Rozdział 11. Configuration and Tuning

Unfortunately, there are still many vendors that do not provide schematics for their drivers to the open source community because they regard such information as trade secrets. Consequently, the developers of FreeBSD and other operating systems are left with two choices: develop the drivers by a long and pain-staking process of reverse engineering or using the existing driver binaries available for Microsoft® Windows® platforms.

FreeBSD provides "native" support for the Network Driver Interface Specification (NDIS). It includes ndisgen(8) which can be used to convert a Windows® XP driver into a format that can be used on FreeBSD. Because the ndis(4) driver uses a Windows® XP binary, it only runs on i386™ and amd64 systems. PCI, CardBus, PCMCIA, and USB devices are supported.

To use ndisgen(8) , three things are needed:

Download the .SYS and .INF files for the specific NIC. Generally, these can be found on the driver CD or at the vendor’s website. The following examples use W32DRIVER.SYS and W32DRIVER.INF .

The driver bit width must match the version of FreeBSD. For FreeBSD/i386, use a Windows® 32-bit driver. For FreeBSD/amd64, a Windows® 64-bit driver is needed.

The next step is to compile the driver binary into a loadable kernel module. As root , use ndisgen(8) :

This command is interactive and prompts for any extra information it requires. A new kernel module will be generated in the current directory. Use kldload(8) to load the new module:

In addition to the generated kernel module, the ndis.ko and if_ndis.ko modules must be loaded. This should happen automatically when any module that depends on ndis(4) is loaded. If not, load them manually, using the following commands:

The first command loads the ndis(4) miniport driver wrapper and the second loads the generated NIC driver.

Check dmesg(8) to see if there were any load errors. If all went well, the output should be similar to the following:

From here, ndis0 can be configured like any other NIC.

To configure the system to load the ndis(4) modules at boot time, copy the generated module, W32DRIVER_SYS.ko , to /boot/modules . Then, add the following line to /boot/loader.conf :

To verify that an Ethernet card is configured correctly, ping(8) the interface itself, and then ping(8) another machine on the LAN:

To test network resolution, use the host name instead of the IP address. If there is no DNS server on the network, /etc/hosts must first be configured. To this purpose, edit /etc/hosts to add the names and IP addresses of the hosts on the LAN, if they are not already there. For more information, refer to hosts(5) and to /usr/shared/examples/etc/hosts .

When troubleshooting hardware and software configurations, check the simple things first. Is the network cable plugged in? Are the network services properly configured? Is the firewall configured correctly? Is the NIC supported by FreeBSD? Before sending a bug report, always check the Hardware Notes, update the version of FreeBSD to the latest STABLE version, check the mailing list archives, and search the Internet.

If the card works, yet performance is poor, read through tuning(7) . Also, check the network configuration as incorrect network settings can cause slow connections.

Some users experience one or two device timeout messages, which is normal for some cards. If they continue, or are bothersome, determine if the device is conflicting with another device. Double check the cable connections. Consider trying another card.

To resolve watchdog timeout errors, first check the network cable. Many cards require a PCI slot which supports bus mastering. On some old motherboards, only one PCI slot allows it, usually slot 0. Check the NIC and the motherboard documentation to determine if that may be the problem.

No route to host messages occur if the system is unable to route a packet to the destination host. This can happen if no default route is specified or if a cable is unplugged. Check the output of netstat -rn and make sure there is a valid route to the host. If there is not, read “Gateways and Routes” .

ping: sendto: Permission denied error messages are often caused by a misconfigured firewall. If a firewall is enabled on FreeBSD but no rules have been defined, the default policy is to deny all traffic, even ping(8) . Refer to Firewalls for more information.

Sometimes performance of the card is poor or below average. In these cases, try setting the media selection mode from autoselect to the correct media selection. While this works for most hardware, it may or may not resolve the issue. Again, check all the network settings, and refer to tuning(7) .

A log server is a system that has been configured to accept logging information from other hosts. Before configuring a log server, check the following:

On the log server, edit /etc/syslog.conf to specify the name of the client to receive log entries from, the logging facility to be used, and the name of the log to store the host’s log entries. This example adds the hostname of B , logs all facilities, and stores the log entries in /var/log/logclient.log .

When adding multiple log clients, add a similar two-line entry for each client. More information about the available facilities may be found in syslog.conf(5) .

Next, configure /etc/rc.conf :

The first entry starts syslogd at system boot. The second entry allows log entries from the specified client. The -v -v increases the verbosity of logged messages. This is useful for tweaking facilities as administrators are able to see what type of messages are being logged under each facility.

Multiple -a options may be specified to allow logging from multiple clients. IP addresses and whole netblocks may also be specified. Refer to syslogd(8) for a full list of possible options.

Finally, create the log file:

At this point, syslogd should be restarted and verified:

If a PID is returned, the server restarted successfully, and client configuration can begin. If the server did not restart, consult /var/log/messages for the error.

A logging client sends log entries to a logging server on the network. The client also keeps a local copy of its own logs.

Once a logging server has been configured, edit /etc/rc.conf on the logging client:

The first entry enables syslogd on boot up. The second entry prevents logs from being accepted by this client from other hosts ( -s ) and increases the verbosity of logged messages.

Next, define the logging server in the client’s /etc/syslog.conf . In this example, all logged facilities are sent to a remote system, denoted by the @ symbol, with the specified hostname:

After saving the edit, restart syslogd for the changes to take effect:

To test that log messages are being sent across the network, use logger(1) on the client to send a message to syslogd:

This message should now exist both in /var/log/messages on the client and /var/log/logclient.log on the log server.

If no messages are being received on the log server, the cause is most likely a network connectivity issue, a hostname resolution issue, or a typo in a configuration file. To isolate the cause, ensure that both the logging server and the logging client are able to ping each other using the hostname specified in their /etc/rc.conf . If this fails, check the network cabling, the firewall ruleset, and the hostname entries in the DNS server or /etc/hosts on both the logging server and clients. Repeat until the ping is successful from both hosts.

If the ping succeeds on both hosts but log messages are still not being received, temporarily increase logging verbosity to narrow down the configuration issue. In the following example, /var/log/logclient.log on the logging server is empty and /var/log/messages on the logging client does not indicate a reason for the failure. To increase debugging output, edit the syslogd_flags entry on the logging server and issue a restart:

Debugging data similar to the following will flash on the console immediately after the restart:

In this example, the log messages are being rejected due to a typo which results in a hostname mismatch. The client’s hostname should be logclient , not logclien . Fix the typo, issue a restart, and verify the results:

At this point, the messages are being properly received and placed in the correct file.

As with any network service, security requirements should be considered before implementing a logging server. Log files may contain sensitive data about services enabled on the local host, user accounts, and configuration data. Network data sent from the client to the server will not be encrypted or password protected. If a need for encryption exists, consider using security/stunnel , which will transmit the logging data over an encrypted tunnel.

Local security is also an issue. Log files are not encrypted during use or after log rotation. Local users may access log files to gain additional insight into system configuration. Setting proper permissions on log files is critical. The built-in log rotator, newsyslog, supports setting permissions on newly created and rotated log files. Setting log files to mode 600 should prevent unwanted access by local users. Refer to newsyslog.conf(5) for additional information.

How a FreeBSD system accesses the Internet Domain Name System (DNS) is controlled by resolv.conf(5) .

The most common entries to /etc/resolv.conf are:

A typical /etc/resolv.conf looks like this:

When using DHCP, dhclient(8) usually rewrites /etc/resolv.conf with information received from the DHCP server.

/etc/hosts is a simple text database which works in conjunction with DNS and NIS to provide host name to IP address mappings. Entries for local computers connected via a LAN can be added to this file for simplistic naming purposes instead of setting up a named(8) server. Additionally, /etc/hosts can be used to provide a local record of Internet names, reducing the need to query external DNS servers for commonly accessed names.

The format of /etc/hosts is as follows:

For example:

Consult hosts(5) for more information.

The vfs.vmiodirenable sysctl(8) variable may be set to either 0 (off) or 1 (on). It is set to 1 by default. This variable controls how directories are cached by the system. Most directories are small, using just a single fragment (typically 1 K) in the file system and typically 512 bytes in the buffer cache. With this variable turned off, the buffer cache will only cache a fixed number of directories, even if the system has a huge amount of memory. When turned on, this sysctl(8) allows the buffer cache to use the VM page cache to cache the directories, making all the memory available for caching directories. However, the minimum in-core memory used to cache a directory is the physical page size (typically 4 K) rather than 512 bytes. Keeping this option enabled is recommended if the system is running any services which manipulate large numbers of files. Such services can include web caches, large mail systems, and news systems. Keeping this option on will generally not reduce performance, even with the wasted memory, but one should experiment to find out.

The vfs.write_behind sysctl(8) variable defaults to 1 (on). This tells the file system to issue media writes as full clusters are collected, which typically occurs when writing large sequential files. This avoids saturating the buffer cache with dirty buffers when it would not benefit I/O performance. However, this may stall processes and under certain circumstances should be turned off.

The vfs.hirunningspace sysctl(8) variable determines how much outstanding write I/O may be queued to disk controllers system-wide at any given instance. The default is usually sufficient, but on machines with many disks, try bumping it up to four or five megabytes . Setting too high a value which exceeds the buffer cache’s write threshold can lead to bad clustering performance. Do not set this value arbitrarily high as higher write values may add latency to reads occurring at the same time.

There are various other buffer cache and VM page cache related sysctl(8) values. Modifying these values is not recommended as the VM system does a good job of automatically tuning itself.

The vm.swap_idle_enabled sysctl(8) variable is useful in large multi-user systems with many active login users and lots of idle processes. Such systems tend to generate continuous pressure on free memory reserves. Turning this feature on and tweaking the swapout hysteresis (in idle seconds) via vm.swap_idle_threshold1 and vm.swap_idle_threshold2 depresses the priority of memory pages associated with idle processes more quickly then the normal pageout algorithm. This gives a helping hand to the pageout daemon. Only turn this option on if needed, because the tradeoff is essentially pre-page memory sooner rather than later which eats more swap and disk bandwidth. In a small system this option will have a determinable effect, but in a large system that is already doing moderate paging, this option allows the VM system to stage whole processes into and out of memory easily.

Turning off IDE write caching reduces write bandwidth to IDE disks, but may sometimes be necessary due to data consistency issues introduced by hard drive vendors. The problem is that some IDE drives lie about when a write completes. With IDE write caching turned on, IDE hard drives write data to disk out of order and will sometimes delay writing some blocks indefinitely when under heavy disk load. A crash or power failure may cause serious file system corruption. Check the default on the system by observing the hw.ata.wc sysctl(8) variable. If IDE write caching is turned off, one can set this read-only variable to 1 in /boot/loader.conf in order to enable it at boot time.

For more information, refer to ata(4) .

The SCSI_DELAY kernel configuration option may be used to reduce system boot times. The defaults are fairly high and can be responsible for 15 seconds of delay in the boot process. Reducing it to 5 seconds usually works with modern drives. The kern.cam.scsi_delay boot time tunable should be used. The tunable and kernel configuration option accept values in terms of milliseconds and not seconds .

Meta-data updates are updates to non-content data like inodes or directories. There are two traditional approaches to writing a file system’s meta-data back to disk.

Historically, the default behavior was to write out meta-data updates synchronously. If a directory changed, the system waited until the change was actually written to disk. The file data buffers (file contents) were passed through the buffer cache and backed up to disk later on asynchronously. The advantage of this implementation is that it operates safely. If there is a failure during an update, meta-data is always in a consistent state. A file is either created completely or not at all. If the data blocks of a file did not find their way out of the buffer cache onto the disk by the time of the crash, fsck(8) recognizes this and repairs the file system by setting the file length to 0 . Additionally, the implementation is clear and simple. The disadvantage is that meta-data changes are slow. For example, rm -r touches all the files in a directory sequentially, but each directory change will be written synchronously to the disk. This includes updates to the directory itself, to the inode table, and possibly to indirect blocks allocated by the file. Similar considerations apply for unrolling large hierarchies using tar -x .

The second approach is to use asynchronous meta-data updates. This is the default for a UFS file system mounted with mount -o async . Since all meta-data updates are also passed through the buffer cache, they will be intermixed with the updates of the file content data. The advantage of this implementation is there is no need to wait until each meta-data update has been written to disk, so all operations which cause huge amounts of meta-data updates work much faster than in the synchronous case. This implementation is still clear and simple, so there is a low risk for bugs creeping into the code. The disadvantage is that there is no guarantee for a consistent state of the file system. If there is a failure during an operation that updated large amounts of meta-data, like a power failure or someone pressing the reset button, the file system will be left in an unpredictable state. There is no opportunity to examine the state of the file system when the system comes up again as the data blocks of a file could already have been written to the disk while the updates of the inode table or the associated directory were not. It is impossible to implement a fsck(8) which is able to clean up the resulting chaos because the necessary information is not available on the disk. If the file system has been damaged beyond repair, the only choice is to reformat it and restore from backup.

The usual solution for this problem is to implement dirty region logging , which is also referred to as journaling . Meta-data updates are still written synchronously, but only into a small region of the disk. Later on, they are moved to their proper location. Because the logging area is a small, contiguous region on the disk, there are no long distances for the disk heads to move, even during heavy operations, so these operations are quicker than synchronous updates. Additionally, the complexity of the implementation is limited, so the risk of bugs being present is low. A disadvantage is that all meta-data is written twice, once into the logging region and once to the proper location, so performance "pessimization" might result. On the other hand, in case of a crash, all pending meta-data operations can be either quickly rolled back or completed from the logging area after the system comes up again, resulting in a fast file system startup.

Kirk McKusick, the developer of Berkeley FFS, solved this problem with Soft Updates. All pending meta-data updates are kept in memory and written out to disk in a sorted sequence ("ordered meta-data updates"). This has the effect that, in case of heavy meta-data operations, later updates to an item "catch" the earlier ones which are still in memory and have not already been written to disk. All operations are generally performed in memory before the update is written to disk and the data blocks are sorted according to their position so that they will not be on the disk ahead of their meta-data. If the system crashes, an implicit "log rewind" causes all operations which were not written to the disk appear as if they never happened. A consistent file system state is maintained that appears to be the one of 30 to 60 seconds earlier. The algorithm used guarantees that all resources in use are marked as such in their blocks and inodes. After a crash, the only resource allocation error that occurs is that resources are marked as "used" which are actually "free". fsck(8) recognizes this situation, and frees the resources that are no longer used. It is safe to ignore the dirty state of the file system after a crash by forcibly mounting it with mount -f . In order to free resources that may be unused, fsck(8) needs to be run at a later time. This is the idea behind the background fsck(8) : at system startup time, only a snapshot of the file system is recorded and fsck(8) is run afterwards. All file systems can then be mounted "dirty", so the system startup proceeds in multi-user mode. Then, background fsck(8) is scheduled for all file systems where this is required, to free resources that may be unused. File systems that do not use Soft Updates still need the usual foreground fsck(8) .

The advantage is that meta-data operations are nearly as fast as asynchronous updates and are faster than logging , which has to write the meta-data twice. The disadvantages are the complexity of the code, a higher memory consumption, and some idiosyncrasies. After a crash, the state of the file system appears to be somewhat "older". In situations where the standard synchronous approach would have caused some zero-length files to remain after the fsck(8) , these files do not exist at all with Soft Updates because neither the meta-data nor the file contents have been written to disk. Disk space is not released until the updates have been written to disk, which may take place some time after running rm(1) . This may cause problems when installing large amounts of data on a file system that does not have enough free space to hold all the files twice.

The kern.maxfiles sysctl(8) variable can be raised or lowered based upon system requirements. This variable indicates the maximum number of file descriptors on the system. When the file descriptor table is full, file: table is full will show up repeatedly in the system message buffer, which can be viewed using dmesg(8) .

Each open file, socket, or fifo uses one file descriptor. A large-scale production server may easily require many thousands of file descriptors, depending on the kind and number of services running concurrently.

In older FreeBSD releases, the default value of kern.maxfiles is derived from maxusers in the kernel configuration file. kern.maxfiles grows proportionally to the value of maxusers . When compiling a custom kernel, consider setting this kernel configuration option according to the use of the system. From this number, the kernel is given most of its pre-defined limits. Even though a production machine may not have 256 concurrent users, the resources needed may be similar to a high-scale web server.

The read-only sysctl(8) variable kern.maxusers is automatically sized at boot based on the amount of memory available in the system, and may be determined at run-time by inspecting the value of kern.maxusers . Some systems require larger or smaller values of kern.maxusers and values of 64 , 128 , and 256 are not uncommon. Going above 256 is not recommended unless a huge number of file descriptors is needed. Many of the tunable values set to their defaults by kern.maxusers may be individually overridden at boot-time or run-time in /boot/loader.conf . Refer to loader.conf(5) and /boot/defaults/loader.conf for more details and some hints.

In older releases, the system will auto-tune maxusers if it is set to 0 . ^{[ 1 ]} . When setting this option, set maxusers to at least 4 , especially if the system runs Xorg or is used to compile software. The most important table set by maxusers is the maximum number of processes, which is set to 20 + 16 * maxusers . If maxusers is set to 1 , there can only be 36 simultaneous processes, including the 18 or so that the system starts up at boot time and the 15 or so used by Xorg. Even a simple task like reading a manual page will start up nine processes to filter, decompress, and view it. Setting maxusers to 64 allows up to 1044 simultaneous processes, which should be enough for nearly all uses. If, however, the error is displayed when trying to start another program, or a server is running with a large number of simultaneous users, increase the number and rebuild.

The kern.ipc.soacceptqueue sysctl(8) variable limits the size of the listen queue for accepting new TCP connections. The default value of 128 is typically too low for robust handling of new connections on a heavily loaded web server. For such environments, it is recommended to increase this value to 1024 or higher. A service such as sendmail(8) , or Apache may itself limit the listen queue size, but will often have a directive in its configuration file to adjust the queue size. Large listen queues do a better job of avoiding Denial of Service (DoS) attacks.

The net.inet.ip.portrange.* sysctl(8) variables control the port number ranges automatically bound to TCP and UDP sockets. There are three ranges: a low range, a default range, and a high range. Most network programs use the default range which is controlled by net.inet.ip.portrange.first and net.inet.ip.portrange.last , which default to 1024 and 5000 , respectively. Bound port ranges are used for outgoing connections and it is possible to run the system out of ports under certain circumstances. This most commonly occurs when running a heavily loaded web proxy. The port range is not an issue when running a server which handles mainly incoming connections, such as a web server, or has a limited number of outgoing connections, such as a mail relay. For situations where there is a shortage of ports, it is recommended to increase net.inet.ip.portrange.last modestly. A value of 10000 , 20000 or 30000 may be reasonable. Consider firewall effects when changing the port range. Some firewalls may block large ranges of ports, usually low-numbered ports, and expect systems to use higher ranges of ports for outgoing connections. For this reason, it is not recommended that the value of net.inet.ip.portrange.first be lowered.

TCP bandwidth delay product limiting can be enabled by setting the net.inet.tcp.inflight.enable sysctl(8) variable to 1 . This instructs the system to attempt to calculate the bandwidth delay product for each connection and limit the amount of data queued to the network to just the amount required to maintain optimum throughput.

This feature is useful when serving data over modems, Gigabit Ethernet, high speed WAN links, or any other link with a high bandwidth delay product, especially when also using window scaling or when a large send window has been configured. When enabling this option, also set net.inet.tcp.inflight.debug to 0 to disable debugging. For production use, setting net.inet.tcp.inflight.min to at least 6144 may be beneficial. Setting high minimums may effectively disable bandwidth limiting, depending on the link. The limiting feature reduces the amount of data built up in intermediate route and switch packet queues and reduces the amount of data built up in the local host’s interface queue. With fewer queued packets, interactive connections, especially over slow modems, will operate with lower Round Trip Times . This feature only effects server side data transmission such as uploading. It has no effect on data reception or downloading.

Adjusting net.inet.tcp.inflight.stab is not recommended. This parameter defaults to 20 , representing 2 maximal packets added to the bandwidth delay product window calculation. The additional window is required to stabilize the algorithm and improve responsiveness to changing conditions, but it can also result in higher ping(8) times over slow links, though still much lower than without the inflight algorithm. In such cases, try reducing this parameter to 15 , 10 , or 5 and reducing net.inet.tcp.inflight.min to a value such as 3500 to get the desired effect. Reducing these parameters should be done as a last resort only.

A vnode is the internal representation of a file or directory. Increasing the number of vnodes available to the operating system reduces disk I/O. Normally, this is handled by the operating system and does not need to be changed. In some cases where disk I/O is a bottleneck and the system is running out of vnodes, this setting needs to be increased. The amount of inactive and free RAM will need to be taken into account.

To see the current number of vnodes in use:

To see the maximum vnodes:

If the current vnode usage is near the maximum, try increasing kern.maxvnodes by a value of 1000 . Keep an eye on the number of vfs.numvnodes . If it climbs up to the maximum again, kern.maxvnodes will need to be increased further. Otherwise, a shift in memory usage as reported by top(1) should be visible and more memory should be active.

In some cases, resuming from a suspend operation will cause the mouse to fail. A known work around is to add hint.psm.0.flags="0x3000" to /boot/loader.conf .

ACPI has three suspend to RAM (STR) states, S1 - S3 , and one suspend to disk state (STD), called S4 . STD can be implemented in two separate ways. The S4 BIOS is a BIOS-assisted suspend to disk and S4 OS is implemented entirely by the operating system. The normal state the system is in when plugged in but not powered up is "soft off" ( S5 ).

Use sysctl hw.acpi to check for the suspend-related items. These example results are from a Thinkpad:

Use acpiconf -s to test S3 , S4 , and S5 . An s4bios of one ( 1 ) indicates S4 BIOS support instead of S4 operating system support.

When testing suspend/resume, start with S1 , if supported. This state is most likely to work since it does not require much driver support. No one has implemented S2 , which is similar to S1 . Next, try S3 . This is the deepest STR state and requires a lot of driver support to properly reinitialize the hardware.

A common problem with suspend/resume is that many device drivers do not save, restore, or reinitialize their firmware, registers, or device memory properly. As a first attempt at debugging the problem, try:

This test emulates the suspend/resume cycle of all device drivers without actually going into S3 state. In some cases, problems such as losing firmware state, device watchdog time out, and retrying forever, can be captured with this method. Note that the system will not really enter S3 state, which means devices may not lose power, and many will work fine even if suspend/resume methods are totally missing, unlike real S3 state.

Harder cases require additional hardware, such as a serial port and cable for debugging through a serial console, a Firewire port and cable for using dcons(4) , and kernel debugging skills.

To help isolate the problem, unload as many drivers as possible. If it works, narrow down which driver is the problem by loading drivers until it fails again. Typically, binary drivers like nvidia.ko , display drivers, and USB will have the most problems while Ethernet interfaces usually work fine. If drivers can be properly loaded and unloaded, automate this by putting the appropriate commands in /etc/rc.suspend and /etc/rc.resume . Try setting hw.acpi.reset_video to 1 if the display is messed up after resume. Try setting longer or shorter values for hw.acpi.sleep_delay to see if that helps.

Try loading a recent Linux® distribution to see if suspend/resume works on the same hardware. If it works on Linux®, it is likely a FreeBSD driver problem. Narrowing down which driver causes the problem will assist developers in fixing the problem. Since the ACPI maintainers rarely maintain other drivers, such as sound or ATA, any driver problems should also be posted to the FreeBSD-CURRENT mailing list and mailed to the driver maintainer. Advanced users can include debugging printf(3) s in a problematic driver to track down where in its resume function it hangs.

Finally, try disabling ACPI and enabling APM instead. If suspend/resume works with APM, stick with APM, especially on older hardware (pre-2000). It took vendors a while to get ACPI support correct and older hardware is more likely to have BIOS problems with ACPI.

Most system hangs are a result of lost interrupts or an interrupt storm. Chipsets may have problems based on boot, how the BIOS configures interrupts before correctness of the APIC (MADT) table, and routing of the System Control Interrupt (SCI).

Interrupt storms can be distinguished from lost interrupts by checking the output of vmstat -i and looking at the line that has acpi0 . If the counter is increasing at more than a couple per second, there is an interrupt storm. If the system appears hung, try breaking to DDB ( CTRL + ALT + ESC on console) and type show interrupts .

When dealing with interrupt problems, try disabling APIC support with hint.apic.0.disabled="1" in /boot/loader.conf .

Panics are relatively rare for ACPI and are the top priority to be fixed. The first step is to isolate the steps to reproduce the panic, if possible, and get a backtrace. Follow the advice for enabling options DDB and setting up a serial console in “Entering the DDB Debugger from the Serial Line” or setting up a dump partition. To get a backtrace in DDB, use tr . When handwriting the backtrace, get at least the last five and the top five lines in the trace.

Then, try to isolate the problem by booting with ACPI disabled. If that works, isolate the ACPI subsystem by using various values of debug.acpi.disable . See acpi(4) for some examples.

First, try setting hw.acpi.disable_on_poweroff="0" in /boot/loader.conf . This keeps ACPI from disabling various events during the shutdown process. Some systems need this value set to 1 (the default) for the same reason. This usually fixes the problem of a system powering up spontaneously after a suspend or poweroff.

Some BIOS vendors provide incorrect or buggy bytecode. This is usually manifested by kernel console messages like this:

Often, these problems may be resolved by updating the BIOS to the latest revision. Most console messages are harmless, but if there are other problems, like the battery status is not working, these messages are a good place to start looking for problems.