Roy Marples [Tue, 9 Aug 2022 16:56:44 +0000 (17:56 +0100)]
Respect DHCPv6 for prior
Roy Marples [Tue, 9 Aug 2022 16:32:49 +0000 (17:32 +0100)]
Amend prior so that NTP servers are removed if the option vanishes
Roy Marples [Tue, 9 Aug 2022 16:26:37 +0000 (17:26 +0100)]
hooks: Support timesyncd
Fixes #106
Roy Marples [Mon, 8 Aug 2022 15:47:55 +0000 (16:47 +0100)]
inet6: Support RFC 4862 5.5.3.e in regards to Valid Lifetime
1. If the received Valid Lifetime is greater than 2 hours or
greater than RemainingLifetime, set the valid lifetime of the
corresponding address to the advertised Valid Lifetime.
2. If RemainingLifetime is less than or equal to 2 hours, ignore
the Prefix Information option with regards to the valid
lifetime.
3. Otherwise, reset the valid lifetime of the corresponding
address to 2 hours.
Roy Marples [Fri, 29 Jul 2022 07:00:09 +0000 (08:00 +0100)]
options: Ensure that we correctly read auth tokens.
Roy Marples [Fri, 29 Jul 2022 06:22:59 +0000 (07:22 +0100)]
dhcpcd-run-hooks.8: its and not it's for posessive.
Thanks to perkelix.
Fixes #110.
Adam Dinwoodie [Fri, 29 Jul 2022 06:14:41 +0000 (07:14 +0100)]
Always skip hooks ending in ~ (#113)
dhcpcd-run-hooks is intended to skip hooks with filenames ending in `~`,
but the test only works if `$skip_hooks` is defined and not empty.
Refactor the test such that files ending in `~` are always skipped, as
appears to be the intent of this code.
Roy Marples [Thu, 2 Jun 2022 05:36:18 +0000 (06:36 +0100)]
DHCP: Fill chaddr if hwlen really is 16 bytes
This isn't normally a problem as ethernet is 6 bytes in length.
Spotted by Steve Noonan.
Roy Marples [Wed, 20 Apr 2022 16:51:53 +0000 (17:51 +0100)]
Linux: Use netlink to workout address flags
There seems to be a race with netlink announcing the address vs
it's availability via /proc.
Fixes #76.
Wu Xiaotian [Thu, 31 Mar 2022 22:46:20 +0000 (06:46 +0800)]
add support for loongarch64 (#93)
Roy Marples [Sat, 12 Mar 2022 11:15:09 +0000 (11:15 +0000)]
duid: Disable use of machine UUID by default.
Fixes #89.
Roy Marples [Thu, 24 Feb 2022 12:53:02 +0000 (12:53 +0000)]
hooks: Use command -v rather than type
command -v is now required by POSIX, type is technically an extension.
Roy Marples [Thu, 24 Feb 2022 06:34:10 +0000 (06:34 +0000)]
Bump dates for prior
Martin-Éric Racine [Thu, 24 Feb 2022 06:28:21 +0000 (06:28 +0000)]
Fixes manual page typos found by Lintian
Roy Marples [Sat, 13 Nov 2021 14:10:31 +0000 (14:10 +0000)]
dhcpcd.8: remove reference to 02-dump hook script
This hook script was removed a while ago.
Fixes #72.
Ben Woods [Sat, 13 Nov 2021 13:53:58 +0000 (21:53 +0800)]
Add epair(4) to the ignored interface types list (#73)
The FreeBSD epair(4) interface type can be thought of as a virtual patch cable.
The epair interfaces come in pairs, with network traffic being passed between the 2 pairs.
The basic intent is to provide connectivity between two virtual network stack instances.
This interface type does not typically require an IP address, as it is passing traffic
between other virtual networks/interfaces which obtain their own IP address.
Therefore, add this interface type to the list that will be ignored by dhcpcd.
Roy Marples [Fri, 12 Nov 2021 14:24:11 +0000 (14:24 +0000)]
privsep: Only set RLIMIT_NOFILES for supported platforms
Basically only for NetBSD, DragonFlyBSD, kqueue or epoll platforms.
All others will use poll(2) and return EINVAL if nfds > RLIMIT_NOFILES.
Trying to work out our nfs limit relies on already listening to
all FD's which may not be the case and was problematic.
Luckily all platforms aside from Solaris have new file creation
sandboxed away aside from accept(2) so this should still be safe.
Veronika Kremneva [Wed, 27 Oct 2021 19:53:22 +0000 (22:53 +0300)]
src/privsep-linux.c: Add support of Synopsys ARC64 architecture (#66)
Signed-off-by: Veronika Kremneva <kremneva@synopsys.com>
Roy Marples [Wed, 20 Oct 2021 10:21:38 +0000 (11:21 +0100)]
privsep: and again on OpenBSD
Roy Marples [Wed, 20 Oct 2021 09:23:19 +0000 (10:23 +0100)]
privsep: Fix compile on Dragonfly BSD
Roy Marples [Mon, 18 Oct 2021 14:25:35 +0000 (15:25 +0100)]
README.md: Remove non existant man page links.
They have not been online for sometime and are unlikely to return.
While here, correct the commit log location.
Fixes #39
Roy Marples [Thu, 14 Oct 2021 05:31:11 +0000 (06:31 +0100)]
DHCP: Set INFORM state only when we are about to inform
As we may not find an address right away.
While here, tidy up and comment code about deriving the address
when none is specified for inform and static setups.
Kaelan Mikowicz [Thu, 14 Oct 2021 05:14:45 +0000 (22:14 -0700)]
DHCP: Fix infinite INFORM messages
When an address is configured on the interface, do not send an
INFORM message if the interface is already bound.
Fixes #48
Philip Paeps [Tue, 12 Oct 2021 08:07:28 +0000 (16:07 +0800)]
privsep: Fix getting interface VLANID on BSD (#58)
Include relevant headers for the definition of SIOCGETVLAN. This fixes
DHCPv6 Prefix Delegation where a prefix is delegated to several VLAN
interfaces on the same physical interface.
Roy Marples [Fri, 1 Oct 2021 15:25:53 +0000 (16:25 +0100)]
DHCP6: Don't spam the log when a RA repeatedly triggers an INFORM
This can occur if the RA reduces the prefix times in accordance
with it's own lifetimes for example.
dhcpcd only checks if the RA contents have changed to trigger a
new INFORM.
As such, only log about new INFORMs.
Fixes #46.
Marian Cingel [Tue, 31 Aug 2021 12:33:20 +0000 (13:33 +0100)]
privsep: Enforce proper alignment of serialized struct cmsghdr
Fixes #43.
Roy Marples [Tue, 31 Aug 2021 09:57:44 +0000 (10:57 +0100)]
DHCP6: Only send FQDN for SOLICIT, REQUEST, RENEW, or REBIND messages.
As per RFC 4704 section 5.
Fixes #44.
Stefan Sperling [Tue, 31 Aug 2021 09:31:15 +0000 (11:31 +0200)]
Fix rt_cmp_dest() for equivalent network prefixes with different netmasks. (#52)
When rt_add() decides that it must delete+add a route in order to change
the routing table entry, a wrong RB tree lookup result can throw it off
the rails. In the case observed, a static /64 prefix was deleted from
vlan1 while dhcpcd intended to delete its reject route bound to lo0.
Given two routes in the table, the loopback reject route installed by
dhcpd for my /48 prefix, and a cloning route for a /64 prefix on vlan1:
2001:db8::/48 ::1 UGR 0 0 32768 56 lo0
2001:db8::/64 2001:db8::1 UCn 1 2 - 4 vlan1
When searching the OS routing table dhcpcd attempts to tell routes apart
based only on the masked destination address. In the above case the masked
destinations look identical. The only difference is the length of the netmask.
The function rt_cmp_dest() didn't detect this and returned the /64 route
while dhcpcd was in fact searching for the /48 route.
This patch fixes the lookup by running rt_cmp_netmask() if the masked
destination comparison via sa_cmp() leaves us with a tie. With this change
dhcpcd deletes the /48 route as intended, and leaves the /64 route alone.
I had to move the rt_cmp_dest() function down since it needs to use the
static helper function rt_cmp_netmask(), which happened to be defined
just below rt_cmp_dest().
Why am I using an overlapping static prefix? The answer is that my ISP
assigns a static /48 prefix but won't route IPv6 unless my router sends
a DHCPv6 request when it connects via PPPoE. I configure static IPv6 subnets
on LAN interfaces and have configured dhcpcd to obtain a /48 prefix lease
without setting addresses on any internal interfaces.
My dhcpcd.conf contains:
ipv6only
noipv6rs
duid
persistent
option rapid_commit
require dhcp_server_identifier
script ""
allowinterfaces pppoe0
interface pppoe0
ia_pd 1 /2001:db8::/48
This problem was found on OpenBSD, in case that matters for reproduction
of the issue.
Peter Fabinski [Tue, 31 Aug 2021 09:29:25 +0000 (05:29 -0400)]
Update man pages for default metric changes (#51)
* Update man pages for metric changes in
3547ed19
* Make metric man changes additive
Craig McQueen [Fri, 20 Aug 2021 07:57:18 +0000 (17:57 +1000)]
Fix parsing of hex/octal escapes in strings (#42)
Roy Marples [Tue, 20 Jul 2021 10:06:32 +0000 (11:06 +0100)]
Merge pull request #38 from concatime/posix-shell
configure: test -a|o is not POSIX
Roy Marples [Mon, 19 Jul 2021 20:25:45 +0000 (21:25 +0100)]
Build as dhcpcd-9.99 now we have a dhcpcd-9 branch.
Roy Marples [Mon, 19 Jul 2021 20:22:44 +0000 (21:22 +0100)]
IPv4LL: Don't remove statically assigned addresses
Fixes #37.
While here, don't open a DHCP port for static addresses either
when not in manager mode.
Issam E. Maghni [Tue, 4 May 2021 16:33:02 +0000 (12:33 -0400)]
configure: test -a|o is not POSIX
Roy Marples [Thu, 18 Mar 2021 12:14:54 +0000 (12:14 +0000)]
FreeBSD: Finish EVFILT_PROCDESC implementation.
Roy Marples [Wed, 17 Mar 2021 20:16:20 +0000 (20:16 +0000)]
Fix more new eloop fallout.
Roy Marples [Wed, 17 Mar 2021 20:13:02 +0000 (20:13 +0000)]
privsep: Implement pdfork(2) for capsicum(4)
Capsicum does not allow waitpid(2).
As such we need to use pdfork(2) and watch for the descriptor to
hangup.
Roy Marples [Wed, 17 Mar 2021 15:19:27 +0000 (15:19 +0000)]
privsep: Fix FreeBSD for prior
Roy Marples [Wed, 17 Mar 2021 15:12:22 +0000 (15:12 +0000)]
privsep: Fix Linux support for prior
Roy Marples [Wed, 17 Mar 2021 15:11:53 +0000 (15:11 +0000)]
privsep: Remove debug from prior
Roy Marples [Wed, 17 Mar 2021 15:03:52 +0000 (15:03 +0000)]
privsep: Implement some process management
This also allows us to wait until all processes have exited to
avoid a fast restart which complains addresses are in use.
Roy Marples [Wed, 17 Mar 2021 14:23:55 +0000 (14:23 +0000)]
privsep: Don't close control proxy on hangup
Instead wait for it to be closed by the master process for a
clean shtudown.
Roy Marples [Wed, 10 Mar 2021 13:09:39 +0000 (13:09 +0000)]
control: Fix working with new eloop
Fabrice Fontaine [Mon, 8 Mar 2021 10:05:16 +0000 (11:05 +0100)]
src/privsep-linux.c: add support for m68k (#29)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Mon, 8 Mar 2021 09:43:50 +0000 (09:43 +0000)]
README.md: remove dead reference to phabricator
Roy Marples [Sat, 20 Feb 2021 16:25:05 +0000 (16:25 +0000)]
control: Cannot write with hangup
Roy Marples [Tue, 16 Feb 2021 16:43:18 +0000 (16:43 +0000)]
eloop: delete events from kqueue/epoll directly
Rather than relying on close(2) being called.
Whilst a bit less performant with many open/close, there is also
no guarantee that close(2) will actually be called as
shutdown(2) could be used instead.
Roy Marples [Tue, 16 Feb 2021 16:40:35 +0000 (16:40 +0000)]
log a diagnostic here
Roy Marples [Tue, 9 Feb 2021 11:36:07 +0000 (11:36 +0000)]
control: Don't log error deleting fds from eloop.
They might not be added as they are inactive and its just noise
as no operation is done anymore even for kqueue or epoll.
Roy Marples [Tue, 9 Feb 2021 10:11:54 +0000 (10:11 +0000)]
BSD: Find the correct interface for tunneled routes
Should disard a harmless diagnostic.
Roy Marples [Mon, 8 Feb 2021 19:25:08 +0000 (19:25 +0000)]
control: GC control deletion
Just free it entirely.
Fabrice Fontaine [Mon, 8 Feb 2021 06:23:54 +0000 (07:23 +0100)]
src/privsep-linux.c: add support for arc (#28)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_{ARCOMPACT,ARCV2} is only defined
since kernel 5.2 and
https://github.com/torvalds/linux/commit/
67f2a8a29311841ba6ab9b0e2d1b8f1e9978cd84
Detection of arc compact and arc v2 have been "copy/pasted" from
https://github.com/wbx-github/uclibc-ng/commit/
afab56958f1cbb47b831ee3ebff231dfbae74af2
Fixes:
- http://autobuild.buildroot.org/results/
d29083700a80dd647621eed06faeeae03f0587d3
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Fabrice Fontaine [Sun, 7 Feb 2021 08:43:14 +0000 (09:43 +0100)]
src/privsep-linux.c: add support for or1k (#27)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_OPENRISC is defined since kernel 3.7:
https://github.com/torvalds/linux/commit/
e2bebb4ae6d9ac4ffc524db67f7ecb205a173f77
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Fri, 5 Feb 2021 22:41:16 +0000 (22:41 +0000)]
Force TOP as we know which directory we are in.
Roy Marples [Fri, 5 Feb 2021 21:56:40 +0000 (21:56 +0000)]
control fd can be both read and write.
Fabrice Fontaine [Fri, 5 Feb 2021 08:25:12 +0000 (09:25 +0100)]
src/privsep-linux: fix build on sparc (#26)
Fix the following build failure:
privsep-linux.c:203: warning: "AUDIT_ARCH_SPARC64" redefined
203 | # define AUDIT_ARCH_SPARC64
|
In file included from privsep-linux.c:35:
/srv/storage/autobuild/run/instance-0/output-1/host/sparc64-buildroot-linux-gnu/sysroot/usr/include/linux/audit.h:392: note: this is the location of the previous definition
392 | #define AUDIT_ARCH_SPARC64 (EM_SPARCV9|__AUDIT_ARCH_64BIT)
|
In file included from privsep-linux.c:36:
privsep-linux.c:215:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
215 | BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
| ^~~~~~~~~~~~~~~~~~
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Fabrice Fontaine [Fri, 5 Feb 2021 08:05:54 +0000 (09:05 +0100)]
src/privsep-linux.c: add support for sh (#25)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_{SH,SHEL,SH64,SHEL64} are defined at
least since kernel 3.7 and
https://github.com/torvalds/linux/commit/
607ca46e97a1b6594b29647d98a32d545c24bdff
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Wed, 3 Feb 2021 20:52:01 +0000 (20:52 +0000)]
Fix tests
Roy Marples [Tue, 2 Feb 2021 17:09:25 +0000 (17:09 +0000)]
Rename Privileged Actioneer to Privileged Proxy
Sadly actioneer is not a real word.
Roy Marples [Tue, 2 Feb 2021 16:54:09 +0000 (16:54 +0000)]
Rename Master to Manager
Roy Marples [Tue, 2 Feb 2021 15:10:20 +0000 (15:10 +0000)]
Fix prior for epoll.
Roy Marples [Tue, 2 Feb 2021 14:41:54 +0000 (14:41 +0000)]
eloop: Make the API more like native poll/kqueue/epoll
Just have the one callback, but return an abstracted event mask
to work out if we can read/write have something else.
Log diagnostics if the event mask is unexpected.
While here add more logging if we fail to register an event to
monitor.
Roy Marples [Sun, 31 Jan 2021 21:12:26 +0000 (21:12 +0000)]
Add __NR_pselect6_time64 to priory
Roy Marples [Sun, 31 Jan 2021 21:07:43 +0000 (21:07 +0000)]
Linux: allow pselect(2) to work in SECCOMP
The default is ppoll(2), but we still allow epoll(7) so
allowing pselect(2) makes perfect sense and allows the user
to pick the polling mechanism they want.
Fabrice Fontaine [Sun, 31 Jan 2021 20:57:40 +0000 (21:57 +0100)]
src/privsep-linux.c: add support for microblaze (#24)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_MICROBLAZE is only defined since
kernel 3.18 and
https://github.com/torvalds/linux/commit/
ce5d112827e5c2e9864323d0efd7ec2a62c6dce0
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Sun, 31 Jan 2021 10:33:21 +0000 (10:33 +0000)]
A belated welcome to 2021
Fabrice Fontaine [Sun, 31 Jan 2021 10:12:21 +0000 (11:12 +0100)]
src/privsep-linux.c: add support for nios2 (#23)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_NIOS2 is only defined since kernel
5.2 and
https://github.com/torvalds/linux/commit/
1660aac45e5b49a5ace29fb5b73254617533fcbd
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Sun, 31 Jan 2021 10:08:11 +0000 (10:08 +0000)]
Typo with prior
Roy Marples [Sun, 31 Jan 2021 07:37:03 +0000 (07:37 +0000)]
Linux: Add support for NDS32BE SECCOMP
Fabrice Fontaine [Sun, 31 Jan 2021 02:35:10 +0000 (03:35 +0100)]
src/privsep-linux.c: add support for xtensa (#22)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_XTENSA is only defined since kernel
5.0 and
https://github.com/torvalds/linux/commit/
98c3115a4ec56f03056efd9295e0fcb4c5c57a85
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Sat, 30 Jan 2021 12:21:15 +0000 (12:21 +0000)]
import-src: Improve target so we don't create empty files
Roy Marples [Sat, 30 Jan 2021 11:46:06 +0000 (11:46 +0000)]
import-src: adjust build target so all .in files are pre-built
While here, if we cannot find ypbind to work out how the hook script
should be configured, guess according to OS.
Fabrice Fontaine [Sat, 30 Jan 2021 11:10:08 +0000 (12:10 +0100)]
src/privsep-linux.c: add support for nds32 (#21)
Fix the following build failure:
privsep-linux.c:206:4: error: #error "Platform does not support seccomp filter yet"
# error "Platform does not support seccomp filter yet"
^~~~~
In file included from privsep-linux.c:36:
privsep-linux.c:213:38: error: 'SECCOMP_AUDIT_ARCH' undeclared here (not in a function); did you mean 'SECCOMP_ALLOW_ARG'?
BPF_JUMP(BPF_JMP + BPF_JEQ + BPF_K, SECCOMP_AUDIT_ARCH, 1, 0),
^~~~~~~~~~~~~~~~~~
It should be noted that AUDIT_ARCH_NDS32 is only defined since kernel
5.2 and
https://github.com/torvalds/linux/commit/
fa562447e154334523daa44c0b60625d71a345f5
Fixes:
- http://autobuild.buildroot.org/results/
af8ba07ea0c12ab8cd24d528ef98db05521f3d36
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Roy Marples [Sat, 30 Jan 2021 11:04:53 +0000 (11:04 +0000)]
eloop: optimise the pselect code so it's not a wrapper for ppoll
This makes the code smaller yet and also use less memory then ppoll!
Still, the API blows chunks and we still have arbitary fd limits
which we'll realistically never hit.
Also, some BSD's note potential issues with select on the same fd
across processes so ppoll is still the winner.
Roy Marples [Sat, 30 Jan 2021 01:37:15 +0000 (01:37 +0000)]
eloop: Default to using ppoll(2) again.
Even though we now have fully working kqueue(2) and epoll(7)
with our privsep code, ppoll(2) is faster and smaller for our
workload.
This time though, we will keep the code here as it's fully working
and while fixing kqueue we also fixed other bugs in dhcpcd as a result
so it's not time wasted at all.
Roy Marples [Fri, 29 Jan 2021 22:10:21 +0000 (01:10 +0300)]
eloop: Fix logic error with epoll reading test
Roy Marples [Fri, 29 Jan 2021 11:13:43 +0000 (11:13 +0000)]
eloop: Always free the pollfd buffer when clearing events
And set the flag to re-create it before the next run.
This allows us to always ensure it's small if the prior run
used more signals and fds.
Roy Marples [Fri, 29 Jan 2021 06:48:17 +0000 (06:48 +0000)]
eloop: fix a memory leak when clearing events
Roy Marples [Fri, 29 Jan 2021 06:46:39 +0000 (06:46 +0000)]
Whitespace
Roy Marples [Thu, 28 Jan 2021 23:25:26 +0000 (23:25 +0000)]
eloop: fixup a compile warning
Roy Marples [Thu, 28 Jan 2021 23:24:49 +0000 (23:24 +0000)]
BSD: Plug a memory leak
Roy Marples [Thu, 28 Jan 2021 17:26:20 +0000 (17:26 +0000)]
Linux: Implement epoll(7) for eloop (again)
eloop allows for O(1) processing of active fd's.
The problems with the prior implementation have now been fixed.
Roy Marples [Thu, 28 Jan 2021 17:26:08 +0000 (17:26 +0000)]
control: Don't log the write db didn't exist when freeing it.
Roy Marples [Thu, 28 Jan 2021 13:17:57 +0000 (13:17 +0000)]
privsep: Restore the poll maxfd + 1 for state engine
It wasn't fixed, it was using kqueue so avoided!
Roy Marples [Thu, 28 Jan 2021 12:51:47 +0000 (12:51 +0000)]
DragonFly: cast away more kqueue compile warnings
Roy Marples [Thu, 28 Jan 2021 12:39:07 +0000 (12:39 +0000)]
FreeBSD: cast away some kqueue warnings
NetBSD uses size_t, FreeBSD uses int for changelist length.
Roy Marples [Thu, 28 Jan 2021 12:13:45 +0000 (12:13 +0000)]
privsep: Don't add 1 to max fd open when limiting resources
It shoud no longer be needed for OpenBSD, Linux and Solaris.
Roy Marples [Thu, 28 Jan 2021 12:13:24 +0000 (12:13 +0000)]
OpenBSD: Fix kqueue detection and one error.
Roy Marples [Thu, 28 Jan 2021 11:54:05 +0000 (11:54 +0000)]
BSD: Implement kqueue(2) for eloop (again)
kqueue allows for O(1) processing of active fd's an a more
robust signal handling method without the need to use global
variables to avoid calling functions during signal delivery.
The problems with the prior implemenation have now been fixed.
Roy Marples [Wed, 27 Jan 2021 14:11:52 +0000 (14:11 +0000)]
privsep: move setting signals to after clearing eloop
Roy Marples [Wed, 27 Jan 2021 13:45:34 +0000 (13:45 +0000)]
Whitespace
Roy Marples [Wed, 27 Jan 2021 13:37:31 +0000 (13:37 +0000)]
privsep: Ensure BPF listener pid is logged in debug
This mirrors the inet listener.
Roy Marples [Wed, 27 Jan 2021 13:33:52 +0000 (13:33 +0000)]
control: log an error if we fail to remove the write callback
Roy Marples [Wed, 27 Jan 2021 13:31:51 +0000 (13:31 +0000)]
Whitespace
Roy Marples [Tue, 26 Jan 2021 17:37:29 +0000 (17:37 +0000)]
control: unlink privileged socket when shutting down without privsep
Otherwise we have a connection refused when dhcpcd starts again
as it sees the stale socket file.
Roy Marples [Sun, 24 Jan 2021 22:53:20 +0000 (22:53 +0000)]
Linux: fix a memory leak when dhcpcd exits or the log is reopened
Roy Marples [Sun, 24 Jan 2021 22:22:25 +0000 (22:22 +0000)]
eloop: Allow eloop to process all fds returned from poll(2)
We do this by ensuring the events list or pollfd struct storage
is not modified during the revent processing.
An event with a fd of -1 means it's been deleted and one without
a pollfd struct reference has been newly added.
This also allows us to count down the number of fd's that
returned a revent so we can break the loop early if possible.
This is a really minor optimisation that at best only applies if
more than one revent is returned via poll(2).
In the case on dhcpcd on NetBSD with privsep, the number of
fd's is really low. And on other platforms or without privsep it's
low also (just not as low).
It's only when you run dhcpcd per interface that the number
of fd's starts to creep upwards as you then need one per address
dhcpcd is monitoring (as well as the ARP listener per IPv4 address
for non NetBSD).
However, I use eloop in other code where this could be a good saving
and dhcpcd is where the master version of this lives!
Roy Marples [Mon, 18 Jan 2021 11:31:05 +0000 (11:31 +0000)]
script: Use rt_proto_add to ensure no duplicate interfaces on OpenBSD
OpenBSD allows matching priorities, so we need to take the interfaces
in the order given to ensure uniqueness.
Roy Marples [Mon, 28 Dec 2020 12:59:20 +0000 (12:59 +0000)]
Release dhcpcd-9.4.0
Roy Marples [Mon, 28 Dec 2020 01:59:57 +0000 (01:59 +0000)]
script: ignore inactive interfaces for ordering