MSR_KGSBASE is no longer saved and restored from the guest MSR save area. This
behavior was changed in r271888 so update the comment block to reflect this.
MSR_KGSBASE is accessible from the guest without triggering a VM-exit. The
permission bitmap for MSR_KGSBASE is modified by vmx_msr_guest_init() so get
rid of redundant code in vmx_vminit().
Restructure the MSR handling so it is entirely handled by processor-specific
code. There are only a handful of MSRs common between the two so there isn't
too much duplicate functionality.
The VT-x code has the following types of MSRs:
- MSRs that are unconditionally saved/restored on every guest/host context
switch (e.g., MSR_GSBASE).
- MSRs that are restored to guest values on entry to vmx_run() and saved
before returning. This is an optimization for MSRs that are not used in
host kernel context (e.g., MSR_KGSBASE).
- MSRs that are emulated and every access by the guest causes a trap into
the hypervisor (e.g., MSR_IA32_MISC_ENABLE).
adrian [Sat, 20 Sep 2014 01:22:17 +0000 (01:22 +0000)]
Fix up the EDMA RX setup path to correctly initialise and reset the RX FIFO.
The original code was .. well, slightly more than incorrect.
It showed up as stalled RX queues if the NIC needed to be frequently
reinitialised (eg during scans.)
This is inspired by work done by Matt Dillon over at the DragonflyBSD
project.
So:
* track when EDMA RX has been stopped and when the MAC has been reset;
* re-initialise the ring only after a reset;
* track whether RX has been stopped/started - just for debugging now;
* don't bother with the RX EOL stuff for EDMA - we don't need the
interrupt at all. We also don't need to disable/enable the interrupt
or start DMA - once new frames are pushed into the ring via the
normal RX path, it'll just restart RX DMA on its own.
Tested:
* AR9380, STA mode
* AR9380, AP mode
* AR9485, STA mode
* AR9462, STA mode
Fix partition alignment and image rounding when any of -P (block size),
-T (track size) or -H (number of heads) is given:
o scheme_metadata() always rounded to the block size. This is not
always valid (e.g. vtoc8 that must have partitions start at cylinder
boundaries).
o The bsd and vtoc8 schemes "resized" the image to make it match the
geometry, but since the geometry is an approximation and the size
of the image computed from cylinders * heads * sectors is always
smaller than the original image size, the partition information ran
out of bounds.
The fix is to have scheme_metadata() simply pass it's arguments to the
per-scheme metadata callback, so that schemes not only know where the
metadata is to go, but also what the current block address is. It's now
up to the per-scheme callback to reserve room for metadata and to make
sure alignment and rounding is applied.
The BSD scheme now has the most elaborate alignment and rounding. Just
to make the point: partitions are aligned on block boundaries, but the
image is rounded to the next cyclinder boundary.
vtoc8 now properly has all partitions aligned (and rounded) to the
cyclinder boundary.
Obtained from: Juniper Networks, Inc.
MFC after: 3 days
Put the various atf_tc_pack_t variables generated by ATF_TC to use in
the macros_h_test.c file so that we prevent some build warnings (and
thus some build errors) with clang and -Wunused.
It's now possible to scroll up the 500 hard-coded lines of history, not
just a fraction of them. For instance, one can reach the top of the boot
process.
Sometimes, when scrolling or when changing the screen size (by changing
the font or loading a KMS driver for instance), one could see the
history cycling (old content appeared below latest lines). This is
fixed.
Now, when the resolution changes are more lines can be shown, the
displayed area is adjusted so that, if the screen was filled with
content before, it's filled with content after as well: more history
is visible, instead of having blank lines below the previously visible
content.
- Provide igb_get_counter() to return counters that are not collected,
but taken from hardware.
- Mechanically convert to if_inc_counter() the rest of counters.
Revert r270808, which were changes to common code (i40e_alloc.h).
Rather than #define-ing common code function calls to OS-dependent
ones, make the osdep versions match the common code expectations,
adjust the FreeBSD specific code to use those, and remove the
#defines.
In the FreeBSD specific code, use "i40e_mem_reserved" for the
now expected but unused argument to i40e_allocate_dma_mem().
Reviewed by: gnn, eric.joyner intel.com
MFC after: 3 days
- Provide if_get_counter() method for vtnet(4).
- Do not accumulate statistics on every tick.
- Accumulate statistics in vtnet_setup_stat_sysctl()
and in vtnet_get_counter().
will [Thu, 18 Sep 2014 17:28:21 +0000 (17:28 +0000)]
Start the process of cleaning up FreeBSD's firewire driver.
sys/dev/firewire/firewire.c:
sys/dev/firewire/firewire.h:
sys/dev/firewire/firewirereg.h:
sys/dev/firewire/fwcrom.c:
sys/dev/firewire/fwdev.c:
sys/dev/firewire/fwdma.c:
sys/dev/firewire/fwmem.c:
sys/dev/firewire/fwohci.c:
sys/dev/firewire/fwohci_pci.c:
sys/dev/firewire/fwohcivar.h:
sys/dev/firewire/if_fwe.c:
sys/dev/firewire/if_fwip.c:
sys/dev/firewire/sbp.c:
sys/dev/firewire/sbp_targ.c:
Unifdef the code, removing support for DragonflyBSD
and FreeBSD prior to version 5.
adrian [Thu, 18 Sep 2014 16:20:17 +0000 (16:20 +0000)]
Fix the handling of EOP in status descriptors for if_igb(4) and don't
double-free mbufs.
Like ixgbe(4) chipsets, EOP is only set on the final descriptor
in a chain of descriptors. So, to free the whole list of descriptors,
we should free the current slot _and_ the assembled list of descriptors
that make up the fragment list.
The existing code was setting discard once it saw EOP + an error status;
it then freed all the subsequent descriptors until the next EOP. That's
totally the wrong order.
- Use if_inc_counter() to increment various counters.
- Do not ever set a counter to a value. For those counters
that we don't increment, but return directly from hardware
create cases in if_get_counter() method.