grub installed on /dev/sda or not, so
there are two variations.  I suspect based on Bios settings the system
will either go to hd or to usb to boot, and that will be sda



-- 
 "The most violent element in society is ignorance" rEG

"Who died and made you the superuser?"  Brooklinux

"keep rocking in the non-free world" Neilznotyoung

Re: Possibly erroneous "device not present" message during boot

[The Wanderer]

On 2017-04-18 at 15:42, Brian wrote:

> On Tue 18 Apr 2017 at 20:55:43 +0200, Pascal Hambourg wrote:
> 
>> Le 17/04/2017 à 22:46, GiaThnYgeia a écrit :
>>> 
>>> I hope this makes some sense to someone.
>> 
>> Not to me. TL;DR.
> 
> TL is "too loquacious"? DR is "drivel"?

In case you're not being facetious: "TL;DR" is standard Internet
shorthand for "too long; didn't read".

The reaction which it represents is a peculiar combination of
"frequently understandable" with "usually kind of rude".

-- 
   The Wanderer

The reasonable man adapts himself to the world; the unreasonable one
persists in trying to adapt the world to himself. Therefore all
progress depends on the unreasonable man. -- George Bernard Shaw



signature.asc
Description: OpenPGP digital signature

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 19 Apr 2017 at 00:12:15 +0200, Pascal Hambourg wrote:

> Le 18/04/2017 à 21:14, Brian a écrit :
> >
> >However, my main point was that the kernel
> >identifying a card as a /dev/sdX disk does not imply it will necessarily
> >be visible to GRUB and bootable by it.
> 
> If the card reader is idendified as /dev/sdX, then it probably behaves as a
> generic USB mass storage device and most BIOS can see it.

A card in one of the slots of my card reader is identified as /dev/sdc1.
The reader itself is seen by lsusb. The BIOS doesn't see it and does not
have an option to boot from it. So I suppose my reader is not "most
BIOS". What is it behaving as?

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 18/04/2017 à 21:14, Brian a écrit :


However, my main point was that the kernel
identifying a card as a /dev/sdX disk does not imply it will necessarily
be visible to GRUB and bootable by it.


If the card reader is idendified as /dev/sdX, then it probably behaves 
as a generic USB mass storage device and most BIOS can see it.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Tue 18 Apr 2017 at 20:55:43 +0200, Pascal Hambourg wrote:

> Le 17/04/2017 à 22:46, GiaThnYgeia a écrit :
> >
> >I hope this makes some sense to someone.
> 
> Not to me. TL;DR.

TL is "too loquacious"? DR is "drivel"?

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Tue 18 Apr 2017 at 20:42:40 +0200, Pascal Hambourg wrote:

> Le 17/04/2017 à 21:04, Brian a écrit :
> >On Sat 15 Apr 2017 at 00:07:39 +0200, Pascal Hambourg wrote:
> >
> >>No. The only bug is that the system installed on the SD card boots while it
> >>should not.
> >
> >Why shouldn't the system boot? It has the same kernel that is on the
> >hard disk
> 
> By chance. In a more general case, it would not boot.

Agreed. I've said as much elsewhere. Indeed, the OP never uses anything
but Jessie so luck was involved - there is no kernel/initrd mismatch.

What is intriguing is that the search error message was not observed at
the store.

> If the distribution installed on the SD card had a different kernel, or
> located in a different path, or if the initramfs on the internal drive did
> not include generic drivers but only the ones needed to mount the root
> filesystem from the internal drive (MODULES=dep instead of MODULES=most in
> /etc/initramfs-tools/initramfs.conf), it would not boot.

True. I've experienced all of this.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Tue 18 Apr 2017 at 20:32:38 +0200, Pascal Hambourg wrote:

> Le 17/04/2017 à 21:05, Brian a écrit :
> >On Mon 17 Apr 2017 at 08:48:50 +, Curt wrote:
> >
> >>On 2017-04-17, David Wright  wrote:
> >>>
> >>>Oddly, this ancient laptop (Acer) has a slot that makes the SD card
> >>>look like a /dev/sdX disk, and the BIOS is happy to boot from it.
> >>
> >>I have an Acer X1430 with an internal "Multi-in-1 Media Card Reader" for
> >>which an SD card inserted into the reader is recognized by the kernel as
> >>as a /dev/sdX disk.
> >
> >You have to be careful here to distinguish between what the OS can
> >recognise and what GRUB can see. The kernel will be using the Multimedia
> >Card device drivers to detect the device
> 
> Not always. If the card reader identifies itself as a generic USB mass
> storage device, then the kernel will use the same drivers as for USB drives
> and assign the same kind of name /dev/sd*. It seems that in this case, most
> BIOS recognize it and can boot from it.
> 
> I guess that the kernel will use MMC device drivers only if the card reader
> identifies itself as a native SD/MMC card reader.

Seems reasonable and sensible. However, my main point was that the kernel
identifying a card as a /dev/sdX disk does not imply it will necessarily
be visible to GRUB and bootable by it. 

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 17/04/2017 à 22:46, GiaThnYgeia a écrit :


I hope this makes some sense to someone.


Not to me. TL;DR.


Does a usb memory stick really have an MBR?


Yes, if you write an MBR structure into the first sector.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 18/04/2017 à 00:04, Richard Owlett a écrit :


My *minimal* criterion was the ability to read *AND* write the device.


Well then, installing a Debian system on it and booting it was overkill.
You could have just performed some reads and writes on the device from 
any of the systems installed on the internal drive.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 17/04/2017 à 21:04, Brian a écrit :

On Sat 15 Apr 2017 at 00:07:39 +0200, Pascal Hambourg wrote:


No. The only bug is that the system installed on the SD card boots while it
should not.


Why shouldn't the system boot? It has the same kernel that is on the
hard disk


By chance. In a more general case, it would not boot.

If the distribution installed on the SD card had a different kernel, or 
located in a different path, or if the initramfs on the internal drive 
did not include generic drivers but only the ones needed to mount the 
root filesystem from the internal drive (MODULES=dep instead of 
MODULES=most in /etc/initramfs-tools/initramfs.conf), it would not boot.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 17/04/2017 à 21:05, Brian a écrit :

On Mon 17 Apr 2017 at 08:48:50 +, Curt wrote:


On 2017-04-17, David Wright  wrote:


Oddly, this ancient laptop (Acer) has a slot that makes the SD card
look like a /dev/sdX disk, and the BIOS is happy to boot from it.


I have an Acer X1430 with an internal "Multi-in-1 Media Card Reader" for
which an SD card inserted into the reader is recognized by the kernel as
as a /dev/sdX disk.


You have to be careful here to distinguish between what the OS can
recognise and what GRUB can see. The kernel will be using the Multimedia
Card device drivers to detect the device


Not always. If the card reader identifies itself as a generic USB mass 
storage device, then the kernel will use the same drivers as for USB 
drives and assign the same kind of name /dev/sd*. It seems that in this 
case, most BIOS recognize it and can boot from it.


I guess that the kernel will use MMC device drivers only if the card 
reader identifies itself as a native SD/MMC card reader.

Re: Possibly erroneous "device not present" message during boot

[Lisi Reisz]

On Monday 17 April 2017 23:04:57 Richard Owlett wrote:
> Please note my usage of "criterion" rather than "criteria".

Sorry Richard.  I would certainly have noted if you hadn't.  It would have 
screamed at me.  I'm afraid that I don't always notice in passing when words 
are used correctly!  Nor do I see any need to note, as requested, that you 
know the difference between one and more of something. ;-)

Lisi

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Mon 17 Apr 2017 at 17:04:57 -0500, Richard Owlett wrote:

> On 04/17/2017 02:04 PM, Brian wrote:
> >On Sat 15 Apr 2017 at 00:07:39 +0200, Pascal Hambourg wrote:
> >
> >>Le 14/04/2017 à 19:05, Richard Owlett a écrit :
> >>
> >>>Does any of this justify a bug report.
> >>
> >>No. The only bug is that the system installed on the SD card boots while it
> >>should not.
> >
> >Why shouldn't the system boot? It has the same kernel that is on the
> >hard disk, And the GRUB versions are the same. Does it matter where the
> >booted kernel resides?
> >
> >Ok. The OP was lucky. Another arrangement between the booted OS and GRUB
> >on the hard disk would possibly have led to a loss of sale for the store.
> >Never using anything but Jessie installs (which is what the OP does) is
> >a disadvantage when GRUB is involved.
> 
> Not quite correct.
> I performed what *I* considered a *WORST CASE TEST*.
> It passed.
> My *minimal* criterion was the ability to read *AND* write the device.

You haven't quite absorbed and understood

https://lists.debian.org/debian-user/2017/04/msg00430.html

I'd suggest you get to grip with its contents.

> CAVEAT LECTORS
>  Before retirement I frequently served in QA/QC roles
>NOTE BENE: QC is NOT identical to QA
>though "pickers of nits" frequent both
> 
> Please note my usage of "criterion" rather than "criteria".

I did. I also noted your use of "minimal" rather than "minimum". (I would
not normally point out things like this but times are hard. :) )

-- 
Brian

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/17/2017 04:11 PM, GiaThnYgeia wrote:



Richard Owlett:

 4. All installs in this thread have been done using DVD 1 of
13 of Debian 8.6.0 - thus all intrinsically use the same
kernel.


Is this with the actual dvd or the image of the 1st DVD on a usb-stick?


In this case a tautology.


I assume once everything was installed and rebooted there was an upgrade


*ERROR* - All is from DVD - I have MINIMAL available band width ;/


to the latest Jessie which I am pretty sure incorporated a new version
of grub-xxx and updates the grub.cfg based on the current state of that
boot.  During the installation it asks you to select and confirm where
you want grub installed, but when it is upgraded as a package it bases
its installation on previous options, except it reviews that the
installed systems on partitions are actually there.


Does any of this justify a bug report. Especially as I do not
have the bandwidth to do a netinstall of a pre-release version?


If your bug reporting is functional, go ahead it is FREE!  :)


You err.
A spurious bug report may cost me nothing.
*HOWEVER* it costs OTHERS!


I am saying this because it only works with the appropriate network/mail
setup that Debian dictates.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/17/2017 02:04 PM, Brian wrote:

On Sat 15 Apr 2017 at 00:07:39 +0200, Pascal Hambourg wrote:


Le 14/04/2017 à 19:05, Richard Owlett a écrit :


Does any of this justify a bug report.


No. The only bug is that the system installed on the SD card boots while it
should not.


Why shouldn't the system boot? It has the same kernel that is on the
hard disk, And the GRUB versions are the same. Does it matter where the
booted kernel resides?

Ok. The OP was lucky. Another arrangement between the booted OS and GRUB
on the hard disk would possibly have led to a loss of sale for the store.
Never using anything but Jessie installs (which is what the OP does) is
a disadvantage when GRUB is involved.



Not quite correct.
I performed what *I* considered a *WORST CASE TEST*.
It passed.
My *minimal* criterion was the ability to read *AND* write the device.

CAVEAT LECTORS
 Before retirement I frequently served in QA/QC roles
   NOTE BENE: QC is NOT identical to QA
   though "pickers of nits" frequent both

Please note my usage of "criterion" rather than "criteria".
Side note: I am an alumnus of "Venatores Verborum"

may dictator lapidum rip

Re: Possibly erroneous "device not present" message during boot

[GiaThnYgeia]

Richard Owlett:
>  4. All installs in this thread have been done using DVD 1 of
> 13 of Debian 8.6.0 - thus all intrinsically use the same
> kernel.

Is this with the actual dvd or the image of the 1st DVD on a usb-stick?
I assume once everything was installed and rebooted there was an upgrade
to the latest Jessie which I am pretty sure incorporated a new version
of grub-xxx and updates the grub.cfg based on the current state of that
boot.  During the installation it asks you to select and confirm where
you want grub installed, but when it is upgraded as a package it bases
its installation on previous options, except it reviews that the
installed systems on partitions are actually there.

> Does any of this justify a bug report. Especially as I do not
> have the bandwidth to do a netinstall of a pre-release version?

If your bug reporting is functional, go ahead it is FREE!  :)
I am saying this because it only works with the appropriate network/mail
setup that Debian dictates.

-- 
 "The most violent element in society is ignorance" rEG

"Who died and made you the superuser?"  Brooklinux

"keep rocking in the non-free world" Neilznotyoung

Re: Possibly erroneous "device not present" message during boot

[GiaThnYgeia]

Richard Owlett:
> On 04/12/2017 12:13 PM, Pascal Hambourg wrote:
>> Le 12/04/2017 à 17:14, Richard Owlett a écrit :
>>>
>>> Whether initiated after power-on OR a restart the observed sequence is:
>>> 1. Appearance of the Grub2 menu with a choice of 4 instances of Debian.
>>> 2. Select instance installed on the SD card.
>>> 3. Screen clears, this message appears against the Debian 8 background.
>>>error:  no such device: 380e2a6d-f851-4fd1-9db2-869a0982b511.
>>>Press any key to continue ...
>>> 4. Otherwise the instance of Debian on the SD card boots routinely.
>>
>> This is a GRUB error message which cannot find the UUID specified in a
>> "search" command.
>>
>> In order to investigate, can you
>> - report the menu entry code for the SD card system in
>> /boot/grub/grub.cfg (the one from the system on the hard drive owning
>> GRUB, not the one on the SD card) ;
> 
> On theory "too much better than too little" I see:
> 
> ### BEGIN /etc/grub.d/20_linux_xen ###
> 
> ### END /etc/grub.d/20_linux_xen ###

Wild shot of the blind helping the half blind here.

If the uuid is correct but if sda and sdb are reversed the wrong
partition is found on the crossed uuid.
I couldn't reproduce it myself but there had been a couple of instances
while messing with usb-mem-sticks and installing grub to the stick as if
it was a real disk, that this happened.  Booting from the bios command
to choose between hdd or usb seemed more consistent than when bios was
adjusted to automatically try to boot from usb if it existed before
trying the hdd.  I don't know whether your setup can relate.  To make
matters worse when the usb installed grub was crossed and the hdd became
sdb I updated grub-pc which goes through the whole procedure of seeking
installed systems on all drives and updating the usb-installed grub to
where now it would seek hd installed systems in sdb!  So now the correct
booting order would not work and the default primary installed debian
was not found on sda1 as it did not exist on the hd.

So, I am speculating such crossing may have taken place between store
and home that produced the mistake, sda and sdb were reversed.  Now if
you have 1 hdd and 2 usb drives/mem-sticks things can get really messy
as grub is silly enough to think this is a fixed order of things.  As
you unplug sdb and reboot sdc has now become sdb even though it was
never moved.  But how did you install something on the usbdisk?  Was it
from a live usb-stick?  If that stick is gone then your installed grub
thinks that the HD is the usb/live stick that installed debian.

I hope this makes some sense to someone.  Does a usb memory stick really
have an MBR?  It may act like it.




-- 
 "The most violent element in society is ignorance" rEG

"Who died and made you the superuser?"  Brooklinux

"keep rocking in the non-free world" Neilznotyoung

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Mon 17 Apr 2017 at 08:48:50 +, Curt wrote:

> On 2017-04-17, David Wright  wrote:
> >
> > Oddly, this ancient laptop (Acer) has a slot that makes the SD card
> > look like a /dev/sdX disk, and the BIOS is happy to boot from it.
> 
> I have an Acer X1430 with an internal "Multi-in-1 Media Card Reader" for
> which an SD card inserted into the reader is recognized by the kernel as
> as a /dev/sdX disk.

You have to be careful here to distinguish between what the OS can
recognise and what GRUB can see. The kernel will be using the Multimedia
Card device drivers to detect the device and assign a name to it.  Check
the modules are loaded with

 lsmod | grep mmc

On the other hand, GRUB can only boot from a device when information
is provided to it by the BIOS [1]. Using the normal GRUB setup [2] that
Debian provides, the command

 ls

at the GRUB command prompt (press "c" within the GRUB menu for this)
will show whether a device equivalent to the kernel's /dev/sdX is
discovered.

[1] Not completely correct, but it will do for the point I want to make.

[2] The "normal" setup does not load GRUB's nativedisk module.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 00:07:39 +0200, Pascal Hambourg wrote:

> Le 14/04/2017 à 19:05, Richard Owlett a écrit :
> 
> >Does any of this justify a bug report.
> 
> No. The only bug is that the system installed on the SD card boots while it
> should not.

Why shouldn't the system boot? It has the same kernel that is on the
hard disk, And the GRUB versions are the same. Does it matter where the
booted kernel resides?

Ok. The OP was lucky. Another arrangement between the booted OS and GRUB
on the hard disk would possibly have led to a loss of sale for the store.
Never using anything but Jessie installs (which is what the OP does) is
a disadvantage when GRUB is involved.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Curt]

On 2017-04-17, David Wright  wrote:
>
> Oddly, this ancient laptop (Acer) has a slot that makes the SD card
> look like a /dev/sdX disk, and the BIOS is happy to boot from it.

I have an Acer X1430 with an internal "Multi-in-1 Media Card Reader" for
which an SD card inserted into the reader is recognized by the kernel as
as a /dev/sdX disk.

I would be happy to trade with the OP if he would be willing to come out
of retirement.

-;)

> Cheers,
> David.
>
>


-- 
"It might be a vision--of a shell, of a wheelbarrow, of a fairy kingdom on the
far side of the hedge; or it might be the glory of speed; no one knew." --Mrs.
Ramsay, speculating on why her little daughter might be dashing about, in "To
the Lighthouse," by Virginia Woolf.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 23:52:30 (+0200), Pascal Hambourg wrote:
> Le 15/04/2017 à 22:50, David Wright a écrit :
> >On Sat 15 Apr 2017 at 19:14:24 (+0200), Pascal Hambourg wrote:
> >>Le 15/04/2017 à 16:28, David Wright a écrit :
> >>>On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:
> >>>
> A SD card reader such as the one the OP has just exposes the SD card
> as what it is, a SD/MMC card (/dev/mmcblk*).
> >>>
> >>>I assumed that the OP, writing about a laptop, had no card reader,
> >>>and was inserting the SD card directly into the computer.
> >>
> >>How do you name that slot on the laptop or desktop where you insert
> >>the SD card, if not an embedded card reader ?
> >
> >A slot, or an SD slot (as on this Dell), or a micro SD slot.
> 
> Behind the slot is a SD card reader. Lenovo names it a card reader.

Ok, ok. You're angling for congratulations for guessing what
Amazon and I call a card reader is what you prefer to call a
USB-to-SD card adapter.

Congratulations.

> >You seem to have had some difficulty with this part of my post:
> >
> >| It's not clear to me why an SD is being used in this way as the OP
> >| has at last revealed that the computer is unable to boot from an SD
> >| plugged in directly.
> 
> IIUC, at first the OP did not know that the computer was unable to
> boot from the internal SD card reader and discovered it during this
> thread.

_I_ didn't know whether the OP knew, suspected, or guessed one way
or the other when taking the card on approval. The later assertion
could be correct or incorrect: the experiment and observations
aren't reported. Other people in other places have reported
difficulties booting with /dev/mmcblk-style devices so I'm not
optimistic.

I can't experiment with my own hardware because my mmcblk-style
Dell laptop has a fault (which will probably lead to its demise)
meaning that the screen is only occasionally readable for about
one second at a time, and its VGA output is only turned on after
the kernel has loaded. So no CMOS screens, no Boot Menu, and no
Grub menu, as I don't have photographic recall.

> >(Of course, an SD card can be made to look like a
> >| USB stick just by sticking it in a card reader. Then it will boot.)
> 
> Obviously not with the OP's laptop internal card reader, which does
> not make the SD card look like a USB mass storage class device.

In the laptop's slot, it would appear not. If the methodology I
outlined elsewhere in the thread fails, then it would seem that
whatever is behind this slot needs an OS to read the card, whereas
the BIOS can be persuaded to read a USB stick without the need
for any OS.

Oddly, this ancient laptop (Acer) has a slot that makes the SD card
look like a /dev/sdX disk, and the BIOS is happy to boot from it.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sun 16 Apr 2017 at 15:31:56 +0100, Brian wrote:

> What GRUB makes of the OS device could be seen with
> 
>  grub-probe OS_device  (/dev/sdc1, say)

Correction:

 grub-probe -d  -t drive

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Thu 13 Apr 2017 at 20:05:22 +0200, Pascal Hambourg wrote:

> Ok then, here is my full theory. You're not going to like it.
> 
> We now know that the BIOS does not expose the SD card so GRUB cannot see it
> and the search command fails and prints the error message.
> 
> But then how does the system on the SD card boot ?
> 
> The purpose of the search command is to set the $root variable value with
> the device havind the searched UUID. As it fails, the variable remains
> unmodified and keeps its previous value. The menu entry does not set a
> default value for $root (which is rather unusual in my experience), so the
> current value comes from a previous assignement in grub.cfg. It's probably
> the root device of the system owning GRUB on the hard disk drive.

$root not being set bothered me. Could this (from grub-pc's changelog.gz)
be the reason?

 * util/grub.d/00_header.in: Only set root variable when GRUB_DRIVE
   could be identified by update-grub.

A way to identify how the OS sees the card is to use 'lsblk' before
and after inserting the card in the card reader.

What GRUB makes of the OS device could be seen with

 grub-probe OS_device  (/dev/sdc1, say)

'Twould be nice to know.

> (In GRUB's shell, you can print the $root value with the "set" command, and
> check that this partition contains a /boot/grub/vmlinuz-3.16.0-4-686-pae
> file directory with the command "ls /boot/grub".)
> 
> GRUB uses $root as a default device used in paths which do not specify a
> device, such as the ones in the linux and initrd commands of the menu entry.
> By chance, it seems that this location contains the same kernel as the one
> on the SD card, so GRUB can load and boot it from the hard disk drive. But
> GRUB passes the SD card partition as the root device root=/dev/mmcblk0p1 to
> the kernel command line, so the kernel comes from the disk and the root
> filesystem comes from the SD card. Unlike GRUB, the kernel does not rely on
> the BIOS to access the SD card.
> 
> David is right : you don't really boot from the SD card.
> GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem is on
> the SD card.

The ultimate outcome gives every sign of success. However, strange
(and very often unwelcome) things can happen when the kernels on
the HDD and the booting device are different. Upgrading the HDD OS
to Stretch while leaving the distribution on the card at Jessie
would very likely not lead to a good experience.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 16:30:29 -0500, David Wright wrote:

> On Sat 15 Apr 2017 at 19:57:32 (+0100), Brian wrote:
> > On Sat 15 Apr 2017 at 13:08:30 -0500, Richard Owlett wrote:
> > 
> > > On 04/15/2017 12:24 PM, Brian wrote:
> > > >
> > > >Did you overlook this question? You have said your machine does not
> > > >offer booting from an SD card. Your answer will be interesting.
> > > 
> > > Thought I'd answered it elsewhere.
> > 
> > Nope. There has been no mention of booting *directly* *from* the SD card
> > until this subthread.
> > 
> > > It's on the menu that exists due to the grub on MBR of /dev/sda .
> > 
> > So - the card is in its slot on your machine. You do 'update-grub'.
> > There is now an entry in GRUB's menu. That is fine. This is what you are
> > booting from? Your grub.cfg looks similar to what you posted before?
> > 
> > If GRUB has been installed to the MBR of the SD card it has absolutely
> > no bearing on the existence of the entry in GRUB's menu. It may as well
> > not be there when GRUB on the MBR of /dev/sda constructs its grub.cfg.
> > 
> > > There is an fschk error of some sort that flies by too fast.
> > > Otherwise, runs from SD card.
> > 
> > I like "simple"; I'm lost.
> 
> One can avoid all this messing about with Grub by just copying a
> netinst ISO onto the SD card instead of a USB stick.
> 
> But then you need, as I've pointed out just now, to insert the
> SD card into the slot _before_ booting, _and_ entering the CMOS
> Setup Menu to make sure the device has highest booting priority.
> (This is irrespective of how you wrote the SD card.)

Nobody in this thread, apart from the OP, has a Lenovo and an SD card.
He is in the best position to test and report on this suggestion. Less
than twenty minutes work.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 22:50, David Wright a écrit :

On Sat 15 Apr 2017 at 19:14:24 (+0200), Pascal Hambourg wrote:

Le 15/04/2017 à 16:28, David Wright a écrit :

On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:


A SD card reader such as the one the OP has just exposes the SD card
as what it is, a SD/MMC card (/dev/mmcblk*).


I assumed that the OP, writing about a laptop, had no card reader,
and was inserting the SD card directly into the computer.


How do you name that slot on the laptop or desktop where you insert
the SD card, if not an embedded card reader ?


A slot, or an SD slot (as on this Dell), or a micro SD slot.


Behind the slot is a SD card reader. Lenovo names it a card reader.


You seem to have had some difficulty with this part of my post:

| It's not clear to me why an SD is being used in this way as the OP
| has at last revealed that the computer is unable to boot from an SD
| plugged in directly.


IIUC, at first the OP did not know that the computer was unable to boot 
from the internal SD card reader and discovered it during this thread.



(Of course, an SD card can be made to look like a
| USB stick just by sticking it in a card reader. Then it will boot.)


Obviously not with the OP's laptop internal card reader, which does not 
make the SD card look like a USB mass storage class device.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 19:57:32 (+0100), Brian wrote:
> On Sat 15 Apr 2017 at 13:08:30 -0500, Richard Owlett wrote:
> 
> > On 04/15/2017 12:24 PM, Brian wrote:
> > >
> > >Did you overlook this question? You have said your machine does not
> > >offer booting from an SD card. Your answer will be interesting.
> > 
> > Thought I'd answered it elsewhere.
> 
> Nope. There has been no mention of booting *directly* *from* the SD card
> until this subthread.
> 
> > It's on the menu that exists due to the grub on MBR of /dev/sda .
> 
> So - the card is in its slot on your machine. You do 'update-grub'.
> There is now an entry in GRUB's menu. That is fine. This is what you are
> booting from? Your grub.cfg looks similar to what you posted before?
> 
> If GRUB has been installed to the MBR of the SD card it has absolutely
> no bearing on the existence of the entry in GRUB's menu. It may as well
> not be there when GRUB on the MBR of /dev/sda constructs its grub.cfg.
> 
> > There is an fschk error of some sort that flies by too fast.
> > Otherwise, runs from SD card.
> 
> I like "simple"; I'm lost.

One can avoid all this messing about with Grub by just copying a
netinst ISO onto the SD card instead of a USB stick.

But then you need, as I've pointed out just now, to insert the
SD card into the slot _before_ booting, _and_ entering the CMOS
Setup Menu to make sure the device has highest booting priority.
(This is irrespective of how you wrote the SD card.)

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 13:16:54 (-0500), Richard Owlett wrote:
> On 04/15/2017 12:14 PM, Pascal Hambourg wrote:
> >Le 15/04/2017 à 16:28, David Wright a écrit :
> >>On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:
> >>
> >>>A SD card reader such as the one the OP has just exposes the SD card
> >>>as what it is, a SD/MMC card (/dev/mmcblk*).
> >>
> >>I assumed that the OP, writing about a laptop, had no card reader,
> >>and was inserting the SD card directly into the computer.
> >
> >How do you name that slot on the laptop or desktop where you insert the
> >SD card, if not an embedded card reader ?
> >
> >>As it happens, if you do this with the ancient laptop I'm typing on,
> >>it has the functionality of a card reader built into it and you get
> >>a /dev/sd*, and that can be boot an SD card directly.
> >
> >So there seems to be at least two kinds of SD card reader : those which
> >expose themselves as a USB mass storage device and those which expose
> >themselves as a SD/MMC device.
> >
> >I have a couple of desktops with an embedded multi-card reader,
> >connected to an internal USB port on the motherboard. But I do not have
> >any SD card (no use), so I never checked to see what kind they are.
> >
> >
> 
> From https://www.cnet.com/products/lenovo-thinkpad-t510/specs/
> I apparently have
> 
> Card Reader
>   Type 5 in 1 card reader
>   Supported Flash  Memory Stick, Memory Stick PRO, MultiMediaCard,
> Memory SD Memory Card, SDHC Memory Card

If I were you, I would try inserting your SD card and _then_ booting
it up. It's quite usual for bootable devices to appear in the BIOS's
Setup Menu only if a device is actually present. (Don't even rely on
a Boot Menu like F12 sometimes gives: check the Setup Menu itself.)
If it's there, promote it to first device to boot, and then boot.
(Don't rely on the promotion to be "sticky". If you ever boot
without a card plugged in, it may be demoted whenever you next plug
one in.)

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 19:14:24 (+0200), Pascal Hambourg wrote:
> Le 15/04/2017 à 16:28, David Wright a écrit :
> >On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:
> >
> >>A SD card reader such as the one the OP has just exposes the SD card
> >>as what it is, a SD/MMC card (/dev/mmcblk*).
> >
> >I assumed that the OP, writing about a laptop, had no card reader,
> >and was inserting the SD card directly into the computer.
> 
> How do you name that slot on the laptop or desktop where you insert
> the SD card, if not an embedded card reader ?

A slot, or an SD slot (as on this Dell), or a micro SD slot.

You seem to have had some difficulty with this part of my post:

| It's not clear to me why an SD is being used in this way as the OP
| has at last revealed that the computer is unable to boot from an SD
| plugged in directly. (Of course, an SD card can be made to look like a
| USB stick just by sticking it in a card reader. Then it will boot.)

I would be happy to go through it and detail all the referents
if you can't judge them from the context.

> >As it happens, if you do this with the ancient laptop I'm typing on,
> >it has the functionality of a card reader built into it and you get
> >a /dev/sd*, and that can be boot an SD card directly.
> 
> So there seems to be at least two kinds of SD card reader : those
> which expose themselves as a USB mass storage device and those which
> expose themselves as a SD/MMC device.
> 
> I have a couple of desktops with an embedded multi-card reader,
> connected to an internal USB port on the motherboard. But I do not
> have any SD card (no use), so I never checked to see what kind they
> are.

I would imagine that most people wouldn't want to disassemble to check
this. The internal connections are likely subminiature anyway, so it
would be difficult to see how any card will appear to the user without
just inserting one.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 21:24, Doug a écrit :


It turns out that there are two kinds of cards that look the same, but
on a Dell laptop I have, one kind won't be recognized and the other
works. I don't remember
which is which, or what they are called. One is SD, the other something
else.


MMC, SDHC, SDXC ?

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 21:04:34 +0200, Pascal Hambourg wrote:

> Le 15/04/2017 à 19:42, Brian a écrit :
> >On Sat 15 Apr 2017 at 19:20:58 +0200, Pascal Hambourg wrote:
> >>
> >>Expert install gives more control, but is far from granting total control.
> >>The Debian installer still has many automated actions that you cannot
> >>control even in expert mode.
> >
> >I had a feeling while writing that this response might come. It's
> >correct. "much more" instead of "total"?
> 
> I would not say "much more", just "more" or even "a bit more".

Contrasting "simple install" with "expert install" and "preseeded
install" I'd agree and use "more" for "expert install". But that wasn't
the original comparison.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Doug]

On 04/15/2017 10:28 AM, David Wright wrote:

On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:

Le 15/04/2017 à 02:37, David Wright a écrit :

Of course, an SD card can be
made to look like a USB stick just by sticking it in a card reader.

I guess you mean "in a USB-to-SD card adapter", which translates a
SD card into a USB mass storage device (/dev/sd*).

Yes, you guess correctly. Writing 40 years ago, I would use those same
words to describe a washing-machine sized object for reading punched
cards. Nowadays, if you type "card reader" into google, you will be
proffered several more sophisticated ones than mine, together with
some different varieties of credit card reader.


A SD card reader such as the one the OP has just exposes the SD card
as what it is, a SD/MMC card (/dev/mmcblk*).

I assumed that the OP, writing about a laptop, had no card reader,
and was inserting the SD card directly into the computer.

As it happens, if you do this with the ancient laptop I'm typing on,
it has the functionality of a card reader built into it and you get
a /dev/sd*, and that can be boot an SD card directly.

Cheers,
David.


It turns out that there are two kinds of cards that look the same, but 
on a Dell laptop I have, one kind won't be recognized and the other 
works. I don't remember
which is which, or what they are called. One is SD, the other something 
else. Of course, I found out the hard way, but it really doesn't matter, 
since I don't need to

use the laptop for that.

--doug

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 19:42, Brian a écrit :

On Sat 15 Apr 2017 at 19:20:58 +0200, Pascal Hambourg wrote:


Expert install gives more control, but is far from granting total control.
The Debian installer still has many automated actions that you cannot
control even in expert mode.


I had a feeling while writing that this response might come. It's
correct. "much more" instead of "total"?


I would not say "much more", just "more" or even "a bit more".

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 13:08:30 -0500, Richard Owlett wrote:

> On 04/15/2017 12:24 PM, Brian wrote:
> >
> >Did you overlook this question? You have said your machine does not
> >offer booting from an SD card. Your answer will be interesting.
> 
> Thought I'd answered it elsewhere.

Nope. There has been no mention of booting *directly* *from* the SD card
until this subthread.

> It's on the menu that exists due to the grub on MBR of /dev/sda .

So - the card is in its slot on your machine. You do 'update-grub'.
There is now an entry in GRUB's menu. That is fine. This is what you are
booting from? Your grub.cfg looks similar to what you posted before?

If GRUB has been installed to the MBR of the SD card it has absolutely
no bearing on the existence of the entry in GRUB's menu. It may as well
not be there when GRUB on the MBR of /dev/sda constructs its grub.cfg.

> There is an fschk error of some sort that flies by too fast.
> Otherwise, runs from SD card.

I like "simple"; I'm lost.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[songbird]

Richard Owlett wrote:
...
> I don't know how many times I've *MIS*read that last sentence ;<
> Mentally I was correcting non-existent typos, thus totally garbling it.
> I just spent several hours wandering thru grub files and loosely related 
> documentation.
> I've ended up at
> www.gnu.org/software/grub/manual/grub.html#Multi_002dboot-manual-config
> which says "Currently autogenerating config files for multi-boot 
> environments depends on os-prober and has several shortcomings. ..."
> I also discovered there that placing a customized grub in its own 
> partition is not only possible, but recommended in some situations. I 
> had read somewhere that that option had expired with grub-legacy. That, 
> although taking much time to understand, will solve a *MESS* of grub 
> problems for me ;/

  i hope so.  :)

  for me the conceptual trouble started ages ago
when i thought that "update-grub" and os-prober would
go out and find all the other little grubs and get
them all in sync.

  my most recent round of grubismo dealing with the
USB stick educated me a little about that misconception
and helped me figure out the custom menu entry and
chain loading.

  as usual, i still have much to learn...


  songbird

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/15/2017 12:14 PM, Pascal Hambourg wrote:

Le 15/04/2017 à 16:28, David Wright a écrit :

On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:


A SD card reader such as the one the OP has just exposes the SD card
as what it is, a SD/MMC card (/dev/mmcblk*).


I assumed that the OP, writing about a laptop, had no card reader,
and was inserting the SD card directly into the computer.


How do you name that slot on the laptop or desktop where you insert the
SD card, if not an embedded card reader ?


As it happens, if you do this with the ancient laptop I'm typing on,
it has the functionality of a card reader built into it and you get
a /dev/sd*, and that can be boot an SD card directly.


So there seems to be at least two kinds of SD card reader : those which
expose themselves as a USB mass storage device and those which expose
themselves as a SD/MMC device.

I have a couple of desktops with an embedded multi-card reader,
connected to an internal USB port on the motherboard. But I do not have
any SD card (no use), so I never checked to see what kind they are.




From https://www.cnet.com/products/lenovo-thinkpad-t510/specs/
I apparently have

Card Reader
  Type 5 in 1 card reader
  Supported Flash  Memory Stick, Memory Stick PRO, MultiMediaCard,
Memory SD Memory Card, SDHC Memory Card

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/15/2017 12:24 PM, Brian wrote:

On Sat 15 Apr 2017 at 11:52:09 -0500, Richard Owlett wrote:


On 04/15/2017 05:24 AM, Brian wrote:


Now for a big "but". :)

Your previous 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' stanza had

linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1

This one has

linux /boot/vmlinuz-3.16.0-4-686-pae 
root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet

The UUID has changed (I think you said you had done this) but the kernel
now looks for init using the UUID rather than on /dev/mmcblk0p1. How did
you manage to get something significantly different?


During the first install of Debian to the SD card I allowed the installer to
create a single partition filling the whole card. The primary purpose of
that install was a brute force determination of whether or not my individual
machine could read/write an SD card. The machine was a replacement for a
different used machine purchased from the vendor. My accepting the SD card
was dependent on that test.

I had reinstalled Debian to a more reasonably sized partition. That would
explain a changed UUID. During the reinstall I experimented with installing
grub to the MBR of the SD card. No grub was installed the first time.


The UUID change is understandable. But hey - you've altered the ground
rules! Now, it seems, you are exploring booting from the card itself
rather than from GRUB on a hard disk.


I considered a safe test as the BIOS does not list it as a possible boot 
device. Among other goodies I've ordered an USB SD card reader. We'll 
know more in ~1 week.





And, more to the point, is there booting from the card without any error
messages?


Did you overlook this question? You have said your machine does not
offer booting from an SD card. Your answer will be interesting.



Thought I'd answered it elsewhere.
It's on the menu that exists due to the grub on MBR of /dev/sda .
There is an fschk error of some sort that flies by too fast.
Otherwise, runs from SD card.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 11:35:14 -0500, Richard Owlett wrote:

> On 04/14/2017 01:19 PM, Brian wrote:
> >[snip]
> >
> >> 2. I only install Grub the *first* time I do a Debian install.
> >>By poor design Grub puts the current install first on menu.
> >>When experimenting with configuration as I do, the least
> >>likely install to be functional is the latest.
> >>This requires me to run update-grub on the "good" install.
> >
> >That's ok. I tend to be more promiscuous; usually on a whim, like
> >wanting to put a particular entry at the top of GRUB's menu list.
> >
> >When you do 'update-grub' do you still get no "set root=" line for the
> >SD card in the grub.cfg?
> >[snip]
> 
> I don't know how many times I've *MIS*read that last sentence ;<
> Mentally I was correcting non-existent typos, thus totally garbling it.
> I just spent several hours wandering thru grub files and loosely related
> documentation.
> I've ended up at
> www.gnu.org/software/grub/manual/grub.html#Multi_002dboot-manual-config
> which says "Currently autogenerating config files for multi-boot
> environments depends on os-prober and has several shortcomings. ..."
> I also discovered there that placing a customized grub in its own partition
> is not only possible, but recommended in some situations. I had read
> somewhere that that option had expired with grub-legacy. That, although
> taking much time to understand, will solve a *MESS* of grub problems for me
> ;/

Treading GRUB's byways isn't for me today. Your only problem as far as I
am concerned is the one expressed in your first post. Even that is of no
great consequence and the visibilty of the message is easily dealt with
in a custom GRUB stanza. Adopt the pragmatic approach.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 19:20:58 +0200, Pascal Hambourg wrote:

> Le 15/04/2017 à 12:44, Brian a écrit :
> >
> >simple install=some control of d-i.
> >expert install=total control of d-i.
> 
> Expert install gives more control, but is far from granting total control.
> The Debian installer still has many automated actions that you cannot
> control even in expert mode.

I had a feeling while writing that this response might come. It's
correct. "much more" instead of "total"?

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 15:55, songbird a écrit :


  maybe grub needs something provided by the
modules:

usbms, ehci, uhci or ohci?


Be careful if you're going to use driver modules (USB, PATA, AHCI...) to 
get direct access to a device. It disables access to *all* devices 
through the BIOS, including the device GRUB was booted from and reads 
its own files.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Sat 15 Apr 2017 at 11:52:09 -0500, Richard Owlett wrote:

> On 04/15/2017 05:24 AM, Brian wrote:
> >
> >Now for a big "but". :)
> >
> >Your previous 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' stanza had
> >
> > linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1
> >
> >This one has
> >
> > linux /boot/vmlinuz-3.16.0-4-686-pae 
> > root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet
> >
> >The UUID has changed (I think you said you had done this) but the kernel
> >now looks for init using the UUID rather than on /dev/mmcblk0p1. How did
> >you manage to get something significantly different?
> 
> During the first install of Debian to the SD card I allowed the installer to
> create a single partition filling the whole card. The primary purpose of
> that install was a brute force determination of whether or not my individual
> machine could read/write an SD card. The machine was a replacement for a
> different used machine purchased from the vendor. My accepting the SD card
> was dependent on that test.
> 
> I had reinstalled Debian to a more reasonably sized partition. That would
> explain a changed UUID. During the reinstall I experimented with installing
> grub to the MBR of the SD card. No grub was installed the first time.

The UUID change is understandable. But hey - you've altered the ground
rules! Now, it seems, you are exploring booting from the card itself
rather than from GRUB on a hard disk. 

> >And, more to the point, is there booting from the card without any error
> >messages?

Did you overlook this question? You have said your machine does not
offer booting from an SD card. Your answer will be interesting.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 12:44, Brian a écrit :


simple install=some control of d-i.
expert install=total control of d-i.


Expert install gives more control, but is far from granting total 
control. The Debian installer still has many automated actions that you 
cannot control even in expert mode.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 16:28, David Wright a écrit :

On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:


A SD card reader such as the one the OP has just exposes the SD card
as what it is, a SD/MMC card (/dev/mmcblk*).


I assumed that the OP, writing about a laptop, had no card reader,
and was inserting the SD card directly into the computer.


How do you name that slot on the laptop or desktop where you insert the 
SD card, if not an embedded card reader ?



As it happens, if you do this with the ancient laptop I'm typing on,
it has the functionality of a card reader built into it and you get
a /dev/sd*, and that can be boot an SD card directly.


So there seems to be at least two kinds of SD card reader : those which 
expose themselves as a USB mass storage device and those which expose 
themselves as a SD/MMC device.


I have a couple of desktops with an embedded multi-card reader, 
connected to an internal USB port on the motherboard. But I do not have 
any SD card (no use), so I never checked to see what kind they are.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/15/2017 05:24 AM, Brian wrote:

On Fri 14 Apr 2017 at 13:33:40 -0500, Richard Owlett wrote:


On 04/14/2017 12:24 PM, Brian wrote:


Everything GRUB knows about devices comes from what the BIOS tells it.
They are more than just good friends. :)

It appears from 'ls' at a GRUB prompt that your GRUB does not know about
your SD card. Booting takes place but GRUB takes its time to think about
what it should do about not finding something it has been told to search
for. In the end, it decides to go ahead, but in some cases it wouldn't.
That would dispel your present mood of happiness.

While we are it it: your update-grub stanza does not contain a line with
set root=" in it. Could this possibly be a copy and paste error? I ask
because the line is present on Jessie and testing when the device is a
USB stick.


Just to eliminate any source of copy errors, here is the full contents of
grub.cfg created just prior to my most recent post
(https://lists.debian.org/debian-user/2017/04/msg00468.html). It refers to
the 2 new installs I mentioned in that post.


Thank you.


### BEGIN /etc/grub.d/30_os-prober ###
menuentry 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' --class gnu-linux
--class gnu --class os $menuentry_id_option
'osprober-gnulinux-simple-e57b2c64-74ec-4184-af71-d807e07f07dd' {
insmod part_msdos
insmod ext2
if [ x$feature_platform_search_hint = xy ]; then
  search --no-floppy --fs-uuid --set=root
e57b2c64-74ec-4184-af71-d807e07f07dd
else
  search --no-floppy --fs-uuid --set=root
e57b2c64-74ec-4184-af71-d807e07f07dd
fi
linux /boot/vmlinuz-3.16.0-4-686-pae
root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet
initrd /boot/initrd.img-3.16.0-4-686-pae
}


Definitely no "set root=" line. We will have to give GRUB credit for
knowing what it is doing. The absence of this line probably accounts for
your previous successful booting. With such a line you would possibly
have got "cannot get C/H/S values" as an error message. This throws
booting back to the GRUB menu.

Now for a big "but". :)

Your previous 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' stanza had

 linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1

This one has

 linux /boot/vmlinuz-3.16.0-4-686-pae 
root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet

The UUID has changed (I think you said you had done this) but the kernel
now looks for init using the UUID rather than on /dev/mmcblk0p1. How did
you manage to get something significantly different?


During the first install of Debian to the SD card I allowed the 
installer to create a single partition filling the whole card. The 
primary purpose of that install was a brute force determination of 
whether or not my individual machine could read/write an SD card. The 
machine was a replacement for a different used machine purchased from 
the vendor. My accepting the SD card was dependent on that test.


I had reinstalled Debian to a more reasonably sized partition. That 
would explain a changed UUID. During the reinstall I experimented with 
installing grub to the MBR of the SD card. No grub was installed the 
first time.






And, more to the point, is there booting from the card without any error
messages?

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/14/2017 01:19 PM, Brian wrote:

[snip]


 2. I only install Grub the *first* time I do a Debian install.
By poor design Grub puts the current install first on menu.
When experimenting with configuration as I do, the least
likely install to be functional is the latest.
This requires me to run update-grub on the "good" install.


That's ok. I tend to be more promiscuous; usually on a whim, like
wanting to put a particular entry at the top of GRUB's menu list.

When you do 'update-grub' do you still get no "set root=" line for the
SD card in the grub.cfg?
[snip]


I don't know how many times I've *MIS*read that last sentence ;<
Mentally I was correcting non-existent typos, thus totally garbling it.
I just spent several hours wandering thru grub files and loosely related 
documentation.

I've ended up at
www.gnu.org/software/grub/manual/grub.html#Multi_002dboot-manual-config
which says "Currently autogenerating config files for multi-boot 
environments depends on os-prober and has several shortcomings. ..."
I also discovered there that placing a customized grub in its own 
partition is not only possible, but recommended in some situations. I 
had read somewhere that that option had expired with grub-legacy. That, 
although taking much time to understand, will solve a *MESS* of grub 
problems for me ;/

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Sat 15 Apr 2017 at 11:05:12 (+0200), Pascal Hambourg wrote:
> Le 15/04/2017 à 02:37, David Wright a écrit :
> >
> >Of course, an SD card can be
> >made to look like a USB stick just by sticking it in a card reader.
> 
> I guess you mean "in a USB-to-SD card adapter", which translates a
> SD card into a USB mass storage device (/dev/sd*).

Yes, you guess correctly. Writing 40 years ago, I would use those same
words to describe a washing-machine sized object for reading punched
cards. Nowadays, if you type "card reader" into google, you will be
proffered several more sophisticated ones than mine, together with
some different varieties of credit card reader.

> A SD card reader such as the one the OP has just exposes the SD card
> as what it is, a SD/MMC card (/dev/mmcblk*).

I assumed that the OP, writing about a laptop, had no card reader,
and was inserting the SD card directly into the computer.

As it happens, if you do this with the ancient laptop I'm typing on,
it has the functionality of a card reader built into it and you get
a /dev/sd*, and that can be boot an SD card directly.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[songbird]

Brian wrote:
> songbird wrote:
>
>>   what i would do for grins is unplug the
>> devices other than the SSD and the installation 
>> media and then do a base system (simple install 
>> - not expert) and see what the installer does 
>> detect and writes in the fstab and grub menu 
>> (it may even boot).  you may then use those 
>
> This is cargo cult.

  thus my expression "for grins".  ;)

  i noticed my own lack of previous comprehension
as i thought all along the OP was putting things
on an SSD device and not a SD chip.  my SD reader
is on a USB header device with many other things.

  maybe grub needs something provided by the
modules:

usbms, ehci, uhci or ohci?

  these can be tested at the command line and
then doing ls -l to see if the device shows up
or not...


> simple install=some control of d-i.
> expert install=total control of d-i.
>
> In both cases the OP does not want to install a boot loader to the
> device.

  dd can get rid of it easily enough.

  i think the possible confusion for a more
ancient machine may be cleared up enough to 
get workable bits.  how we get to workable
bits may be a different route, but in the end
if they work we have gotten to the cheese in
this particular maze of twisty-turny passages
even if they all look alike (1's and 0's).


  songbird

Re: Possibly erroneous "device not present" message during boot

[songbird]

Brian wrote:
...
> And, more to the point, is there booting from the card without any error
> messages?

  ...

  and as additional info you can adjust some things
in grub via editing /etc/default/grub (and running
update-grub) if you don't like how the menu is being
generated.


  songbird

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Fri 14 Apr 2017 at 20:02:11 -0400, songbird wrote:

>   what i would do for grins is unplug the
> devices other than the SSD and the installation 
> media and then do a base system (simple install 
> - not expert) and see what the installer does 
> detect and writes in the fstab and grub menu 
> (it may even boot).  you may then use those 

This is cargo cult.

simple install=some control of d-i.
expert install=total control of d-i.

In both cases the OP does not want to install a boot loader to the
device.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Fri 14 Apr 2017 at 13:33:40 -0500, Richard Owlett wrote:

> On 04/14/2017 12:24 PM, Brian wrote:
> >
> >Everything GRUB knows about devices comes from what the BIOS tells it.
> >They are more than just good friends. :)
> >
> >It appears from 'ls' at a GRUB prompt that your GRUB does not know about
> >your SD card. Booting takes place but GRUB takes its time to think about
> >what it should do about not finding something it has been told to search
> >for. In the end, it decides to go ahead, but in some cases it wouldn't.
> >That would dispel your present mood of happiness.
> >
> >While we are it it: your update-grub stanza does not contain a line with
> >set root=" in it. Could this possibly be a copy and paste error? I ask
> >because the line is present on Jessie and testing when the device is a
> >USB stick.
> 
> Just to eliminate any source of copy errors, here is the full contents of
> grub.cfg created just prior to my most recent post
> (https://lists.debian.org/debian-user/2017/04/msg00468.html). It refers to
> the 2 new installs I mentioned in that post.

Thank you.

> ### BEGIN /etc/grub.d/30_os-prober ###
> menuentry 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' --class gnu-linux
> --class gnu --class os $menuentry_id_option
> 'osprober-gnulinux-simple-e57b2c64-74ec-4184-af71-d807e07f07dd' {
>   insmod part_msdos
>   insmod ext2
>   if [ x$feature_platform_search_hint = xy ]; then
> search --no-floppy --fs-uuid --set=root
> e57b2c64-74ec-4184-af71-d807e07f07dd
>   else
> search --no-floppy --fs-uuid --set=root
> e57b2c64-74ec-4184-af71-d807e07f07dd
>   fi
>   linux /boot/vmlinuz-3.16.0-4-686-pae
> root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet
>   initrd /boot/initrd.img-3.16.0-4-686-pae
> }

Definitely no "set root=" line. We will have to give GRUB credit for
knowing what it is doing. The absence of this line probably accounts for
your previous successful booting. With such a line you would possibly
have got "cannot get C/H/S values" as an error message. This throws
booting back to the GRUB menu.

Now for a big "but". :)

Your previous 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' stanza had

 linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1

This one has

 linux /boot/vmlinuz-3.16.0-4-686-pae 
root=UUID=e57b2c64-74ec-4184-af71-d807e07f07dd ro quiet

The UUID has changed (I think you said you had done this) but the kernel
now looks for init using the UUID rather than on /dev/mmcblk0p1. How did
you manage to get something significantly different?

And, more to the point, is there booting from the card without any error
messages?

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 15/04/2017 à 02:37, David Wright a écrit :


Of course, an SD card can be
made to look like a USB stick just by sticking it in a card reader.


I guess you mean "in a USB-to-SD card adapter", which translates a SD 
card into a USB mass storage device (/dev/sd*).
A SD card reader such as the one the OP has just exposes the SD card as 
what it is, a SD/MMC card (/dev/mmcblk*).

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Fri 14 Apr 2017 at 12:05:38 (-0500), Richard Owlett wrote:
> On 04/13/2017 05:55 PM, Brian wrote:
> >On Thu 13 Apr 2017 at 20:05:22 +0200, Pascal Hambourg wrote:
> >
> >>David is right : you don't really boot from the SD card.
> >
> >The OP never claimed he was booting from the SD card. He particularly
> >said he did not install GRUB to the card.
> >
> >>GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem is on
> >>the SD card.
> >
> >Yes. That's what the linux line says too.
> >
> 
> I'll try to clarify some details.
> 
> My installation protocol.
>  1. I always use "Expert" as that way the installer will
> do fewer things I'm not aware of.
>  2. I only install Grub the *first* time I do a Debian install.
> By poor design Grub puts the current install first on menu.

When are you going to let this drop?

On Thu, 05 Jul 2012 08:23:19 (-0500), Richard Owlett wrote:

| What I need:
|   1. newest install to be on bottom of the list

On Sat, 24 Nov 2012 15:33:49 (-0600), Richard Owlett wrote:

| The GRUB team made many reasonable design decisions.
| Though they match vast MAJORITY of users, their choices annoy me every
| time I boot ;<
|
| What would fit best with my habits is that the precedence of OS to
| boot would be
| "first installed -> first boot choice" *NOT* "last installed -> first
| boot choice".

On Sat, 02 Feb 2013 08:47:37 (-0600), Richard Owlett wrote:

| The latest installation goes to the top of the displayed list and is
| also selected as the default OS to boot.
| That is unsatisfactory as the first OS will will always be closest to
| a standard install - i.e. most likely to run.

On Sun, 24 Jul 2016 13:53:59 (-0500), Richard Owlett wrote:

| What I *REQUIRE* is that GRUB2's menu list available OS monotonically
| by partition number. For reasons of logic and sanity the menu items
| should be in the order of
| sda1, sda2, ..., sdaMAX. I could work with sdaMAX, ..., sda2, sda1 .
|
| The purpose AND rationale SHALL be that the default OS choice *SHALL
| BE* the first OS installed.



> When experimenting with configuration as I do, the least
> likely install to be functional is the latest.
> This requires me to run update-grub on the "good" install.
>  3. Similarly a swap partition is specified only on the first
> install as the installer insists on destroying the UUID of
> the existing swap partition. It is simpler to edit only the
> fstab of latest install than to edit those for all other
> installs each time.
>  4. All installs in this thread have been done using DVD 1 of
> 13 of Debian 8.6.0 - thus all intrinsically use the same
> kernel.
> 
> I've done some additional observations and test installs.
> 1. The BIOS of the Lenovo T510 can be directed to boot from
>the CD/DVD drive, hard disk, or any attached USB flash
>drive. It *cannot* be directed to boot from the SD card.
> 2. I did an install to a USB flash drive including installing
>Grub2 to the MBR of that flash drive. When selecting the
>SD card from the grub menu I see nothing different.
> 3. I did a new install to the SD card specifying a different set
>of packages and installing grub to the MBR of the SD card.
>Once again no behavioral difference.

Observation 1 would make that inevitable, wouldn't it. Unless, of
course, you stick your SD card in a card reader. These little gizmos
used to be bundled with SD cards when you bought them; I don't know if
they still are. I have a large one that came free with the first
(only) SD card I bought, and a couple of tiny ones that came with
SDmicros bought a long while back. The latter also come with
converters so that you can put an SDmicro in the SD reader or an SD
slot.

> Does any of this justify a bug report. Especially as I do not
> have the bandwidth to do a netinstall of a pre-release version?

You've posted your grub.cfg but I haven't bothered to try and
reverse engineer how you produced it. I haven't seen any
behaviour yet that I would call a bug. What did you have in mind?
You could say whether you think the bug is in the _generation_ of
grub.cfg or in its _execution_ when you boot.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Fri 14 Apr 2017 at 20:02:11 (-0400), songbird wrote:
> Richard Owlett wrote:
> ...
> [file a bug?]
> 
>   i'm not sure this is a bug.  you've
> done an expert install, but not let grub
> finish so by my estimate you've left the
> install where manual patching is needed.

I would agree with that. It's not clear to me why an SD is being used
in this way as the OP has at last revealed that the computer is unable
to boot from an SD plugged in directly. (Of course, an SD card can be
made to look like a USB stick just by sticking it in a card reader.
Then it will boot.)

>   right now this thread is showing up via
> go ogle so people will find it if they have
> similar issues and they can maybe sort it
> out.
> 
>   what i would do for grins is unplug the
> devices other than the SSD and the installation 
> media and then do a base system (simple install 
> - not expert) and see what the installer does 
> detect and writes in the fstab and grub menu 
> (it may even boot).  you may then use those 
> things in the previously installed grub and
> fstab files (for the system which boots and
> runs but does not mount the SSD).

Googling may now throw up this thread when people are searching for
SSD installation. SSDs are a different animal from SD cards.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[songbird]

Richard Owlett wrote:
...
[file a bug?]

  i'm not sure this is a bug.  you've
done an expert install, but not let grub
finish so by my estimate you've left the
install where manual patching is needed.

  right now this thread is showing up via
go ogle so people will find it if they have
similar issues and they can maybe sort it
out.

  what i would do for grins is unplug the
devices other than the SSD and the installation 
media and then do a base system (simple install 
- not expert) and see what the installer does 
detect and writes in the fstab and grub menu 
(it may even boot).  you may then use those 
things in the previously installed grub and
fstab files (for the system which boots and
runs but does not mount the SSD).

  if you do this report the contents of the 
fstab and the grub.cfg


> Especially as I do not
> have the bandwidth to do a netinstall of a pre-release version?

  that's really unfortunate.


  songbird

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 14/04/2017 à 19:05, Richard Owlett a écrit :


 2. I only install Grub the *first* time I do a Debian install.
By poor design Grub puts the current install first on menu.


It's not poor design. It is perfectly normal that the system which 
installs GRUB is first in the menu. If you don't like it, you can change it.



When experimenting with configuration as I do, the least
likely install to be functional is the latest.


When experimenting, it is wise not to replace the existing working boot 
loader with a new one which depends on the newly installed system. It 
has nothing to do with GRUB's supposedly poor design.



 3. Similarly a swap partition is specified only on the first
install as the installer insists on destroying the UUID of
the existing swap partition.


No, it doesn't insist. It just marks the existing swaps as to be used by 
default, which in turns marks them to be formatted with a new UUID. You 
are free to mark them as not to be used and they won't be reformatted. 
It just insist on using a swap, but it does not have to be any of the 
existing ones.



Does any of this justify a bug report.


No. The only bug is that the system installed on the SD card boots while 
it should not.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/14/2017 12:33 PM, Brian wrote:

On Thu 13 Apr 2017 at 10:58:10 -0500, Richard Owlett wrote:


On 04/12/2017 01:49 PM, songbird wrote:


 another thing you can do is construct your own
menu entry and put it in /etc/grub.d/40_custom file


I don't see that being informative for this problem.


Quite the opposite; this is a very useful suggestion. Imagine that
update-grub didn't give you something workable. You would (or should)
be glad of the opportunity to roll your own. I do it all the time with
removable devices.

In your case, if grub-install had given you a "root=" line in a stanza
it is possible you would not be booting the SD card. A 40_custom file
needn't have it in.



I'll add it to my TODO list.
I've never attempted to edit /etc/grub.d/40_custom file or its cousins.
I've just received some need documentation for something I'm doing for 
someone else.

Anyone have a few spare 100 hour days ;/

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Fri 14 Apr 2017 at 12:05:38 -0500, Richard Owlett wrote:

> On 04/13/2017 05:55 PM, Brian wrote:
> >On Thu 13 Apr 2017 at 20:05:22 +0200, Pascal Hambourg wrote:
> >
> >>David is right : you don't really boot from the SD card.
> >
> >The OP never claimed he was booting from the SD card. He particularly
> >said he did not install GRUB to the card.
> >
> >>GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem is on
> >>the SD card.
> >
> >Yes. That's what the linux line says too.
> 
> I'll try to clarify some details.

Your previous descriptions were good; clarification cannot do any harm.

> My installation protocol.
>  1. I always use "Expert" as that way the installer will
> do fewer things I'm not aware of.

Very wise. If you are familiar with each stage of the installation you
can more more easily spot anomalies. "Expert"=="more control".

>  2. I only install Grub the *first* time I do a Debian install.
> By poor design Grub puts the current install first on menu.
> When experimenting with configuration as I do, the least
> likely install to be functional is the latest.
> This requires me to run update-grub on the "good" install.

That's ok. I tend to be more promiscuous; usually on a whim, like
wanting to put a particular entry at the top of GRUB's menu list.

When you do 'update-grub' do you still get no "set root=" line for the
SD card in the grub.cfg?

>  3. Similarly a swap partition is specified only on the first
> install as the installer insists on destroying the UUID of
> the existing swap partition. It is simpler to edit only the
> fstab of latest install than to edit those for all other
> installs each time.

That's ok too. I often don't bother with a swap partition. Some of my
machines have less than 4 GB of disk space. I see space for packages as
more important than swap space. 

>  4. All installs in this thread have been done using DVD 1 of
> 13 of Debian 8.6.0 - thus all intrinsically use the same
> kernel.

Yes.

> I've done some additional observations and test installs.
> 1. The BIOS of the Lenovo T510 can be directed to boot from
>the CD/DVD drive, hard disk, or any attached USB flash
>drive. It *cannot* be directed to boot from the SD card.

I'm at a disadvantage here. I do not have an SD card. I do not have a
USB hub. However, I thought USB sticks and SD cards were the same (or
at least similar) when it came to booting.

> 2. I did an install to a USB flash drive including installing
>Grub2 to the MBR of that flash drive. When selecting the
>SD card from the grub menu I see nothing different.

Does that not substantiate what I said just now? 

> 3. I did a new install to the SD card specifying a different set
>of packages and installing grub to the MBR of the SD card.
>Once again no behavioral difference.

I cannot believe your install to the card would be capable of criticism.

> Does any of this justify a bug report. Especially as I do not
> have the bandwidth to do a netinstall of a pre-release version?

I mentioned "bug" in another mail. I wish I hadn't. It is not that bugs
in GRUB or the Debian grub package do not arise, but I'd want to know
more about SD cards before committing myself.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/14/2017 12:24 PM, Brian wrote:

On Wed 12 Apr 2017 at 10:14:59 -0500, Richard Owlett wrote:


On 04/12/2017 07:39 AM, songbird wrote:


- the bios may not be set correctly to find the device.


I do not see how the bios might be involved.
Grub2 has been installed to the MBR of /dev/sda .
The only known reference to Grub2 is in the instance of Debian residing on
/dev/sda1 .


Everything GRUB knows about devices comes from what the BIOS tells it.
They are more than just good friends. :)

It appears from 'ls' at a GRUB prompt that your GRUB does not know about
your SD card. Booting takes place but GRUB takes its time to think about
what it should do about not finding something it has been told to search
for. In the end, it decides to go ahead, but in some cases it wouldn't.
That would dispel your present mood of happiness.

While we are it it: your update-grub stanza does not contain a line with
set root=" in it. Could this possibly be a copy and paste error? I ask
because the line is present on Jessie and testing when the device is a
USB stick.



Just to eliminate any source of copy errors, here is the full contents 
of grub.cfg created just prior to my most recent post 
(https://lists.debian.org/debian-user/2017/04/msg00468.html). It refers 
to the 2 new installs I mentioned in that post.





#
# DO NOT EDIT THIS FILE
#
# It is automatically generated by grub-mkconfig using templates
# from /etc/grub.d and settings from /etc/default/grub
#

### BEGIN /etc/grub.d/00_header ###
if [ -s $prefix/grubenv ]; then
  set have_grubenv=true
  load_env
fi
if [ "${next_entry}" ] ; then
   set default="${next_entry}"
   set next_entry=
   save_env next_entry
   set boot_once=true
else
   set default="0"
fi

if [ x"${feature_menuentry_id}" = xy ]; then
  menuentry_id_option="--id"
else
  menuentry_id_option=""
fi

export menuentry_id_option

if [ "${prev_saved_entry}" ]; then
  set saved_entry="${prev_saved_entry}"
  save_env saved_entry
  set prev_saved_entry=
  save_env prev_saved_entry
  set boot_once=true
fi

function savedefault {
  if [ -z "${boot_once}" ]; then
saved_entry="${chosen}"
save_env saved_entry
  fi
}
function load_video {
  if [ x$feature_all_video_module = xy ]; then
insmod all_video
  else
insmod efi_gop
insmod efi_uga
insmod ieee1275_fb
insmod vbe
insmod vga
insmod video_bochs
insmod video_cirrus
  fi
}

if [ x$feature_default_font_path = xy ] ; then
   font=unicode
else
insmod part_msdos
insmod ext2
set root='hd0,msdos1'
if [ x$feature_platform_search_hint = xy ]; then
  search --no-floppy --fs-uuid --set=root --hint-bios=hd0,msdos1 
--hint-efi=hd0,msdos1 --hint-baremetal=ahci0,msdos1 
8fe2965b-01ed-4227-9155-bc3e47ddf077

else
  search --no-floppy --fs-uuid --set=root 
8fe2965b-01ed-4227-9155-bc3e47ddf077

fi
font="/usr/share/grub/unicode.pf2"
fi

if loadfont $font ; then
  set gfxmode=auto
  load_video
  insmod gfxterm
  set locale_dir=$prefix/locale
  set lang=en_US
  insmod gettext
fi
terminal_output gfxterm
if [ "${recordfail}" = 1 ] ; then
  set timeout=-1
else
  if [ x$feature_timeout_style = xy ] ; then
set timeout_style=menu
set timeout=5
  # Fallback normal timeout code in case the timeout_style feature is
  # unavailable.
  else
set timeout=5
  fi
fi
### END /etc/grub.d/00_header ###

### BEGIN /etc/grub.d/05_debian_theme ###
insmod part_msdos
insmod ext2
set root='hd0,msdos1'
if [ x$feature_platform_search_hint = xy ]; then
  search --no-floppy --fs-uuid --set=root --hint-bios=hd0,msdos1 
--hint-efi=hd0,msdos1 --hint-baremetal=ahci0,msdos1 
8fe2965b-01ed-4227-9155-bc3e47ddf077

else
  search --no-floppy --fs-uuid --set=root 
8fe2965b-01ed-4227-9155-bc3e47ddf077

fi
insmod png
if background_image /usr/share/images/desktop-base/lines-grub.png; then
  set color_normal=white/black
  set color_highlight=black/white
else
  set menu_color_normal=cyan/blue
  set menu_color_highlight=white/blue
fi
### END /etc/grub.d/05_debian_theme ###

### BEGIN /etc/grub.d/10_linux ###
function gfxmode {
set gfxpayload="${1}"
}
set linux_gfx_mode=
export linux_gfx_mode
menuentry 'Debian GNU/Linux' --class debian --class gnu-linux --class 
gnu --class os $menuentry_id_option 
'gnulinux-simple-8fe2965b-01ed-4227-9155-bc3e47ddf077' {

load_video
insmod gzio
if [ x$grub_platform = xxen ]; then insmod xzio; insmod lzopio; fi
insmod part_msdos
insmod ext2
set root='hd0,msdos1'
if [ x$feature_platform_search_hint = xy ]; then
	  search --no-floppy --fs-uuid --set=root --hint-bios=hd0,msdos1 
--hint-efi=hd0,msdos1 --hint-baremetal=ahci0,msdos1 
8fe2965b-01ed-4227-9155-bc3e47ddf077

else
	  search --no-floppy --fs-uuid --set=root 
8fe2965b-01ed-4227-9155-bc3e47ddf077

fi
echo'Loading Linux 3.16.0-4-686-pae ...'
	linux	/boot/vmlinuz-3.16.0-4-686-pae 
root=UUID=8fe2965b-01ed-4227-9155-bc3e47ddf077 ro  quiet

echo'Loading initial ramdisk ...'

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Thu 13 Apr 2017 at 10:58:10 -0500, Richard Owlett wrote:

> On 04/12/2017 01:49 PM, songbird wrote:
> >
> >  another thing you can do is construct your own
> >menu entry and put it in /etc/grub.d/40_custom file
> 
> I don't see that being informative for this problem.

Quite the opposite; this is a very useful suggestion. Imagine that
update-grub didn't give you something workable. You would (or should)
be glad of the opportunity to roll your own. I do it all the time with
removable devices.

In your case, if grub-install had given you a "root=" line in a stanza
it is possible you would not be booting the SD card. A 40_custom file
needn't have it in.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 12 Apr 2017 at 10:14:59 -0500, Richard Owlett wrote:

> On 04/12/2017 07:39 AM, songbird wrote:
> >
> > - the bios may not be set correctly to find the device.
> 
> I do not see how the bios might be involved.
> Grub2 has been installed to the MBR of /dev/sda .
> The only known reference to Grub2 is in the instance of Debian residing on
> /dev/sda1 .

Everything GRUB knows about devices comes from what the BIOS tells it.
They are more than just good friends. :)

It appears from 'ls' at a GRUB prompt that your GRUB does not know about
your SD card. Booting takes place but GRUB takes its time to think about
what it should do about not finding something it has been told to search
for. In the end, it decides to go ahead, but in some cases it wouldn't.
That would dispel your present mood of happiness.

While we are it it: your update-grub stanza does not contain a line with
set root=" in it. Could this possibly be a copy and paste error? I ask
because the line is present on Jessie and testing when the device is a
USB stick.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/13/2017 05:55 PM, Brian wrote:

On Thu 13 Apr 2017 at 20:05:22 +0200, Pascal Hambourg wrote:


David is right : you don't really boot from the SD card.


The OP never claimed he was booting from the SD card. He particularly
said he did not install GRUB to the card.


GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem is on
the SD card.


Yes. That's what the linux line says too.



I'll try to clarify some details.

My installation protocol.
 1. I always use "Expert" as that way the installer will
do fewer things I'm not aware of.
 2. I only install Grub the *first* time I do a Debian install.
By poor design Grub puts the current install first on menu.
When experimenting with configuration as I do, the least
likely install to be functional is the latest.
This requires me to run update-grub on the "good" install.
 3. Similarly a swap partition is specified only on the first
install as the installer insists on destroying the UUID of
the existing swap partition. It is simpler to edit only the
fstab of latest install than to edit those for all other
installs each time.
 4. All installs in this thread have been done using DVD 1 of
13 of Debian 8.6.0 - thus all intrinsically use the same
kernel.

I've done some additional observations and test installs.
1. The BIOS of the Lenovo T510 can be directed to boot from
   the CD/DVD drive, hard disk, or any attached USB flash
   drive. It *cannot* be directed to boot from the SD card.
2. I did an install to a USB flash drive including installing
   Grub2 to the MBR of that flash drive. When selecting the
   SD card from the grub menu I see nothing different.
3. I did a new install to the SD card specifying a different set
   of packages and installing grub to the MBR of the SD card.
   Once again no behavioral difference.

Does any of this justify a bug report. Especially as I do not
have the bandwidth to do a netinstall of a pre-release version?

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Thu 13 Apr 2017 at 20:05:22 +0200, Pascal Hambourg wrote:

> David is right : you don't really boot from the SD card.

The OP never claimed he was booting from the SD card. He particularly
said he did not install GRUB to the card.

> GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem is on
> the SD card.

Yes. That's what the linux line says too.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Thu 13 Apr 2017 at 12:28:58 -0500, David Wright wrote:

> Not being expert with Grub myself, I don't know how it reacts to a
> failure during the search. Just a quick glance at my own cfg shows

For the third time in this thread: GRUB reacts by shrugging its
shoulders and booting whatever the linux and initrd say. If they
are not sensible you can forget about booting. You can judge the
importance of a search line in your grub.cfg after doing this.

This is so easy to test. Just do some deletions by editing the GRUB
menu when it appears. Anything between iffi would do for a start.

> that there are "set root" commands very early on in the file. So,
> again I'm guessing, this may explain where the kernel is being
> read from: one of those early commands has given Grub a fallback
> device.

What effect a USB hub or card reader has on the booting process
I do not know. It could be a factor.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 13/04/2017 à 19:33, Richard Owlett a écrit :

On 04/13/2017 10:48 AM, Pascal Hambourg wrote:


USB drive ? Then (hd1) is not the SD card ? If you boot with no USB
drive plugged in, there is no (hd1), right ?


Correct on all counts. IIUC the SD card slot is connected to a USB hub.
When not the booted system, it automatically mounted in the same
apparent manner as a USB flash drive.


I don't think so, because it appears as /dev/mmcblk0 in grub.cfg, i.e. a 
MMC/SD card. A USB mass storage device would appear as /dev/sdb or so.



Then the SD card does not
appear as a usable device for GRUB and my twisted theory may still be
correct after all.


Ok then, here is my full theory. You're not going to like it.

We now know that the BIOS does not expose the SD card so GRUB cannot see 
it and the search command fails and prints the error message.


But then how does the system on the SD card boot ?

The purpose of the search command is to set the $root variable value 
with the device havind the searched UUID. As it fails, the variable 
remains unmodified and keeps its previous value. The menu entry does not 
set a default value for $root (which is rather unusual in my 
experience), so the current value comes from a previous assignement in 
grub.cfg. It's probably the root device of the system owning GRUB on the 
hard disk drive.


(In GRUB's shell, you can print the $root value with the "set" command, 
and check that this partition contains a 
/boot/grub/vmlinuz-3.16.0-4-686-pae file directory with the command "ls 
/boot/grub".)


GRUB uses $root as a default device used in paths which do not specify a 
device, such as the ones in the linux and initrd commands of the menu 
entry. By chance, it seems that this location contains the same kernel 
as the one on the SD card, so GRUB can load and boot it from the hard 
disk drive. But GRUB passes the SD card partition as the root device 
root=/dev/mmcblk0p1 to the kernel command line, so the kernel comes from 
the disk and the root filesystem comes from the SD card. Unlike GRUB, 
the kernel does not rely on the BIOS to access the SD card.


David is right : you don't really boot from the SD card.
GRUB is on the HDD. The kernel is on the HDD. Only the root filesystem 
is on the SD card.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/13/2017 10:48 AM, Pascal Hambourg wrote:

Le 13/04/2017 à 13:40, Richard Owlett a écrit :

On 04/12/2017 02:03 PM, Pascal Hambourg wrote:

Le 12/04/2017 à 20:33, Richard Owlett a écrit :

On 04/12/2017 12:13 PM, Pascal Hambourg wrote:


[snip]

- while the SD card is inserted, in the GRUB menu press "c" to enter a
GRUB shell and report the result of the "ls" command.


(hd0) (hd0, msdos9) (hd0, msdos8) (hd0, msdos7) (hd0, msdos6)
(hd0, msdos5) (hd0, msdos1) (hd1) (hd1, msdos1)


Some BIOS expose a removable media only when booting from it, so my
theory was that your BIOS did not expose the SD card and GRUB could
not see it. The result of ls proves me wrong.

Can you also report the output of "ls (hd1,msdos1)" to check what GRUB
sees of the partition on the card ?


It gives the partitioning description, UUID, and label of the USB flash
drive that was plugged in. It is FAT formatted and used to exchange
information with my Windows machine.


USB drive ? Then (hd1) is not the SD card ? If you boot with no USB
drive plugged in, there is no (hd1), right ?


Correct on all counts. IIUC the SD card slot is connected to a USB hub. 
When not the booted system, it automatically mounted in the same 
apparent manner as a USB flash drive.



Then the SD card does not
appear as a usable device for GRUB and my twisted theory may still be
correct after all.


Any suggested reading on the "BASH like shell" I just used?


The GRUB manual at 


It was what prompted me to ask ;/


Sorry, I do not know any better documentation.

Re: Possibly erroneous "device not present" message during boot

[David Wright]

On Thu 13 Apr 2017 at 17:48:41 (+0200), Pascal Hambourg wrote:
> Le 13/04/2017 à 13:40, Richard Owlett a écrit :
> >On 04/12/2017 02:03 PM, Pascal Hambourg wrote:
> >>Le 12/04/2017 à 20:33, Richard Owlett a écrit :
> >>>On 04/12/2017 12:13 PM, Pascal Hambourg wrote:
> >
> >[snip]
> - while the SD card is inserted, in the GRUB menu press "c" to enter a
> GRUB shell and report the result of the "ls" command.
> >>>
> >>>(hd0) (hd0, msdos9) (hd0, msdos8) (hd0, msdos7) (hd0, msdos6)
> >>>(hd0, msdos5) (hd0, msdos1) (hd1) (hd1, msdos1)
> >>
> >>Some BIOS expose a removable media only when booting from it, so my
> >>theory was that your BIOS did not expose the SD card and GRUB could
> >>not see it. The result of ls proves me wrong.
> >>
> >>Can you also report the output of "ls (hd1,msdos1)" to check what GRUB
> >>sees of the partition on the card ?
> >
> >It gives the partitioning description, UUID, and label of the USB flash
> >drive that was plugged in. It is FAT formatted and used to exchange
> >information with my Windows machine.
> 
> USB drive ? Then (hd1) is not the SD card ? If you boot with no USB
> drive plugged in, there is no (hd1), right ? Then the SD card does
> not appear as a usable device for GRUB and my twisted theory may
> still be correct after all.

This is just a guess. The OP says they "booted to the SD card". So
presumably the SD card was plugged in, and we have seen mention of
/dev/mmcblk0p1 which means it was plugged in when Grub made the
grub.cfg file. But I'm not yet convinced that the machine is actually
able to boot from an SD card (unless you put it in a USB card reader).

The boot stanza contains:

search --no-floppy --fs-uuid --set=root 380e2a6d-f851-4fd1-9db2-869a0982b511
[...]
linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1
initrd /boot/initrd.img-3.16.0-4-686-pae

So it's possible that the kernel and initrd are coming from the USB
stick that the OP has left in, or even the hard drive.

I normally expect to see
search --no-floppy --foo --set=root bar
[...]
linux /boot/vmlinuz-... root=foo=bar

where foo depends on whether you're using /dev/ or UUID or LABEL,
and bar is the value appropriate to foo.

That's not what the OP posted. Also their response

> It gives the partitioning description, UUID, and label of the USB
> flash drive that was plugged in. It is FAT formatted and used to
> exchange information with my Windows machine.

sadly lacks the actual information asked for as is often the case.
What is the UUID/LABEL of the USB device? I'm not bothered about
the Windows machine.

> >>>Any suggested reading on the "BASH like shell" I just used?
> >>
> >>The GRUB manual at 
> >
> >It was what prompted me to ask ;/
> 
> Sorry, I do not know any better documentation.

I'm surprised the OP isn't best qualified to answer this question;
battling with Grub goes back at least five years.

Not being expert with Grub myself, I don't know how it reacts to a
failure during the search. Just a quick glance at my own cfg shows
that there are "set root" commands very early on in the file. So,
again I'm guessing, this may explain where the kernel is being
read from: one of those early commands has given Grub a fallback
device.

Cheers,
David.

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Thu 13 Apr 2017 at 10:17:23 -0500, Richard Owlett wrote:

> On 04/12/2017 02:23 PM, Brian wrote:
> >
> >So try this:
> >When the GRUB menu displays press the "e" key and remove the
> >
> >  search --no-floppy --fs-uuid --set=root 
> > 380e2a6d-f851-4fd1-9db2-869a0982b511
> >
> >Then press the "F10" key. Can you still boot? Do you get the onscreen
> >message you are concerned about?
> >
> Something *ODD* is going on in that area.
> As that string appears in both the "then" and "else" clauses I:
>   1. tried removing them individually leaving a blank line.
>  For both cases grub crashed with a syntax error.

I'm not sure this would be a good idea without reconstructing the if...fi
phrasing.

>   2. removed entire if-then-else-fi construct.
>  That booted without the "device not found" message.

This is a much better technique than the one I suggested, which is a
little unclear as to what to remove. A GRUB stanza does not need a
search line or what is between if and fi. It can be useful, but the
linux and initrd lines are sufficient much of the time.

>   3. changed the last digit of the UUID.
>  That demonstrated that the "then" clause was the one executed.

This confirms your second point.

>   4. used gparted to change the UUID of the SD card.
>  That crashed in ways seemingly unrelated to anything.
>  I'm going to update my preferred preseed.cfg and reinstall to
>  both /dev/sda1 and the SD card in a reproducible way.

I'm afraid I'm not going to try this. Off the top my head I'd expect the
device still to boot with a suitable GRUB stanza

"Erroneous" isn't a bad description of the message. As you describe it
the boot proceeds whether or not the "ENTER" key is pressed. It would be
interesting to know why the UUID is not recognised. Perhaps a small bug
in GRUB? It is a minor inconvenience. A LABEL is the way to go but I
think 'update-grub' will only generate a search line with a UUID in it.

> What log files should I save after each test boot.

The issue arises with GRUB. The will be no records in the logs as the
system has not booted.

-- 
Brian.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/12/2017 01:49 PM, songbird wrote:

Pascal Hambourg wrote:

Le 12/04/2017 à 17:14, Richard Owlett a écrit :


Whether initiated after power-on OR a restart the observed sequence is:
1. Appearance of the Grub2 menu with a choice of 4 instances of Debian.
2. Select instance installed on the SD card.
3. Screen clears, this message appears against the Debian 8 background.
   error:  no such device: 380e2a6d-f851-4fd1-9db2-869a0982b511.
   Press any key to continue ...
4. Otherwise the instance of Debian on the SD card boots routinely.


This is a GRUB error message which cannot find the UUID specified in a
"search" command.

In order to investigate, can you
- report the menu entry code for the SD card system in
/boot/grub/grub.cfg (the one from the system on the hard drive owning
GRUB, not the one on the SD card) ;


  yep!



- while the SD card is inserted, in the GRUB menu press "c" to enter a
GRUB shell and report the result of the "ls" command.


  also, check that the mount points exist for
the file system.

  you may think you've installed things to one
location only to later find you've actually
put them someplace else...


I do not understand those 2 questions.
Gparted displays what I expect to see.
All partitions are mountable using the "Places" menu on MATE's panel.




  another thing you can do is construct your own
menu entry and put it in /etc/grub.d/40_custom file


I don't see that being informative for this problem.



here is an example from when i was playing around
last week (you may need to insmod something else(s)
to get the efi, ssd, ext4, whatever partition table
type you use, etc. going):  [and only use the lines from
between the ='s]  adjust as needed/desired.  :)

oh and of course, check your partition tables on the
devices to make sure they're actually what you expect.



???
See comment re your 1st 2 questions.




==
#!/bin/sh
exec tail -n +3 $0
# This file provides an easy way to add custom menu entries.  Simply type the
# menu entries you want to add after this comment.  Be careful not to change
# the 'exec tail' line above.

menuentry "Count to three before launching..." {

echo "1..."
sleep 1
echo "2..."
sleep 1
echo "3..."
sleep 1

set linux_gfx_mode=
export linux_gfx_mode
load_video
insmod gzio
insmod part_msdos
insmod ext2
insmod usb
insmod fat

set root='hd0,msdos1'

echo'Loading Linux 4.9.0-1-686-pae ...'
linux   /boot/vmlinuz-4.9.0-1-686-pae ro root=LABEL=ROOT_007
echo'Loading initial ramdisk ...'
initrd  /boot/initrd.img-4.9.0-1-686-pae
}
==

  and then run update-grub to get your new menu.


  songbird

Re: Possibly erroneous "device not present" message during boot

[Pascal Hambourg]

Le 13/04/2017 à 13:40, Richard Owlett a écrit :

On 04/12/2017 02:03 PM, Pascal Hambourg wrote:

Le 12/04/2017 à 20:33, Richard Owlett a écrit :

On 04/12/2017 12:13 PM, Pascal Hambourg wrote:


[snip]

- while the SD card is inserted, in the GRUB menu press "c" to enter a
GRUB shell and report the result of the "ls" command.


(hd0) (hd0, msdos9) (hd0, msdos8) (hd0, msdos7) (hd0, msdos6)
(hd0, msdos5) (hd0, msdos1) (hd1) (hd1, msdos1)


Some BIOS expose a removable media only when booting from it, so my
theory was that your BIOS did not expose the SD card and GRUB could
not see it. The result of ls proves me wrong.

Can you also report the output of "ls (hd1,msdos1)" to check what GRUB
sees of the partition on the card ?


It gives the partitioning description, UUID, and label of the USB flash
drive that was plugged in. It is FAT formatted and used to exchange
information with my Windows machine.


USB drive ? Then (hd1) is not the SD card ? If you boot with no USB 
drive plugged in, there is no (hd1), right ? Then the SD card does not 
appear as a usable device for GRUB and my twisted theory may still be 
correct after all.



Any suggested reading on the "BASH like shell" I just used?


The GRUB manual at 


It was what prompted me to ask ;/


Sorry, I do not know any better documentation.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/12/2017 02:23 PM, Brian wrote:

On Wed 12 Apr 2017 at 13:33:02 -0500, Richard Owlett wrote:


On theory "too much better than too little" I see:

### BEGIN /etc/grub.d/20_linux_xen ###

### END /etc/grub.d/20_linux_xen ###

### BEGIN /etc/grub.d/30_os-prober ###
menuentry 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' --class gnu-linux
--class gnu --class os $menuentry_id_option
'osprober-gnulinux-simple-380e2a6d-f851-4fd1-9db2-869a0982b511' {
insmod part_msdos
insmod ext2
if [ x$feature_platform_search_hint = xy ]; then
  search --no-floppy --fs-uuid --set=root
380e2a6d-f851-4fd1-9db2-869a0982b511
else
  search --no-floppy --fs-uuid --set=root
380e2a6d-f851-4fd1-9db2-869a0982b511
fi
linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1
initrd /boot/initrd.img-3.16.0-4-686-pae
}



GRUB is content to boot whatever is on the SD (as you observe) so must
be happy enough with the stanza above; there is no problem with booting
there. Actually, GRUB only uses

 linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1

and

 initrd /boot/initrd.img-3.16.0-4-686-pae

to get everything up and running. Without those two lines you would not
be booting.

So try this:
When the GRUB menu displays press the "e" key and remove the

  search --no-floppy --fs-uuid --set=root 380e2a6d-f851-4fd1-9db2-869a0982b511

Then press the "F10" key. Can you still boot? Do you get the onscreen
message you are concerned about?




Something *ODD* is going on in that area.
As that string appears in both the "then" and "else" clauses I:
  1. tried removing them individually leaving a blank line.
 For both cases grub crashed with a syntax error.
  2. removed entire if-then-else-fi construct.
 That booted without the "device not found" message.
  3. changed the last digit of the UUID.
 That demonstrated that the "then" clause was the one executed.
  4. used gparted to change the UUID of the SD card.
 That crashed in ways seemingly unrelated to anything.
 I'm going to update my preferred preseed.cfg and reinstall to
 both /dev/sda1 and the SD card in a reproducible way.

What log files should I save after each test boot.

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/12/2017 02:11 PM, songbird wrote:

Richard Owlett wrote:
...

  ran across http://forums.debian.net/viewtopic.php?f=17=128777
not sure if will help or not, but does point to what i
was thinking (bios, uefi, partition issue or module not
being loaded by grub).


It will take me a while to visually parse that page to make it legible. 
Its miss-use of graphics is a prime example of why I avoid blogs in 
favor of mailing lists and USENET groups. YMMV ;)

Re: Possibly erroneous "device not present" message during boot

[Richard Owlett]

On 04/12/2017 02:03 PM, Pascal Hambourg wrote:

Le 12/04/2017 à 20:33, Richard Owlett a écrit :

On 04/12/2017 12:13 PM, Pascal Hambourg wrote:


[snip]

- while the SD card is inserted, in the GRUB menu press "c" to enter a
GRUB shell and report the result of the "ls" command.


(hd0) (hd0, msdos9) (hd0, msdos8) (hd0, msdos7) (hd0, msdos6)
(hd0, msdos5) (hd0, msdos1) (hd1) (hd1, msdos1)


Some BIOS expose a removable media only when booting from it, so my
theory was that your BIOS did not expose the SD card and GRUB could
not see it. The result of ls proves me wrong.

Can you also report the output of "ls (hd1,msdos1)" to check what GRUB
sees of the partition on the card ?


It gives the partitioning description, UUID, and label of the USB flash 
drive that was plugged in. It is FAT formatted and used to exchange 
information with my Windows machine.





Any suggested reading on the "BASH like shell" I just used?


The GRUB manual at 


It was what prompted me to ask ;/


describes some of the commands, but not all. The shell can display the
command list with "help" and information about a specific command with
"help ".

Re: Possibly erroneous "device not present" message during boot

[songbird]

Richard Owlett wrote:
...

  ran across http://forums.debian.net/viewtopic.php?f=17=128777
not sure if will help or not, but does point to what i
was thinking (bios, uefi, partition issue or module not
being loaded by grub).

...
> Any suggested reading on the "BASH like shell" I just used?
> My initial DuckDuckGo search was *NOT* encouraging ;/
> If retirement is not for learning, what use is it?

  a simple thing like a "help" or "?" command working 
would be great huh?  :)  i ended up at the grub website
trying to figure things out.

  i think hitting tab key once or twice may give full 
list of commands available.  i don't want to reboot right
now to find out...


  songbird

Re: Possibly erroneous "device not present" message during boot

[Brian]

On Wed 12 Apr 2017 at 13:33:02 -0500, Richard Owlett wrote:

> On theory "too much better than too little" I see:
> 
> ### BEGIN /etc/grub.d/20_linux_xen ###
> 
> ### END /etc/grub.d/20_linux_xen ###
> 
> ### BEGIN /etc/grub.d/30_os-prober ###
> menuentry 'Debian GNU/Linux (8.6) (on /dev/mmcblk0p1)' --class gnu-linux
> --class gnu --class os $menuentry_id_option
> 'osprober-gnulinux-simple-380e2a6d-f851-4fd1-9db2-869a0982b511' {
>   insmod part_msdos
>   insmod ext2
>   if [ x$feature_platform_search_hint = xy ]; then
> search --no-floppy --fs-uuid --set=root
> 380e2a6d-f851-4fd1-9db2-869a0982b511
>   else
> search --no-floppy --fs-uuid --set=root
> 380e2a6d-f851-4fd1-9db2-869a0982b511
>   fi
>   linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1
>   initrd /boot/initrd.img-3.16.0-4-686-pae
> }
> 

GRUB is content to boot whatever is on the SD (as you observe) so must
be happy enough with the stanza above; there is no problem with booting
there. Actually, GRUB only uses

 linux /boot/vmlinuz-3.16.0-4-686-pae root=/dev/mmcblk0p1

and

 initrd /boot/initrd.img-3.16.0-4-686-pae

to get everything up and running. Without those two lines you would not
be booting.

So try this:
When the GRUB menu displays press the "e" key and remove the

  search --no-floppy --fs-uuid --set=root 380e2a6d-f851-4fd1-9db2-869a0982b511

Then press the "F10" key. Can you still boot? Do you get the onscreen
message you are concerned about?