upgrade libdrm-2.4.75 to libdrm-2.4.109 #25
Conversation
klausz65
force-pushed
the
update-to-2.4.109
branch
3 times, most recently
from
6c80f5e to
d009fb7
Compare
|
To the best of my knowledge, there is no gfx_private for Sparc, so the LDFLAGS on line 50 would at best cause a load failure (would be a pain to port given gfx_private's intensive use of x86-specific facilities). And I am unaware of any Sparc frame buffer driver that is still around that uses DRM (could possibly be one in the future, but wouldn't it be better to cross that bridge then?). |
klausz65
force-pushed
the
update-to-2.4.109
branch
3 times, most recently
from
8558cca to
c089c8f
Compare
klausz65
force-pushed
the
update-to-2.4.109
branch
from
c089c8f to
0e3b7dc
Compare
|
There's a lot here. Let me look at this later. Thanks. |
|
|
||
| 2. Implement dynamic creation of the new header file: | ||
|
|
||
| - generated_static_table_fourcc.h |
There was a problem hiding this comment.
I generally causes less problems for people if you go ahead and check in the generated file,
and provide a make rule to (optionally) verify that regenerating the file produces the same.
This is the same approach described in UPDATE_LIB.
Oh, looks like you did "git add" that file, right?
Is there a rule to verify if matches what gets generated?
| @@ -179,7 +179,7 @@ ELFEDIT= /usr/bin/elfedit | |||
| ELFSIGN= /usr/bin/elfsign | |||
| DTRACE= /usr/sbin/dtrace -xnolibs | |||
| UNIQ= /usr/bin/uniq | |||
| TAR= /usr/bin/tar | |||
| TAR= /usr/bin/gtar | |||
There was a problem hiding this comment.
Was there a problem with plain old tar?
There was a problem hiding this comment.
I see, we need to unpack an xz file now.
| + * Don't have DRM_CLONEID_NBITS here. | ||
| + */ | ||
| + d &= ~0x3fe00; | ||
| +#endif |
There was a problem hiding this comment.
I'm a bit worried about the removal of these two changes.
When I latest tested this, the library would not correctly determine the type of device was open on some FD because the "clone open" bits cause the minor number to (usually) not match.
I forget which test demonstrates this. Do you need help tracking that down?
I think it was the "drmdevice" test (which we build under cmd/misc1)
| + err = _sun_drm_find_device(min, &path); | ||
| + if (err == 0) | ||
| + return (path); | ||
| + /* else ... (ret NULL at endif) */ |
There was a problem hiding this comment.
This is the 2nd part I referred to above.
There was a problem hiding this comment.
I added both those parts back for you in #26
And add initial SPARC support for drm module