Re: udev applied to SDHC card. Two systems compared.

["Alexander V. Makartsev" <avbetev@gmail.com>]

On 29.12.2021 01:59, peter@easthope.ca wrote:

>     From: "Alexander V. Makartsev" <avbetev@gmail.com>
>     Date: Mon, 27 Dec 2021 22:16:04 +0500
> > It is always better to make udev rules to identify target devices properly.
> >       
> Thanks Alexander.  Definitely my intention.
>
> 'KERNEL=="sd?1"' identifies two volumes, /dev/sda1 and /dev/sdb1.
>
> /dev/sda1 is approximately 6 GB.  'ATTR{size}=="7434240"' 
> distinguishes the SD card from the HDD.  Therefore 'KERNEL=="sd?1", 
> ATTR{size}=="7434240"' should identify the SD volume uniquely.
>     
    My point was, you could've used better or more identifiers to
    distinguish between devices, so there is no other device could ever
    interfere.
    For example, what will happen if you plug in another drive with the
    same size?
    It will be assigned as /dev/sdc1 and will have partition of the same
    size. Your rule will match and as a result will lead to undefined
    behavior.
    It is much better to use identifiers like:
    "ENV{ID_MODEL}" - internal information,
    "ENV{ID_SERIAL_SHORT}" - unique serial number,
    "ENV{ID_FS_UUID}" - unique identifier of filesystem.
    Even if you clone filesystem to another device of same make/model,
    udev still will be able to tell them apart because of unique serial
    number.
    
    Also, there is no information about when exactly "KERNEL"
    and "ATTR{size}" gets populated by udev, so by naming your custom
    rule "99-*.rules"
    you will make it to be processed last, effectively bypassing
    possible race condition between the rules.
    This could be the reason why two systems behave differently.
    
> > ... impossible to answer that question without actually looking at 
> > these systems.
> >       
> Right oh.  "udevadm info" shows that udev identifies the device. The 
> failure isn't in identification.  With configuration paralleling the 
> SD card, a Kingston USB stick works in both systems .  Something 
> peculiar about the SD card. I need to work on that.
>
> Thanks again,                            ... P.
>     
    You can also compare two systems using "udevadm info
    --attribute-walk"
    Systems could have different card-readers based on different
    controllers and different usb hub devices.
    They all could use different kernel modules and trigger different
    set of udev rules.
    
    I'm sorry, if I wasn't clear enough in previous letter. English is
    an every day struggle for me.
    
-- 
With kindest regards, Alexander.

⢀⣴⠾⠻⢶⣦⠀ 
⣾⠁⢠⠒⠀⣿⡁ Debian - The universal operating system
⢿⡄⠘⠷⠚⠋⠀ https://www.debian.org
⠈⠳⣄⠀⠀⠀⠀