| Age | Commit message (Collapse) | Author |
|---|
|
mmap'ed), by properly making these mappings uncached, and correctly accounting
their removal to not get stuck with leftover pte pages.
While there, change a can't happen infinite loop in pmap_destroy() into a
KASSERT.
feedback and ok kettenis@
|
|
interrupts would be counted double, once for the interrupting device and
once for the device at the head of the chain. The handlers would run properly
though. Avoid this by giving each device its own interrupt counter instead
of using the counter provided by the generic interrupt handling code for the
head of the chain.
|
|
|
|
|
|
This code is not yet called as there are still some important parts
not completed.
ok deraadt@, kettenis@ "looks reasonable"
|
|
numbers in comments and the MAKEDEV majors.
|
|
This will enable uvm_map to behave intelligently when allocating.
Prodded by deraadt to commit this.
|
|
This enables future uvm_map allocator to behave intelligently.
|
|
This will enable future uvm_map code to make intelligent decisions during
allocation.
|
|
This will enable future uvm_map to make intelligent allocation decisions.
|
|
This will enable intelligent decisions with future uvm_map.
|
|
This enables future uvm_map code to make intelligent decisions.
Code is not called at the moment.
|
|
Enables future uvm_map code to make intelligent decisions during allocation.
No functional change.
|
|
Enables future uvm_map code to make intelligent decisions.
No functional change.
|
|
- add nearbyint, nearbyintf and nearbyintl implemented using fenv
|
|
|
|
This fixes possible stack overflows under high interrupt load. Note
that this allows traps on kernel code to run with interrupts blocked.
From Christian Ehrhardt
ok guenther@ deraadt@
|
|
if you boot from a disk that ends up behind mpath(4) on sparc64,
it wont be able to resolve it to the actual disk device since a
path driver (eg sym(4)) sits where ofw tells us a disk is.
this diff allows the bootpath code to match on path drivers, and
then asks mpath to swap the path for the disk device.
ok kettenis@
|
|
Therefore set UVM_FLAG_FIXED and enforce this.
ok oga@
|
|
|
|
kernel so that it can use it to identify the root disk. This will be
needed in order to correctly boot from a softraid volume.
ok deraadt@ marco@ krw@
|
|
|
|
|
|
|
|
ok deraadt@
|
|
|
|
and display the symbol name if there is an exact match. The optimizer likes
to store addresses into `d' registers, then copy them to `a' registers and
invoke the function indirectly; previously we'd only get symbol lookup for
`lea' instructions (when storing to an `a' register).
|
|
scsi_link accordingly, for bad things will happen without (and this was a
real pita to debug).
|
|
|
|
would enable inexact trap instead of setting the sticky bit.
OK miod@.
|
|
|
|
not reviewed yet, but it's better to track changes in cvs
|
|
|
|
Delivering FPE with non-masked exceptions doesn't work on Alpha; I
suspect there's a bug in the kernel trap handler.
FE_INEXACT is intentionally left out of FE_ALL_EXCEPT; since the
inexact exception is not being maintained. Otherwise it will lead
to the bogus results.
|
|
(positive interrupt was for me), like we do on other architectures.
This is done here, at the elroy(4) driver level, since this is where shared
PCI interrupts are handled. We could do something similar for dino(4) but
this optimization is probably not very relevant there.
ok deraadt@
|
|
writes in the pte momentarily. Modified the code stolen from amd64.
ok miod
|
|
says; for couple of reasons:
- makes it actually work, since the code clears ~FE_ALL_EXCEPT bits.
- standard requires that.
|
|
so that the userland actually gets the proto.
OK kettenis@, miod@.
|
|
by deraadt@.
|
|
giving systems with non-default values a chance to work again. Pass the brown
paper bag, please.
|
|
possible DMA window. This breaks systems which have physical memory contiguous
to the selected DMA window. A fix is in the works, in the meantime the
spice^WDMA should flow.
|
|
|
|
not make these drivers spew millions of lines of output.
spotted as missing by miod
|
|
ok marco@, deraadt@
|
|
interrupts. It is irreleveant, confuses people and the information is
available in pcidump(8) output anyway.
ok oga@, jsg@, deraadt@
|
|
|
|
feedback & ok guenther@, matthew@
|
|
|
|
arithmetic add.
|
|
|
