| commit | 6f636fe9f94a05952c3759c2f74723c40373d845 | |
| author | DizzyOfCRN <DizzyOfCRN@fb15a70f-31f2-0310-bbcc-cdcc74a49acc> | |
| Tue, 26 Aug 2014 12:19:55 +0000 (26 12:19 +0000) | ||
| committer | DizzyOfCRN <DizzyOfCRN@fb15a70f-31f2-0310-bbcc-cdcc74a49acc> | |
| Tue, 26 Aug 2014 12:19:55 +0000 (26 12:19 +0000) | ||
| tree | 043241f95689e04b749ddd2478ed54751f563ae1 | treesnapshot (tar.gz zip) |
| parent | 9f4e64a4f2ab6882a2637881bf336933398a1373 | commitdiff |
Implemented long ECAM write method.
Making sure the CAM methods are not used if register value is in the extended configuration space,
there are some chipsets that can access extended configuration space with CAM method, but we don't have any support for that.
Pointers to extended configuration space registers made volatile.
CAM method 2 was missing alignement to 32-bit long all along.
ECAM write method does not use atomic write operations nor does it disable multitasking, is this okay? Will compiler always produce code that will use long writes?
git-svn-id: https://svn.aros.org/svn/aros/trunk/AROS@49536 fb15a70f-31f2-0310-bbcc-cdcc74a49acc
Making sure the CAM methods are not used if register value is in the extended configuration space,
there are some chipsets that can access extended configuration space with CAM method, but we don't have any support for that.
Pointers to extended configuration space registers made volatile.
CAM method 2 was missing alignement to 32-bit long all along.
ECAM write method does not use atomic write operations nor does it disable multitasking, is this okay? Will compiler always produce code that will use long writes?
git-svn-id: https://svn.aros.org/svn/aros/trunk/AROS@49536 fb15a70f-31f2-0310-bbcc-cdcc74a49acc