| Age | Commit message (Collapse) | Author |
|---|
|
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
This is needed for the following reason: our memory layout
looks like this:
| |vmlinux | |CFE| * |
| | |vmlinuz | |
^
Since CFE can only load to the spot marked with ‘^’ anyway,
we load vmlinub.elf there which is basically a big rodata
blob containing vmlinuz and minimal code moving it to the
location pointed with vmlinuz above. Another solution would
be to use CFE’s “boot -raw” to place it on the location marked
with ‘*’ above (but the CFE location and size are dynamic, and
since it insists on loading to 0x80001000 anyway, this point
is virtually moot).
Even worse, we probably cannot overwrite CFE space even late,
so we move vmlinuz like this:
| |vmlinux | |CFE| |vmlinuz | |
This way, both “Total memory used by CFE” and “Boot area
(physical)” (the latter as mapped into KSEG0) are retained.
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
This reverts commit f6897b395f6340132ab47a46fe587e8e6ced647a.
|
|
This is needed for the following reason: our memory layout
looks like this:
| |vmlinux | |CFE| * |
| | |vmlinuz | |
^
Since CFE can only load to the spot marked with ‘^’ anyway,
we load vmlinub.elf there which is basically a big rodata
blob containing vmlinuz and minimal code moving it to the
location pointed with vmlinuz above. Another solution would
be to use CFE’s “boot -raw” to place it on the location marked
with ‘*’ above (but the CFE location and size are dynamic, and
since it insists on loading to 0x80001000 anyway, this point
is virtually moot).
Even worse, we probably cannot overwrite CFE space even late,
so we move vmlinuz like this:
| |vmlinux | |CFE| |vmlinuz | |
This way, both “Total memory used by CFE” and “Boot area
(physical)” (the latter as mapped into KSEG0) are retained.
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
This reverts commit 5bd743ad60b85f005235aac2435563b168e012e3.
|
|
This is needed for the following reason: our memory layout
looks like this:
| |vmlinux | |CFE| * |
| | |vmlinuz | |
^
Since CFE can only load to the spot marked with ‘^’ anyway,
we load vmlinub.elf there which is basically a big rodata
blob containing vmlinuz and minimal code moving it to the
location pointed with vmlinuz above. Another solution would
be to use CFE’s “boot -raw” to place it on the location marked
with ‘*’ above (but the CFE location and size are dynamic, and
since it insists on loading to 0x80001000 anyway, this point
is virtually moot).
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
This reverts commit 80d8381811821445bc47cd6e46f1eea423d9fce5.
|
|
This is needed for the following reason: our memory layout
looks like this:
| |vmlinux | |CFE| * |
| | |vmlinuz | |
^
Since CFE can only load to the spot marked with ‘^’ anyway,
we load vmlinub.elf there which is basically a big rodata
blob containing vmlinuz and minimal code moving it to the
location pointed with vmlinuz above. Another solution would
be to use CFE’s “boot -raw” to place it on the location marked
with ‘*’ above (but the CFE location and size are dynamic, and
since it insists on loading to 0x80001000 anyway, this point
is virtually moot).
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
|
|
Make configuration of new targets cheap.
Just add a new file in target/arch/sys-enabled/foo.
See other files for syntax. While doing runtime tests
with the new infrastructure I've updated a lot of other
stuff:
- gcc 4.5.2
- uClibc 0.9.32-rc1 (NPTL)
- strongswan, php, miredo, parted, util-linux-ng, e2fsprogs
I promise, this is the last big fat commit this year ;)
|
|
|
|
|
|
- add kernel support for AGP/DRM/KMS
- update intel driver
- fix MesaLib (add the correct driver)
|
|
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
note: uuid.patch currently still panics ☹
Signed-off-by: Thorsten Glaser <tg@mirbsd.org>
|
|
|
|
abstract version information, can be overwritten for each
target. default is 2.6.36 right now. Only ibmx40 target does use it.
mk/modules.mk is still problematic..
aufs2 need to be ported...
|
|
Conflicts:
package/freeradius-server/Makefile
package/nss/patches/patch-mozilla_security_coreconf_Linux_mk
|
|
|
|
|
|
patches
|
|
|
|
|
|
|
|
|
|
after n0-1 helped to find a way to fix my broken rb433
by using the primary bootloader via shortening a jumper
I got the motivation to get this target working again.
Summarize both targets to newly created rb4xx target.
|
|
|
|
|
|
alice iptv seems to use destination=ip:port instead of client_port=port, like
RFC suggesting it. But destination= is also used by STUN method.
As a quick hack I commented out STUN method to avoid crashes.
Now at least I can see some tv streams. still not 100% good.
|
|
|
|
|
|
|
|
- use e1000 driver by default
- no need for dummy qemu.img, phil is right.
|
|
|
|
|
|
|
|
After this change it is not needed to create
console, tty and null device nodes for nfsroot boot.
|
|
|
|
|
|
|
|
|
|
- fix image build for different root filesystems
- get rid of duplicate symlink to include dir
- finetune mpd config file
|
|
- update some applications
asterisk, curl, strace, radvd, rpm
- fix allconfig
- /dev for full build seems to need more than 64k
|
|
|
|
|
|
|
|
|
|
WARNING: tested only on qemu-x86
|
|
|
