I applied the patches you mentioned below . One of the two patches from
netraverse fails in two files (don't remember which patch failed and
which files were affected). After analyzing the filename.c.rej-files,
I made the supposed changes by hand. They should have added a
include-statement (like the one in kernel/sched.c below).
I could send a patched version of the netraverse-patches that apply
cleanly to kernel-source-2.4.18-6mdk if you like.
Regrard, Uwe
trax wrote:
Just curious what patches you applied.
I tried patching the source of the 2.4.18-6mdk
with
Kernel-Win4Lin3-2.4.18.patch
mki-adapter.patch
kernel-fix-mki.patch
I see it also fails with kernel/sched.c.***
*** 32,37
#include asm/uaccess.h
#include asm/mmu_context.h
extern void timer_bh(void);
extern void tqueue_bh(void);
--- 32,40
#include asm/uaccess.h
#include asm/mmu_context.h
+ #ifdef CONFIG_MKI
+ #include asm/mki.h
+ #endif
extern void timer_bh(void);
extern void tqueue_bh(void);
###
Trying to create a bzImage, produces the following ...
sched.c: In function `schedule':
sched.c:668: warning: implicit declaration of function `CALL_MKI_HOOK'
sched.c:668: `MKI_HOOK_SWITCH_AWAY' undeclared (first use in this
function)
sched.c:668: (Each undeclared identifier is reported only once
sched.c:668: for each function it appears in.)
make[2]: *** [sched.o] Error 1
make[2]: Leaving directory `/usr/src/linux-2.4.18-6mdk-win4lin/kernel'
make[1]: *** [first_rule] Error 2
make[1]: Leaving directory `/usr/src/linux-2.4.18-6mdk-win4lin/kernel'
make: *** [_dir_kernel] Error 2
So did you add, or change anything else to get this to compile?
(no rush, I am running it on and old 2.4.16 kernel, but it would be nice
to use the supermount and other goodies that are in 8.2)
On Mon, 2002-03-18 at 22:34, Uwe Reimann wrote:
Hi Robert, hi list,
I have a patch ready that makes the patches for win4lin compile and
run
under kernel-2.4.18-6mdk. I am not a kernel-hacker, therefore I don't
know wether everything's done right, but, at least for me, it works.
Apply the patch after applying the patches from netraverse.
Regards, Uwe
P.S.: I did send the patch before, but as I did not see the mail show
up
on the list, I resend it.
Robert Shade wrote:
Running 2.4.17.16mdk :
- really use andrea vm_24.
I assume that you mean that you are running win4lin successfully
with this
kernel? (with necessary patches of course) If you are, please send
me the
file mm/vmscan.c (for reasons I will explain below) from your source
tree to
see what is going on.
Yeah, I did look through the mandrake cvs after I posted and saw
that they've
been using it since then. However looking at vm_24 it looks like
the
collision occurs with the win4lin patch as well.
The problem occurs in the swap_out function in the file mm/vmscan.c,
in the
vanilla kernel the arguments to swap_out are:
unsigned int priority, unsigned int gfp_mask, zone_t * classzone
however, in the aa vm_* kernel patches he made the arguments to
swap_out:
zone_t * classzone
The win4lin patch uses the priority and gfp_mask variables like so:
CALL_MKI_HOOK(MKI_HOOK_SWAP, (void *)priority, (void
*)gfp_mask);
as far as I can see, any kernel with the aa VM will fail to accept
any
win4lin patches.
This might be on the edge of the scope of this list, but hopefully
the
maintainer of the kernel rpm is listening.
Thanks.
Robert
On Mon, 2002-03-18 at 22:34, Uwe Reimann wrote:
Hi Robert, hi list,
I have a patch ready that makes the patches for win4lin compile and run
under kernel-2.4.18-6mdk. I am not a kernel-hacker, therefore I don't
know wether everything's done right, but, at least for me, it works.
Apply the patch after applying the patches from netraverse.
Regards, Uwe
P.S.: I did send the patch before, but as I did not see the mail show up
on the list, I resend it.
Robert Shade wrote:
Running 2.4.17.16mdk :
- really use andrea vm_24.
I assume that you mean that you are running win4lin successfully with this
kernel? (with necessary patches of course) If you are, please send me the
file mm/vmscan.c (for reasons I will explain below) from your source tree to
see what is going on.
Yeah, I did look through the mandrake cvs after I posted and saw that they've
been using it since then. However looking at vm_24 it looks like the
collision occurs with the win4lin patch as well.
The problem occurs in the swap_out function in the file mm/vmscan.c, in the
vanilla kernel the arguments to swap_out are:
unsigned int priority, unsigned int gfp_mask, zone_t * classzone
however, in the aa vm_* kernel patches he made the arguments to swap_out:
zone_t * classzone
The win4lin patch uses the priority and gfp_mask variables like so:
CALL_MKI_HOOK(MKI_HOOK_SWAP, (void