freebsd-skq/sys/conf/kern.mk
peter eea63ec45a Major pmap rework to take advantage of the larger address space on amd64
systems.  Of note:
- Implement a direct mapped region using 2MB pages.  This eliminates the
  need for temporary mappings when getting ptes.  This supports up to
  512GB of physical memory for now.  This should be enough for a while.
- Implement a 4-tier page table system.  Most of the infrastructure is
  there for 128TB of userland virtual address space, but only 512GB is
  presently enabled due to a mystery bug somewhere.  The design of this
  was heavily inspired by the alpha pmap.c.
- The kernel is moved into the negative address space(!).
- The kernel has 2GB of KVM available.
- Provide a uma memory allocator to use the direct map region to take
  advantage of the 2MB TLBs.
- Fixed some assumptions in the bus_space macros about the ability
  to fit virtual addresses in an 'int'.

Notable missing things:
- pmap_growkernel() should be able to grow to 512GB of KVM by expanding
  downwards below kernbase.  The kernel must be at the top 2GB of the
  negative address space because of gcc code generation strategies.
- need to fix the >512GB user vm code.

Approved by:	re (blanket)
2003-05-23 05:04:54 +00:00

69 lines
2.3 KiB
Makefile

# $FreeBSD$
#
# Warning flags for compiling the kernel and components of the kernel.
#
# Note that the newly added -Wcast-qual is responsible for generating
# most of the remaining warnings. Warnings introduced with -Wall will
# also pop up, but are easier to fix.
CWARNFLAGS?= -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes \
-Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual \
-fformat-extensions -std=c99
#
# The following flags are next up for working on:
# -W
#
# On the i386, do not align the stack to 16-byte boundaries. Otherwise GCC
# 2.95 adds code to the entry and exit point of every function to align the
# stack to 16-byte boundaries -- thus wasting approximately 12 bytes of stack
# per function call. While the 16-byte alignment may benefit micro benchmarks,
# it is probably an overall loss as it makes the code bigger (less efficient
# use of code cache tag lines) and uses more stack (less efficient use of data
# cache tag lines)
#
.if ${MACHINE_ARCH} == "i386"
CFLAGS+= -mno-align-long-strings -mpreferred-stack-boundary=2
.endif
#
# On the alpha, make sure that we don't use floating-point registers and
# allow the use of BWX etc instructions (only needed for low-level i/o).
# Also, reserve register t7 to point at per-cpu global variables.
#
.if ${MACHINE_ARCH} == "alpha"
CFLAGS+= -mno-fp-regs -ffixed-8 -Wa,-mev6
.endif
#
# For IA-64, we use r13 for the kernel globals pointer and we only use
# a very small subset of float registers for integer divides.
#
.if ${MACHINE_ARCH} == "ia64"
CFLAGS+= -ffixed-r13 -mfixed-range=f32-f127 -mno-sdata
.endif
#
# For sparc64 we want medlow code model, and we tell gcc to use floating
# point emulation. This avoids using floating point registers for integer
# operations which it has a tendency to do.
#
.if ${MACHINE_ARCH} == "sparc64"
CFLAGS+= -mcmodel=medlow -msoft-float
.endif
#
# For AMD64, use a medium model for now. We'll switch to "kernel"
# once pmap is ready. Be excessively careful to not generate FPU code.
#
.if ${MACHINE_ARCH} == "amd64"
CFLAGS+= -mcmodel=kernel -mno-red-zone \
-mfpmath=387 -mno-sse -mno-sse2 -mno-mmx -mno-3dnow -msoft-float
.endif
#
# GCC 3.0 and above like to do certain optimizations based on the
# assumption that the program is linked against libc. Stop this.
#
CFLAGS+= -ffreestanding