Michael Brown
|
67f122c645
|
Removed pci_find_device_x; it was too ugly to live.
|
19 years ago |
Michael Brown
|
121e7386f6
|
New device probing mechanism written, not tested.
|
20 years ago |
Michael Brown
|
7172d4abfb
|
Moved functions that we want to keep to drivers/bus/pci.c
|
20 years ago |
Michael Brown
|
2ebc56f334
|
Moved in functions from old core/pci.c that we want to keep.
|
20 years ago |
Michael Brown
|
95f755ad7f
|
Renamed pci.c to pci_legacy.c in preparation for new bus managment
infrastructure.
|
20 years ago |
Michael Brown
|
63a0e0e2df
|
Removed references to no-longer-existent CFLAGS options.
|
20 years ago |
Michael Brown
|
5d96e5dad9
|
Changed guard name to avoid clash with INCLUDE_xxx namespace used by
drivers.
|
20 years ago |
Michael Brown
|
378c1b500c
|
Cleaner separation of functionality between post_reloc and
initialise_via_librm.
|
20 years ago |
Michael Brown
|
b9237160ad
|
Almost all information in this file is now obsolete. Writing a new driver
should, with the new build system, just be a case of writing the .c file
and dropping it in place in drivers/net. :)
|
20 years ago |
Michael Brown
|
34a0b5f16f
|
initialise() doesn't take regs
|
20 years ago |
Michael Brown
|
87508aa0b2
|
arch_initialise() is now sufficiently generic that we may as well just
call it initialise() and place it in main.c. :)
|
20 years ago |
Michael Brown
|
997b51c1e6
|
Use librm_arch_initialise for librm, arch_initialise for others.
|
20 years ago |
Michael Brown
|
560027899e
|
Heading towards a sane set of control functions
|
20 years ago |
Michael Brown
|
39cb1b4f9e
|
All the icky stuff is now taken care of by librm_arch_initialise
|
20 years ago |
Michael Brown
|
d189b84536
|
Fix up init ordering.
Make init function table name fixed; it's a static variable anyway.
|
20 years ago |
Michael Brown
|
3714946709
|
arch_rm_main no longer exists.
|
20 years ago |
Michael Brown
|
1937b39079
|
Drag in relocate unless NORELOCATE is defined, since it's now called
only via the initialisation function table.
|
20 years ago |
Michael Brown
|
f06e8c9707
|
Added post-relocation function table.
|
20 years ago |
Michael Brown
|
9abceef561
|
Removed obsolete initialisation calls.
|
20 years ago |
Michael Brown
|
070f6c0b6b
|
Fixed debug messages
|
20 years ago |
Michael Brown
|
43417e6c1b
|
setup16() now calls initialise(); we don't have to do it.
|
20 years ago |
Michael Brown
|
04682f9838
|
Setup now calls initialise() directly, rather than expecting the
prefix to do it.
|
20 years ago |
Michael Brown
|
44e62317be
|
Add librm_base.
|
20 years ago |
Michael Brown
|
dfb9c44994
|
hooks.o is now a single object, rather than having separate hooks.o and
hooks_rm.o
|
20 years ago |
Michael Brown
|
04a99841e6
|
Make prot_call() able to transparently return via the newly installed copy
of librm.
|
20 years ago |
Michael Brown
|
44eee81d11
|
Don't build relocate() under KEEP_IT_REAL
|
20 years ago |
Michael Brown
|
2fa14f2f1d
|
lgdt needs data32 prefix, otherwise it loads only 24 bits. (sgdt
always saves 32 bits).
|
20 years ago |
Michael Brown
|
6a4ac358f3
|
Add INIT_FNs to make sure that librm gets reallocated when needed.
|
20 years ago |
Michael Brown
|
e29e6456b0
|
Don't zero freed base memory; one block will contain librm.
|
20 years ago |
Michael Brown
|
1f6dfd7047
|
Allow ourselves to be freed by basemem.c without being damaged.
|
20 years ago |
Michael Brown
|
97ced37c71
|
Need to provide sizeof(struct free_base_memory_header) to assembly code.
|
20 years ago |
Michael Brown
|
90cf689810
|
Note basemem API change.
|
20 years ago |
Michael Brown
|
ef05051f23
|
Add INIT_LIBRM before INIT_CONSOLE.
|
20 years ago |
Michael Brown
|
48feb91a40
|
Rewritten in a much saner way, now that we don't have to worry about
continually reallocating the real-mode stack.
|
20 years ago |
Michael Brown
|
0574136c6d
|
Split out from etherboot.h
|
20 years ago |
Michael Brown
|
8570814148
|
Removed basemem.c prototypes
|
20 years ago |
Michael Brown
|
fd95189c4d
|
basememsize() updates the BIOS FBMS counter at 40:13.
|
20 years ago |
Michael Brown
|
a800e9657d
|
PM function called by prot_call() should be allowed to modifying
registers returned to RM code.
|
20 years ago |
Michael Brown
|
29f43896aa
|
Add relocate.h
Note that we can't make real-mode calls immediately after relocate_to().
|
20 years ago |
Michael Brown
|
9cc90cbb07
|
First version
|
20 years ago |
Michael Brown
|
c8c7d93757
|
Relocation code is arch-specific.
|
20 years ago |
Michael Brown
|
64d3886c1d
|
relocate() is called by arch_initialise().
|
20 years ago |
Michael Brown
|
8039086ec8
|
Added a lot of explanatory comments.
Preserves alignment requested via _max_align parameter to linker.
|
20 years ago |
Michael Brown
|
19176718be
|
callbacks.h no longer exists
|
20 years ago |
Michael Brown
|
792f34d228
|
#ifdef 0 out the UNDI loader routine, since it refers to the obsolete
callback mechanism.
|
20 years ago |
Michael Brown
|
73dbe41007
|
Old callbacks mechanism is obsolete.
|
20 years ago |
Michael Brown
|
fedd00dd46
|
Removed references to old callback mechanism
|
20 years ago |
Michael Brown
|
9c32083d56
|
callbacks.h no longer exists.
|
20 years ago |
Michael Brown
|
a3faaf405a
|
Callback mechanism now obsolete
|
20 years ago |
Michael Brown
|
a4920c16bd
|
Moved relocate.c to arch-specific dir, since it involves arch-specific
logic (e.g. avoiding memory regions with A20=0).
|
20 years ago |