Alexander Motin [Sat, 14 Mar 2015 21:15:45 +0000 (21:15 +0000)]
Give block I/O interface multiple (8) execution threads.
On parallel random I/O this allows better utilize wide storage pools.
To not confuse prefetcher on linear I/O, consecutive requests are executed
sequentially, following the same logic as was earlier implemented in CTL.
Benchmarks of virtual AHCI disk, backed by ZVOL on RAID10 pool of 4 HDDs,
show ~3.5 times random read performance improvements, while no degradation
on linear I/O.
Steven Hartland [Sat, 14 Mar 2015 17:35:04 +0000 (17:35 +0000)]
Allow zvol_geom_worker to process BIO_DELETE's
If zvol_geom_start is called with a BIO_DELETE from a thread which can
sleep it queues it for later processing by the zvol_geom_worker. The
zvol_geom_worker didn't have a delete case so would simply loose the bio
hence preventing the original caller from every completing. In addition
an other unknown types would suffer the same fate.
Allow zvol_geom_worker to process BIO_DELETE's via zvol_strategy and
return unsupported for all unknown bio types.
Ian Lepore [Sat, 14 Mar 2015 17:08:28 +0000 (17:08 +0000)]
Set the SBUF_INCLUDENUL flag in sbuf_new_for_sysctl() so that sysctl
strings returned to userland include the nulterm byte.
Some uses of sbuf_new_for_sysctl() write binary data rather than strings;
clear the SBUF_INCLUDENUL flag after calling sbuf_new_for_sysctl() in
those cases. (Note that the sbuf code still automatically adds a nulterm
byte in sbuf_finish(), but since it's not included in the length it won't
get copied to userland along with the binary data.)
Remove explicit adding of a nulterm byte in a couple places now that it
gets done automatically by the sbuf drain code.
Ian Lepore [Sat, 14 Mar 2015 16:02:11 +0000 (16:02 +0000)]
Add a new flag, SBUF_INCLUDENUL, and new get/set/clear functions for flags.
The SBUF_INCLUDENUL flag causes the nulterm byte at the end of the string
to be counted in the length of the data. If copying the data using the
sbuf_data() and sbuf_len() functions, or if writing it automatically with
a drain function, the net effect is that the nulterm byte is copied along
with the rest of the data.
Jilles Tjoelker [Sat, 14 Mar 2015 13:45:43 +0000 (13:45 +0000)]
restore: Preserve timestamps to the nanosecond.
The restore utility already knows the full-resolution timestamps, so the
only thing to do is to stop converting the timespecs to timevals and use
futimens() and utimensat().
Dimitry Andric [Sat, 14 Mar 2015 12:40:19 +0000 (12:40 +0000)]
Pull in r231965 from upstream compiler-rt trunk (by Jörg Sonnenberger):
Refactor float to integer conversion to share the same code.
80bit Intel/PPC long double is excluded due to lacking support
for the abstraction. Consistently provide saturation logic.
Extend to long double on 128bit IEEE extended platforms.
Initial patch with test cases from GuanHong Liu.
Reviewed by Steve Canon.
Adrian Chadd [Sat, 14 Mar 2015 08:29:03 +0000 (08:29 +0000)]
Compile some modules - I'm going to eventually just compile all the
modules, but these are a subset of things that are worth playing with
in deployed APs.
Neel Natu [Sat, 14 Mar 2015 00:30:41 +0000 (00:30 +0000)]
Add x86 specific APIs 'lapic_ipi_alloc()' and 'lapic_ipi_free()' to allow IPI
vectors to be dynamically allocated. This allows kernel modules like vmm.ko
to allocate unique IPI slots when loaded (as opposed to hard allocating one
or more vectors).
Also, reorganize the fixed IPI vectors to create a contiguous space for
dynamic IPI allocation.
Properly initialize scope zone id when next hop address stored
directly in the O_FORWARD_IP6 opcode. Use getnameinfo(3) to formatting
the IPv6 addresses of such opcodes.
John Baldwin [Fri, 13 Mar 2015 09:45:06 +0000 (09:45 +0000)]
Simplify string mangling in ifmaybeload().
- Use strlcpy() instead of strcpy().
- Use strlcat() instead of a strlcpy() with a magic number subtracted
from the length.
- Replace strncmp(..., strlen(foo) + 1) with strcmp(...).
John Baldwin [Fri, 13 Mar 2015 09:41:27 +0000 (09:41 +0000)]
Enable bzipfs support in the EFI loader.
- Add bzipfs to the list of supported filesystems in the EFI loader.
- Increase the heap size allocated for the EFI loader from 2MB to 3MB.
John Baldwin [Fri, 13 Mar 2015 09:38:16 +0000 (09:38 +0000)]
The System V ABI for amd64 allows functions to use space in a 128 byte
redzone below the stack pointer for scratch space and requires
interrupt and signal frames to avoid overwriting it. However, EFI uses
the Windows ABI which does not support this. As a result, interrupt
handlers in EFI push their interrupt frames directly on top of the
stack pointer. If the compiler used the red zone in a function in the
EFI loader, then a device interrupt that occurred while that function
was running could trash its local variables. In practice this happens
fairly reliable when using gzipfs as an interrupt during decompression
can trash the local variables in the inflate_table() function
resulting in corrupted output or hangs.
Fix this by disabling the redzone for amd64 EFI binaries. This
requires building not only the loader but any libraries used by the
loader without redzone support.
Thanks to Jilles for pointing me at the redzone once I found the stack
corruption.
Fix `ipfw fwd tablearg'. Use dedicated field nh4 in struct table_value
to obtain IPv4 next hop address in tablearg case.
Add `fwd tablearg' support for IPv6. ipfw(8) uses INADDR_ANY as next hop
address in O_FORWARD_IP opcode for specifying tablearg case. For IPv6 we
still use this opcode, but when packet identified as IPv6 packet, we
obtain next hop address from dedicated field nh6 in struct table_value.
Replace hopstore field in struct ip_fw_args with anonymous union and add
hopstore6 field. Use this field to copy tablearg value for IPv6.
Replace spare1 field in struct table_value with zoneid. Use it to keep
scope zone id for link-local IPv6 addresses. Since spare1 was used
internally, replace spare0 array with two variables spare0 and spare1.
Use getaddrinfo(3)/getnameinfo(3) functions for parsing and formatting
IPv6 addresses in table_value. Use zoneid field in struct table_value
to store sin6_scope_id value.
Since the kernel still uses embedded scope zone id to represent
link-local addresses, convert next_hop6 address into this form before
return from pfil processing. This also fixes in6_localip() check
for link-local addresses.
Adrian Chadd [Fri, 13 Mar 2015 02:16:39 +0000 (02:16 +0000)]
Commit 802.1q configuration support for the AR8327.
This is slightly different to the other switches - the VLAN table
(VTU) programs in the vlan port mapping /and/ the port config
(tagged, untagged, passthrough, any.)
So:
* Add VTU operations to program the VTU (vlan table)
* abstract out the mirror-disable function so it's .. well, a function.
* setup the port to have a dot1q configuration for dot1q - the
port security is VLAN (not per-port VLAN) and requires an entry
in the VLAN table;
* add set_dot1q / get_dot1q to program the VLAN table;
* since the tagged/untagged ports are now programmed into the VTU,
rather than global - plumb the ports /and/ untagged ports bitmaps
through the arswitch API.
John Baldwin [Thu, 12 Mar 2015 17:07:24 +0000 (17:07 +0000)]
Allow the EFI loader to work with large kernels and/or modules
(for example, a large mfsroot). Note that for EFI the kernel and
modules (as well as other metadata files such as splash screens or
memory disk images) are loaded into a statically-sized staging area.
When the EFI loader exits it copies this staging area down to the
location the kernel expects to run at.
- Add bounds checking to the copy routines to fail attempts to access
memory outside of the staging area. Previously loading a combined
kernel + modules larger than the staging size (32MB) would overflow
the staging area trashing whatever memory was afterwards. Under
Intel's OVMF firmware for qemu this resulted in fatal faults in the
firmware itself. Now the attempt will fail with ENOMEM.
- Allow the staging area size to be configured at compile time via
an EFI_STAGING_SIZE variable in src.conf or on the command line.
It accepts the size of the staging area in MB. The default size
remains 32MB.
MFC after: 2 weeks
Sponsored by: Cisco Systems, Inc.
Nathan Whitehorn [Thu, 12 Mar 2015 17:01:30 +0000 (17:01 +0000)]
The H_VIO_SIGNAL hypercall only enables interrupts for future received
packets and does not schedule interrupts for any packets currently
enqueued. Close two races where enqueued packets may not ever trigger
interrupts. The first of these, at adapter initialization time, was
especially severe since a rush of enqueued packets could actually fill
the receive buffer completely, stalling the interface forever.
John Baldwin [Thu, 12 Mar 2015 15:20:05 +0000 (15:20 +0000)]
Revert r279381. The EFI loader needs to use a separate libstand that
is compiled against the ABI EFI expects (specifically, no stack
redzone) so it cannot share libstand with userboot (which must use
the System V ABI).
Using parent DMA tag in drm_pci_alloc(). This can allow drm2 devices to work with Intel DMAR enabled for the system, as long as DMAR is disabled for the drm2 device.
Reset mbuf pointer to NULL in fastroute case to indicate that mbuf was
consumed by filter. This fixes several panics due to accessing to mbuf
after free.
Ed Maste [Thu, 12 Mar 2015 08:40:32 +0000 (08:40 +0000)]
Provide automatic cross-binutils path if no BINUTILS_BOOTSTRAP
The in-tree binutils does not support arm64, so will not work for the
forthcoming FreeBSD arm64 port. BROKEN_OPTIONS will include
BINUTILS_BOOTSTRAP, so provide a default CROSS_BINUTILS_PREFIX for this
case.
Reviewed by: imp
Sponsored by: The FreeBSD Foundation
Warner Losh [Thu, 12 Mar 2015 06:43:58 +0000 (06:43 +0000)]
File names with commas in them cause issues for freebsd-update. We
don't actually use these files at the moment, so eliminate them until
we actually do. In the mean time, freebsd-update will be updated
to eliminate the issues.
Warner Losh [Thu, 12 Mar 2015 03:57:00 +0000 (03:57 +0000)]
Add support for specifying unsupported / broken options that override
any defaults or user specified actions on the command line. This would
be useful for specifying features that are always broken or that
cannot make sense on a specific architecture, like ACPI on pc98 or
EISA on !i386 (!x86 usage of EISA is broken and there's no supported
hardware that could have it in any event). Any items in
BROKEN_OPTIONS are forced to "no" regardless of other settings.
Clients are expected change BROKEN_OPTIONS with +=. It will not
be unset, so other parts of the build system can have visibility
into the options that are broken on this platform, though this
should be very rare.
Randall Stewart [Wed, 11 Mar 2015 20:15:49 +0000 (20:15 +0000)]
You need to have the capabilities and not skip it if you are
not on head.. otherwise the file pointer will be NULL and when
you try to do something with it you will crash. Make the #else
be the old capabilites, and then remove the erroneous ifdefs for
11.
MFC after: 1 week (with the other MFC I was going to do until the panic)
Factor out mbuf hashing code from LAGG driver so that other network
drivers can use it. This avoids some code duplication. Add missing
default case to all switch statements while at it. Also move the
hashing of the IPv6 flow field to layer 4 because the IPv6 flow field
is constant on a per L4 connection basis and not on a per L3 network.