The Xserver has moved to libpciaccess for aeons now, so those
old code pathes aren't needed anymore.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
i128_driver.c: In function ‘I128PreInit’:
i128_driver.c:582:13: warning: declaration of ‘from’ shadows a previous local [-Wshadow]
582 | int from = X_DEFAULT;
| ^~~~
i128_driver.c:440:17: note: shadowed declaration is here
440 | MessageType from;
| ^~~~
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Repeated for each call of the CACHED_UPDATE macro:
i128exa.c: In function ‘i128SetBufCtrl’:
i128exa.c:66:74: warning: suggest braces around empty body in an ‘if’ statement [-Wempty-body]
if (CACHE_DEBUG) I128_EXA_DEBUG(("Updated cache for " #reg "\n")); \
^
i128exa.c:89:5: note: in expansion of macro ‘CACHED_UPDATE’
CACHED_UPDATE(buf_ctrl, BUF_CTRL);
^~~~~~~~~~~~~
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
i128accel.c:95:18: warning: ‘split_size’ defined but not used [-Wunused-const-variable=]
static const int split_size[] = { 0x20, 0x10, 0x08, 0x00 };
^~~~~~~~~~
i128accel.c:94:18: warning: ‘max_size’ defined but not used [-Wunused-const-variable=]
static const int max_size[] = { 0x80, 0x40, 0x20, 0x00 };
^~~~~~~~
i128accel.c:93:18: warning: ‘min_size’ defined but not used [-Wunused-const-variable=]
static const int min_size[] = { 0x62, 0x32, 0x1A, 0x00 };
^~~~~~~~
i128accel.c:73:21: warning: ‘i128alu’ defined but not used [-Wunused-const-variable=]
static const CARD32 i128alu[16] =
^~~~~~~
i128exa.c:450:33: warning: ‘dest_formats’ defined but not used [-Wunused-const-variable=]
static const struct dest_format dest_formats[] = {
^~~~~~~~~~~~
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Apparently, xf86GetOptValString changed their return type from char
to const char between X Server 1.15 and 1.16. In particular, X Server
commit e1e01d2e33c632e395d7e396f73fba8ae606b15a made the change. This
commit will suppress the compilation warning.
Signed-off-by: Kevin Brace <kevinbrace@gmx.com>
ABI Version 12 removes support for multiple PCI domains. If you need to
use this driver on a system with more than one PCI domain, you should
either port this driver to using libpciaccess directly or stick with an
older server.
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Where your typical x86 machine silently ignores writes to non-existing
registers, UltraSPARC machines generate a bus error. So we shouldn't
write to the alpha blender setup register on the Imagine 1 and Imagine 2
that don't seem to have the blender unit.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Mark Kettenis <mark.kettenis@xs4all.nl>
The default cursor realization code seems to work just fine. We just
have to set the HARDWARE_CURSOR_NIBBLE_SWAPPED flag on big endian
platforms.
This isn't really surprising since these cards use fairly standard IBM
or TI RAMDACs.
Signed-off-by: Matthieu Herrb <matthieu.herrb@laas.fr>
If the xf86-vide-i128 is compiled with XSERVER_LIBPCIACCESS, registers and
frmaebuffer memory are mapped through libpciaccess. However, I128UnmapMem
tried to unmap things vy using xf86UnmapVidMem. This failed because the memory
wasn't registered (which only happens if you mapped the memory using
xf86MapPciMem.
Fixes bugzilla #20541
Signed-off-by: Matthieu Herrb <matthieu.herrb@laas.fr>
moving all the accel and card members into the driver ptr, filling in
the exa_major/ exa_minor fields, and always using LoadSubModule so we
can check the module version up front. Only tested on ATI.
