1 |
Holly Bostick wrote: |
2 |
> Peter Ruskin wrote: |
3 |
> |
4 |
>> On Monday 06 December 2004 05:09, Ow Mun Heng wrote: |
5 |
>> |
6 |
> >>PS : Your signature noted that you're runnning 2.6.9-r6-regparm. |
7 |
> |
8 |
>>> What's Regparm |
9 |
>> |
10 |
>> |
11 |
>> |
12 |
>> I just added "-regparm" to "EXTRAVERSION" in the kernel source |
13 |
>> Makefile to remind me that I had set CONFIG_REGPARM=y (Processor type |
14 |
>> and features): |
15 |
>> |
16 |
>> "Use register arguments (EXPERIMENTAL) (REGPARM) |
17 |
>> |
18 |
>> Compile the kernel with -mregparm=3. This uses an different ABI and |
19 |
>> passes the first three arguments of a function call in registers. |
20 |
>> This will probably break binary only modules. |
21 |
> |
22 |
> |
23 |
> Like the ATI drivers, if you use them (which I do), thus presumably also |
24 |
> the nVidia drivers, if you use them. So I don't use this setting myself, |
25 |
> tasty as it looks. |
26 |
> |
27 |
> Just a note, before Ow Mung goes recompiling his kernel to enable this |
28 |
> ;-) . |
29 |
> |
30 |
> Holly |
31 |
|
32 |
Not so much to reply to myself, but on second thought, maybe it was |
33 |
splashutils/fbsplash that is broken by this. |
34 |
|
35 |
Anyway, the point being that there are some modules that seem unrelated, |
36 |
but which one might be using, that will break if regparm is set. So be |
37 |
careful. |
38 |
|
39 |
Holly |
40 |
> |
41 |
>> |
42 |
>> This feature is only enabled for gcc-3.0 and later - earlier compilers |
43 |
>> generate incorrect output with certain kernel constructs |
44 |
>> when-mregparm=3 is used." |
45 |
>> |
46 |
> |
47 |
> |
48 |
> -- |
49 |
> gentoo-user@g.o mailing list |
50 |
> |
51 |
> |
52 |
|
53 |
|
54 |
-- |
55 |
gentoo-user@g.o mailing list |