| Age | Commit message (Collapse) | Author |
|---|
|
This will be useful for making it possible to enable persistent during
user-space.
|
|
Previously LUKS partitions whose fs had the incorrect partition label
would be left open when using persistent-storage=partition.
|
|
|
|
We didn't want this before with live-{rw,sn} and home-{rw,sn} style
persistence, so this was just a left-over. But we still can only allow
one each of root-ov, root-sn and home-sn.
|
|
This will be useful for making it possible to enable persistent during
user-space.
|
|
|
|
|
|
...except that they're broken! But we're getting to that.
|
|
link_files() won't remove symlinks at the destination, which could
cause trouble.
|
|
|
|
This is done by passing the option "linkfiles" to a live.persist
entry. That option is very useful when you only want certain files
to be persistent, not the whole directory they are in, like dot-files
in $home.
|
|
See: https://bugs.busybox.net/show_bug.cgi?id=4520
|
|
|
|
The comma-separated list TYPE... specifies which types of encryption
to allow for persistent media, and whether to allow plaintext media.
The possible TYPEs are the old {none, luks}, with default "none". The
only change is that this allow plaintext and encrypted media to be
used simultaneously.
|
|
Current dependencies are dm-crypt, cryptsetup and askpass.
|
|
|
|
Both overlays and snapshots are now scanned at the same time, and each
device is only mounted once. Passphrases for LUKS volumes are only
asked for once, and unused ones are closed. Also, snapshot files on
encrypted partitions are now supported.
|
|
|
|
|
|
|
|
Since this newline was written to stdout it both messed up the the
display of errors while entering a LUKS passphrase, and caused
find_cow_device to return a bad string under certain
circumstances. Example: You have home-rw on /dev/sdX1. You'll first
be asked for passphrase for "live-rw on /dev/sdX1", which you skip (a
newline has been written to stdout now). Then you're asked for
"home-rw on /dev/sdX1", which is what you want so you enter it.
Thanks to the spurious newline find_cow_device returns "\n/dev/loopY",
which breaks some tests later.
|
|
The comma-separated list TYPE... specifies which types of persistent
storage to use. Current possible TYPEs are {file, filesystem}. For
instance, persistent-storage=filesystem is equivalent to the old
persistent=nofiles, but persistent-storage=file allows to look for
only files, which wasn't possible before. The default is to look for
both, i.e. persistent-storage=filesystem,file.
|
|
|
|
When specified we completely ignore unencrypted media when probing for
persistent media.
|
|
This is the way to control whether to use encrypted persistent media,
instead of the (from this commit on) obsolete persistent=cryptsetup.
|
|
The already duplicated code will be needed a few more times for the upcoming
persistent-media={removable,removable-usb} boot option. Copy/pasting it a few
more times seems the wrong way to go, hence this factorization.
Depending on the place they are used, the newly introduced functions must return
a list of devices named either /sys/block/* or /dev/*. Their first parameter
therefore specifies the wanted output format (default is /dev/* as it is the
most often used).
|
|
This will be needed for the upcoming persistent-media={removable,removable-usb}
boot option.
|
|
consistency with live-config.
|
|
E.g. prefer mounting /dev/sdb1 rather than /dev/sdb.
The isohybrid feature makes both valid devices for being searched for a live
filesystem image. Previous code mounted /dev/sdb on /live/image which makes it
hard to create a second partition on the USB stick.
This brings the need to explicitly check whether the full device is in the
blacklist before even considering its subdevices: the previous blacklist
checking code worked because the full device was always considered before its
subdevices, which is not true anymore.
|
|
Previous blacklist checking was a bit overzealous: e.g. if /dev/sdb1 was
blacklisted (e.g. because it would contain the system-wide persistence COW
device), any other device whose name contains the blacklist one, such as
/dev/sdb11, was also be blacklisted.
|
|
|
|
Previous blacklist checking was a bit overzealous: e.g. if /dev/sdb1 was
blacklisted (e.g. because it would contain the system-wide persistence COW
device), any other device whose name contains the blacklist one, such as
/dev/sdb11, was also be blacklisted.
|
|
|
|
like libata.dma=0.
|
|
|
|
|
|
|
|
(Closes: #565456).
|
|
|
|
not a shell function and causes problems for example within live-snapshot.
|
|
persistent=cryptsetup.
In order to determine if a encrypted partition contains a persistent
partition for live-initramfs, we do have to open it and look into it.
Open means, that the user has to provide the passphase (or key).
A user booting a *default* debian-live system on a machine that has
cryptesetup encrypted partitions should not be bothered by *default*
to enter all his passphrases during the boot process. Also,
encrypted persistency is only usefull when the user knows about it
(= the passphrase to access it). Therefore, it's better to by default
ignore persistency on luks devices.
|
|
Thanks to Rui Miguel Bernardo <rui.bernardo.pt@gmail.com> for the report
and for spotting the bug.
|
|
|
|
|
|
more than one line inwhere_is_mounted();.
|
|
|
|
|
|
|
|
|
|
a path for the persistence files through persistent-path boot parameter (Closes: #512661).
|
