When device becomes non-attachable (for example because it gets mounted,
or used as part of LVM/RAID/whatever), it should be removed from
advertised available devices. The code for removing QubesDB entry was
buggy - the device is actually a directory in QubesDB, not a single
entry.
QubesOS/qubes-issues#1600
(cherry picked from commit 22d6892ec9)
There are already some other rules to ignore not interesting devices.
This includes device-mapper assembled in initramfs manually. 'dmroot'
isn't properly detected as mounted because /dev/mapper/dmroot isn't a
symlink to /dev/dm-0 and /proc/mounts contains the former name, while
udev event the later.
FixesQubesOS/qubes-issues#1586
(cherry picked from commit 6ac3fc3247)
By default dracut set 600 mode. While it isn't a problem for starting a
VM (which is done as root), it is a problem for creating standalone VM,
which has it's own copy of kernel directory.
(cherry picked from commit 2f1c8ea459)
Commit https://github.com/QubesOS/qubes-linux-utils/commit/c1d42f1 --
"qfile-unpacker: do not call fdatasync() at each file" fixing
QubesOS/qubes-issues#1257 -- increased the chance of data loss with
qvm-move-to-vm: Say it nominally succeeds, and *deletes* the files from
the source VM. Soon after, the destination VM or the system could crash,
or an external drive hosting ~/QubesIncoming/srcVM could get unplugged
by accident, all before the data had really been persisted to disk.
But reverting the commit (ignoring the performance issue) wouldn't
completely solve this:
"Calling fsync() does not necessarily ensure that the entry in the
directory containing the file has also reached disk. For that an
explicit fsync() on a file descriptor for the directory is also
needed." - fsync(2)
It gets even worse for "slow symlinks" (whose target is too long to be
stored directly in the inode metadata), apparently they can't be synced
at all individually.
So instead, just call syncfs() once after everything has been unpacked:
+ Should prevent all data loss (if fs and disk are well behaved)
+ Allows caching and reordering -> no slowdown with many small files
- Blocks until any unrelated writes on the filesystem finish :\
(cherry picked from commit 4f59b3df6f)
The filesystem hosting ~/QubesIncoming/srcVM/ needs to support O_TMPFILE
too, in addition to the kernel. If it doesn't, take the use_tmpfile = 0
fallback.
(cherry picked from commit 74a1b4cc50)
It is required to prevent deadlocks in single-threaded select-based IO
programs (namely: qrexec). POSIX API doesn't support checking how much
can be written to pipe/socket without blocking, so to prevent blocking
application must use O_NONBLOCK mode, and somehow deal with non-written
data (buffer it).
QubesOS/qubes-issues#1347
(cherry picked from commit 6a44eaeb09)
POSIX requires that a read(2) which can be proved to occur after a
write() has returned returns the new data.
We want here only that other processes in the same VM will see the
file either fully written, or not see it at all. So ensuring that
linkat(2) is called after write is completed should be enough.
FixesQubesOS/qubes-issues#1257
(cherry picked from commit c1d42f1602)
* origin/pr/1:
archlinux: ensure /usr/sbin is not use to follow archlinux packaging guidelines
archlinux: update to match new dependancies and archlinux /usr/lib guidelines
Since migration to QubesDB, it isn't needed anymore (QubesDB have no
problem with concurrent writes, as transactions are not supported).
This should speedup system startup.
This is preparation for pvgrub support, where all VM kernel files will
be installed inside of VM instead of dom0.
But also the same could be used to prepare VM kernel image from any dom0
kernel.
Recently we've switched all xenstore access to the new interface
(instead of deprecated /proc/xen/xenbus). Mostly because of deadlock in
/proc/xen/xenbus implementation.
Marek Marczykowski-Górecki