slavash [Wed, 5 Dec 2018 14:25:03 +0000 (14:25 +0000)]
mlx5en: Implement backpressure indication.
The backpressure indication is implemented using an unlimited rate type of
mbuf send tag. When the upper layers typically the socket layer has obtained such
a tag, it can then query the destination driver queue for the current
amount of space available in the send queue.
A single mbuf send tag may be referenced multiple times and a refcount has been added
to the mlx5e_priv structure to track its usage. Because the send tag resides
in the mlx5e_channel structure, there is no need to wait for refcounts to reach
zero until the mlx4en(4) driver is detached. The channels structure is persistant
during the lifetime of the mlx5en(4) driver it belongs to and can so be accessed
without any need of synchronization.
The mlx5e_snd_tag structure was extended to contain a type field, because there are now
two different tag types which end up in the driver which need to be distinguished.
slavash [Wed, 5 Dec 2018 14:24:33 +0000 (14:24 +0000)]
mlx5en: Improve configuration of HW LRO.
In order to enable HW LRO, both the "hw_lro" sysctl in the mlx5en(4) config
space must be set, and the ifconfig(8) LRO capability must be set. Any other
settings will disable HW LRO.
slavash [Wed, 5 Dec 2018 14:24:02 +0000 (14:24 +0000)]
mlx5en: Count all transmitted and received bytes.
Add counter for all transmitted and received bytes. Currently only all
transmitted and received packets were counted. Fix description of RX LRO
counters while at it.
slavash [Wed, 5 Dec 2018 14:23:31 +0000 (14:23 +0000)]
mlx5en: Statically allocate and free the channel structure(s).
By allocating the worst case size channel structure array
at attach time we can eliminate various NULL checks in the
fast path. And also reduce the chance for use-after-free
issues in the transmit fast path.
This change is also a requirement for implementing
backpressure support.
slavash [Wed, 5 Dec 2018 14:23:01 +0000 (14:23 +0000)]
mlx5en: Fix race in mlx5e_ethtool_debug_stats().
Writing to the debug stats variable must be locked,
else serialization will be lost which might cause
various kernel panics due to creating and destroying
sysctls out of order.
Make sure the sysctl context is initialized after freeing
the sysctl nodes, else they can be freed twice.
slavash [Wed, 5 Dec 2018 14:21:59 +0000 (14:21 +0000)]
mlx5en: Don't set rate on SQs when the SQ is already stopped.
This can happen when connections are short lived and leads to
a firmware error printout in dmesg, syndrome 0x51cfb0, because
the SQ is in the wrong state.
slavash [Wed, 5 Dec 2018 14:21:28 +0000 (14:21 +0000)]
mlx5en: Fix for inlining issues in transmit path
1) Don't exceed the drivers own hardcoded TX inline limit.
The blueflame register size can be much greater than the hardcoded limit
for inlining. Make sure we don't exceed the drivers own limit, because this
also means that the maximum number of TX fragments becomes invalid and
then memory size assumptions in the TX path no longer hold up.
2) Make sure the mlx5_query_min_inline() function returns an error code.
3) Header inlining is required when using TSO.
4) Catch failure to compute inline header size for TSO.
5) Add support for UDP when computing inline header size.
6) Fix for inlining issues with regards to DSCP.
Make sure we inline 4 bytes beyond the ethernet and/or
VLAN header to workaround a hardware bug extracting
the DSCP field from the IPv4/v6 header.
slavash [Wed, 5 Dec 2018 14:20:57 +0000 (14:20 +0000)]
mlx5en: Remove the DRBR and associated logic in the transmit path.
The hardware queues are deep enough currently and using the DRBR and associated
callbacks only leads to more task switching in the TX path. The is also a race
setting the queue_state which can lead to hung TX rings.
slavash [Wed, 5 Dec 2018 14:20:26 +0000 (14:20 +0000)]
mlx5en: Implement support for bandwidth limiting in by ratio, ETS.
Add support for setting the bandwidth limit as a ratio rather than in bits per
second. The ratio must be an integer number between 1 and 100 inclusivly.
Implement the needed firmware commands and SYSCTLs through mlx5en(4).
slavash [Wed, 5 Dec 2018 13:48:39 +0000 (13:48 +0000)]
mlx5ib: Make sure the congestion work timer does not escape the drain procedure.
If the mlx5_ib_read_cong_stats() function was running when mlx5ib was unloaded,
because this function unconditionally restarts the timer, the timer can still
be pending after the delayed work has been cancelled. To fix this simply loop
on the delayed work cancel procedure as long as it returns non-zero.
slavash [Wed, 5 Dec 2018 13:47:41 +0000 (13:47 +0000)]
mlx5ib: Fix sign extension in mlx5_ib_query_device
"fw_rev_min(dev->mdev)" with type "unsigned short" (16 bits, unsigned) is
promoted in "fw_rev_min(dev->mdev) << 16" to type "int" (32 bits, signed), then
sign-extended to type "unsigned long" (64 bits, unsigned). If
"fw_rev_min(dev->mdev) << 16" is greater than 0x7FFFFFFF, the upper bits of the
result will all be 1.
slavash [Wed, 5 Dec 2018 13:46:39 +0000 (13:46 +0000)]
mlx5: Fixes to allow command polling mode to exist alongside event mode.
A command is either polling or event driven and the mode cannot change
during execution of a command. Make sure the event handler only handle
commands which are not polled. This is done by checking the command mode
in the command handler before completing commands.
slavash [Wed, 5 Dec 2018 13:45:37 +0000 (13:45 +0000)]
mlx5: Add software tx_jumbo_packets counter
This counter will represent transmitted packets which has more than
1518 octets.
The NIC has multiple hardware counters for counting transmitted
packets larger than 1518 octets. Each counter counts the packets
in specific range.
We accumulate those counters to have a single counter.
slavash [Wed, 5 Dec 2018 13:43:37 +0000 (13:43 +0000)]
mlx5: Fix use-after-free in self-healing flow
When the mlx5 health mechanism detects a problem while the driver
is in the middle of init_one or remove_one, the driver needs to prevent
the health mechanism from scheduling future work; if future work
is scheduled, there is a problem with use-after-free: the system WQ
tries to run the work item (which has been freed) at the scheduled
future time.
Prevent this by disabling work item scheduling in the health mechanism
when the driver is in the middle of init_one() or remove_one().
slavash [Wed, 5 Dec 2018 13:40:36 +0000 (13:40 +0000)]
mlx5: Raise fatal IB event when sys error occurs
All other mlx5_events report the port number as 1 based, which is how FW
reports it in the port event EQE. Reporting 0 for this event causes
mlx5_ib to not raise a fatal event notification to registered clients
due to a seemingly invalid port.
All switch cases in mlx5_ib_event that go through the port check are
supposed to set the port now, so just do it once at variable
declaration.
slavash [Wed, 5 Dec 2018 13:39:35 +0000 (13:39 +0000)]
mlx4en: Optimise reception of small packets.
Copy small packets like TCP ACKs into a new mbuf
reusing the existing mbuf to receive a new ethernet
frame. This avoids wasting buffer space for
small sized packets.
slavash [Wed, 5 Dec 2018 13:31:45 +0000 (13:31 +0000)]
mlx4en: Remove the DRBR and associated logic in the transmit path.
The hardware queues are deep enough currently and using the DRBR and associated
callbacks only leads to more task switching in the TX path. The is also a race
setting the queue_state which can lead to hung TX rings.
slavash [Wed, 5 Dec 2018 13:30:48 +0000 (13:30 +0000)]
mlx4: Add board identifier and firmware version to sysctl
In last mlx4 update (r325841) we lost the sysctl to show the
firmware version for mlx4 devices.
Add both board identifier and firmware version under:
sys.device.mlx4_core0.hw sysctl node.
slavash [Wed, 5 Dec 2018 13:27:17 +0000 (13:27 +0000)]
ipoib: Notify on modify QP failure only when relevant
Modify QP can fail and it can be acceptable, like when moving from RST to
ERR state, all the rest are not acceptable and a message to the log
should be printed.
The current code prints on all failures and many messages like:
"Failed to modify QP to ERROR state" appear, even when supported by the
state machine of the QP object.
slavash [Wed, 5 Dec 2018 13:26:47 +0000 (13:26 +0000)]
ipoib: increase the non-cm queue length
When a packet needs fragmentation, it might generate more than 3 fragments.
With the queue length 3, all fragments are generated faster than the
queue is drained, which effectively drops fourth and later fragments on
the floor.
slavash [Wed, 5 Dec 2018 13:26:17 +0000 (13:26 +0000)]
ipoib: Don't do a light flush when MTU is unchanged.
When changing the MTU of ibX network interfaces, check that the MTU was really
changed before requesting an update of the multicast rules. Else we might go
into an infinite loop joining and leaving ibX multicast groups towards the
opensm master interface.
slavash [Wed, 5 Dec 2018 13:25:47 +0000 (13:25 +0000)]
ipoib: correct setting MTU from inside ipoib(4).
It is not enough to set ifnet->if_mtu to change the interface MTU.
System saves the MTU for route in the radix tree, and route cache keeps
the interface MTU as well. Since addition of the multicast group causes
recalculation of MTU, even bringing the interface up changes MTU from
4042 to 1500, which makes the system configuration inconsistent. Worse,
ip_output() prefers route MTU over interface MTU, so large packets are
not fragmented and dropped on floor.
Fix it for ipoib(4) using the same approach (or hack) as was applied
for it_tun/if_tap in r339012. Thanks to bz@ for giving the hint.
slavash [Wed, 5 Dec 2018 13:25:13 +0000 (13:25 +0000)]
ibcore: Fix clearing of bound device interface.
Binding to a loopback device is not allowed. Make sure the destination
device address is global by clearing the bound device interface.
Only do this conditionally, else link local addresses won't work.
slavash [Wed, 5 Dec 2018 13:23:14 +0000 (13:23 +0000)]
ibcore: Fix loopback with rdma-cm.
Trying to validate loopback fails because rtalloc1() resolves system
local addresses to the loopback network interface, lo0. Fix this by
explicitly checking for loopback during validation of the source
and destination network address. If the source address belongs to
a local network interface and is equal to the destination address,
there is no need to run the destination address through rtalloc1().
slavash [Wed, 5 Dec 2018 13:20:51 +0000 (13:20 +0000)]
ibcore: Fix an array index check
The array ib_mad_mgmt_class_table.method_table has MAX_MGMT_CLASS
(80) elements. Hence compare the array index with that value instead
of with IB_MGMT_MAX_METHODS (128). This patch avoids that Coverity
reports the following:
Overrunning array class->method_table of 80 8-byte elements at element index 127
(byte offset 1016) using index convert_mgmt_class(mad_hdr->mgmt_class)
(which evaluates to 127).
slavash [Wed, 5 Dec 2018 13:19:21 +0000 (13:19 +0000)]
ibcore: Don't access invalid port.
The port number in the listen_id_priv has been observed to be zero which
means no port has been selected. The current code lacks a check for invalid
port number.
Changelist:
- Replace netmap passthrough host support with a more general
mechanism to call TXSYNC/RXSYNC from an in-kernel event-loop.
No kernel threads are used to use this feature: the application
is required to spawn a thread (or a process) and issue a
SYNC_KLOOP_START (NIOCCTRL) command in the thread body. The
kernel loop is executed by the ioctl implementation, which returns
to userspace only when a different thread calls SYNC_KLOOP_STOP
or the netmap file descriptor is closed.
- Update the if_ptnet driver to cope with the new data structures,
and prune all the obsolete ptnetmap code.
- Add support for "null" netmap ports, useful to allocate netmap_if,
netmap_ring and netmap buffers to be used by specialized applications
(e.g. hypervisors). TXSYNC/RXSYNC on these ports have no effect.
- Various fixes and code refactoring.
Sponsored by: Sunny Valley Networks
Differential Revision: https://reviews.freebsd.org/D18015
arichardson [Wed, 5 Dec 2018 10:57:57 +0000 (10:57 +0000)]
Fix newvers.sh with BUILD_WITH_STRICT_TMPPATH=1
newvers.sh runs mkfifo which did not exist before this change.
However, I didn't notice before because it is run from a function
where a missing command does cause a noticeable failure.
mmel [Wed, 5 Dec 2018 10:30:53 +0000 (10:30 +0000)]
Tidy up arm64 reloc_jmpslots() implementation.
- don't relocate jump slots multiple times (if LD_BIND_NOW is defined).
- process only R_AARCH64_JUMP_SLOT here, other relocation types are handled
by reloc_plt().
brooks [Wed, 5 Dec 2018 00:46:09 +0000 (00:46 +0000)]
Remove MD __sys_* private symbols.
No references to any of these exist in the tree. The list was also
erratic with different architectures exporting different things
(arm64 and riscv exported none).
vangyzen [Tue, 4 Dec 2018 23:46:43 +0000 (23:46 +0000)]
altq: remove ALTQ3_COMPAT code
This code has apparently never compiled on FreeBSD since its
introduction in 2004 (r130365). It has certainly not compiled
since 2006, when r164033 added #elsif [sic] preprocessor directives.
The code was left in the tree to reduce the diff from upstream (KAME).
Since that upstream is no longer relevant, remove the long-dead code.
This commit is the direct result of:
unifdef -m -UALTQ3_COMPAT sys/net/altq/*
A later commit will do some manual cleanup.
I do not plan to MFC this. If that would help you, go for it.
sobomax [Tue, 4 Dec 2018 21:48:56 +0000 (21:48 +0000)]
Another attempt to fix issue with the DIOCGDELETE ioctl(2) not
handling slightly out-of-bound requests properly (r340187).
Perform range check here rather then rely on g_delete_data() to DTRT.
The g_delete_data() would always return success for requests
starting just the next byte after providers media boundary.
gordon [Tue, 4 Dec 2018 18:28:25 +0000 (18:28 +0000)]
Always treat firmware request and response sizes as unsigned.
This fixes an incomplete bounds check on the guest-supplied request
size where a very large request size could be interpreted as a negative
value and not be caught by the bounds check.
Submitted by: jhb
Reported by: Reno Robert
Approved by: so
Security: FreeBSD-SA-18:14.bhyve
Security: CVE-2018-17160
brooks [Tue, 4 Dec 2018 16:48:47 +0000 (16:48 +0000)]
Normalize COMPAT_43 syscall declarations.
Have ogetkerninfo, ogetpagesize, ogethostname, osethostname, and oaccept
declare o<foo>_args structs rather than non-compat ones. Due to a
failure to use NOARGS in most cases this adds only one new declaration.
No changes required in freebsd32 as only ogetpagesize() is implemented
and it has a 32-bit specific implementation.
ian [Tue, 4 Dec 2018 16:43:50 +0000 (16:43 +0000)]
Fix args cross-threading between gptboot(8) and loader(8) with zfs support.
When loader(8) is built with zfs support enabled, it assumes that any extarg
data present is a zfs_boot_args struct, but if the first-stage loader was
gptboot(8) the extarg data is actually a geli_boot_args struct. Luckily,
zfsboot(8) and gptzfsboot(8) have always passed KARGS_FLAGS_ZFS along with
KARGS_FLAGS_EXTARG, so we can use KARGS_FLAGS_ZFS to decide whether the
extarg data is a zfs_boot_args struct.
To avoid similar problems in the future, gptboot(8) now passes a new
KARGS_FLAGS_GELI to indicate that extarg data is geli_boot_args. In
loader(8), if the neither KARGS_FLAGS_ZFS nor KARGS_FLAGS_GELI is set but
extarg data is present (which will be the case for gptboot compiled before
this change), we now check for the known size of the geli_boot_args struct
passed by the older versions of gptboot as a way of confirming what type of
extarg data is present.
In a semi-related tidying up, since loader's main() has already decided
what type of extarg data is present and set the global 'zargs' var
accordingly, don't repeat the check in extract_currdev, just check whether
zargs is NULL or not.
X-MFC after: a few days, along with prior related changes.
Add ability to request listing and deleting only for dynamic states.
This can be useful, when net.inet.ip.fw.dyn_keep_states is enabled, but
after rules reloading some state must be deleted. Added new flag '-D'
for such purpose.
Retire '-e' flag, since there can not be expired states in the meaning
that this flag historically had.
Also add "verbose" mode for listing of dynamic states, it can be enabled
with '-v' flag and adds additional information to states list. This can
be useful for debugging.
Reimplement how net.inet.ip.fw.dyn_keep_states works.
Turning on of this feature allows to keep dynamic states when parent
rule is deleted. But it works only when the default rule is
"allow from any to any".
Now when rule with dynamic opcode is going to be deleted, and
net.inet.ip.fw.dyn_keep_states is enabled, existing states will reference
named objects corresponding to this rule, and also reference the rule.
And when ipfw_dyn_lookup_state() will find state for deleted parent rule,
it will return the pointer to the deleted rule, that is still valid.
This implementation doesn't support O_LIMIT_PARENT rules.
The refcnt field was added to struct ip_fw to keep reference, also
next pointer added to be able iterate rules and not damage the content
when deleted rules are chained.
Named objects are referenced only when states are going to be deleted to
be able reuse kidx of named objects when new parent rules will be
installed.
ipfw_dyn_get_count() function was modified and now it also looks into
dynamic states and constructs maps of existing named objects. This is
needed to correctly export orphaned states into userland.
ipfw_free_rule() was changed to be global, since now dynamic state can
free rule, when it is expired and references counters becomes 1.
External actions subsystem also modified, since external actions can be
deregisterd and instances can be destroyed. In these cases deleted rules,
that are referenced by orphaned states, must be modified to prevent access
to freed memory. ipfw_dyn_reset_eaction(), ipfw_reset_eaction_instance()
functions added for these purposes.