Sometimes, when we try to reply to messages we don't check return
values. This means we might miss a ECONNRESET, and will get a ENOTCONN
on next command. Treat both the same hence.
sd_event_wait() returning 0 usually means that it timed out, which means it must
have been idle. However, sd_event_wait() may return 0 in case an event was triggered
but it turned out there was nothing to do. Make the check for idle explicit to avoid
this edge-case.
Errors like EPERM from the kernel should certainly not be reason to
exit. Let's try to be defensive here, and try to continue on most send
errors, but possibly tell the sender about it.
We don't actually want a by-path/ symlink for MMC RPMB devices, so just add
them to the blacklist. This will prevent creating wrong by-path links and
blkid'ing those.
Linux 3.10+ exposes RPMB (Replay Protected Memory Block) partitions of MMC
devices [1] ; trying to read them with blkid or other unspecific means will
cause kernel buffer I/O errors and timeouts. So don't run blkid on these.
Also ensure that /dev/disk/by-path creates proper symlinks and exposes the
-rpmb partition separately, instead of letting the "normal" partition symlink
point to the rpbm device (this is a race condition).
[1] http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=090d25fe224c0https://launchpad.net/bugs/1333140
include-what-you-use automatically does this and it makes finding
unnecessary harder to spot. The only content of poll.h is a include
of sys/poll.h so should be harmless.
Otherwise systemd-reboot.service will remove the parameter that was set
before. This was broken in commit "b986229efe2cc96157aa14c37bab7843311bbef1
systemctl: bugfix for systemctl reboot command with argument"
This avoids taking the SysV init script enablement state into account if we
have native units. Otherwise systemctl disable on native unit would not
be respected in the presence of an enabled SysV script.
Also, there's no need to do all the parsing and creation of service files if we
already have a native systemd unit for the processed SysV init script.
On Tue, Feb 10, 2015 at 08:10:43PM +0100, Lennart Poettering wrote:
> Hmm, I think it would be nicer to use be32toh() here instead, since it
> ensures the macro is (to a limited degree) typesafe.
>
> Any chance you could rework that?
From: Paul Martin <paul.martin@codethink.co.uk>
Date: Wed, 11 Feb 2015 11:47:16 +0000
Subject: [PATCH] networkd dhcpv4 logging endian fix
On a big-endian host, systemd-networkd prints out IPv4 network
addresses byte reversed:
Feb 10 16:43:32 hostname systemd-networkd[151]: eth0 : DHCPv4 address 158.1.24.10/16 via 1.1.24.10
The address obtained is 10.24.1.158/16 and the route is
10.24.0.0/16 dev eth0 src 10.24.1.187
The macro ADDRESS_FMT_VAL() unpacks a "struct in_addr" in a
little-endian specific manner.
This patch forces the passed address into host order, then unpacks it.
On an x86 later than i486, compiled with -O2, the only extra overhead
is a single bswap instruction.
Lennart Poettering