1 |
Mark Knecht posted on Tue, 03 Sep 2013 13:25:15 -0700 as excerpted: |
2 |
|
3 |
> I suspect I can disassociate the CUDA stuff if I have to. I don't have |
4 |
> to tell X11 that the 465 is in the machine, and I have 3 PCI Express |
5 |
> slots so I could potentially have 2 VGAs and 1 CUDA. (I think...) |
6 |
|
7 |
There's actually some brand new developments in that regard -- hit the |
8 |
Linux news in the last week, and scheduled for 3.12. |
9 |
|
10 |
So what's the big deal? They're splitting the formerly single-device KMS/ |
11 |
DRM into two separate devices, one of which will be render/compute-only, |
12 |
and thus require lower privs -- classic Unix user/group file permissions, |
13 |
not the root or CAP_SYS_ADMIN that the current device requires, and a |
14 |
second that will be mode-setting/display-controller only, accessible via |
15 |
clone from the current DRM-control node (which is currently unused). |
16 |
Because no new resources are created -- they're simply cloned from the |
17 |
existing control node, required privs here can be reduced as well. The |
18 |
practical effect of the latter will be another step toward allowing user- |
19 |
priv-only X (and wayland). |
20 |
|
21 |
But the former is bigger news in srovider. For just $5 per month, you can |
22 |
get yourself a Cloud server with 512 MB of RAM, 20 GB super-fast SSD, |
23 |
free snapshots, plus backups for a minimal fee. All via a simple |
24 |
graphical interface. |
25 |
|
26 |
And by signing up with this referral link, you can help support this |
27 |
website. |
28 |
|
29 |
If you are reading this, your ad could also be occupying this space. |
30 |
Contact us to make it happen. |
31 |
2 Comments |
32 |
|
33 |
ome ways, as it means any app running as a user/group with suitable file |
34 |
permissions will be able to open compute nodes, no root or CAP_SYS_ADMIN |
35 |
required. That will in turn dramatically open up the possibilities for |
36 |
"ordinary application" use of compute-nodes, thus allowing pretty much |
37 |
any app to use those resources, instead of forcing the severe privilege |
38 |
restrictions currently needed to avoid huge security issues, currently. |
39 |
|
40 |
So why is all this happening now? It turns out that while on x86/amd64, |
41 |
it has historically been the same hardware device handling both rendering |
42 |
and mode-setting, on arm, which is where the big money is now since |
43 |
that's what all the mobile devices seem to run on, these functions are |
44 |
often split into two different pieces of hardware. The existing unified |
45 |
setup could be made to work, but it wasn't a particularly good fit, while |
46 |
the new setup is a natural fit. |
47 |
|
48 |
Meanwhile, while they could have solved their own little hardware problem |
49 |
in their corner of the computing world in just their own drivers, the |
50 |
bigger solution was to split the functionality up in general, thus |
51 |
allowing solutions to a bunch of thorny problems with the existing |
52 |
unified hardware as well. =:^) In addition to the security and compute |
53 |
resource availability issues mentioned above that this solves, it also |
54 |
makes solving problems such as the newer embedded/dedicated gpu pairing |
55 |
on laptops, with the display only attached to the embedded, thus making |
56 |
the dedicated card that's supposed to activate for games and stuff |
57 |
useless, under the current unified setup. Similarly, it'll now be |
58 |
possible to run different user X sessions on different displays driven by |
59 |
the same GPU hardware -- multi-seat-on- single-gpu-multi-display, which |
60 |
currently isn't possible, because only one app can be DRM-master on a gpu |
61 |
at once, and that app must be run by one user, it can't be shared. |
62 |
|
63 |
So the separate render/modeset hardware on arm is driving better and far |
64 |
more flexible solutions for (normally) unified render/modeset hardware on |
65 |
x86/amd64, too! =:^) |
66 |
|
67 |
Original blog announcement: |
68 |
|
69 |
https://dvdhrm.wordpress.com/2013/09/01/splitting-drm-and-kms-device- |
70 |
nodes/ |
71 |
|
72 |
Initial LWN coverage. (Not much there besides the single paragraph |
73 |
pointer to the above at present, but kernel devs often comment there, and |
74 |
LWN will very likely have a feature article covering it in some depth in |
75 |
one of its coming weekly editions, too.) |
76 |
|
77 |
http://lwn.net/Articles/565464 |
78 |
|
79 |
Phoronix (pre-blog announcement coverage): |
80 |
|
81 |
http://www.phoronix.com/scan.php?page=news_item&px=MTQ0OTI |
82 |
|
83 |
(earlier...) |
84 |
|
85 |
http://www.phoronix.com/scan.php?page=news_item&px=MTQ0MzQ |
86 |
|
87 |
(earlier still, July...) |
88 |
|
89 |
http://www.phoronix.com/scan.php?page=news_item&px=MTQwNTA |
90 |
|
91 |
-- |
92 |
Duncan - List replies preferred. No HTML msgs. |
93 |
"Every nonfree program has a lord, a master -- |
94 |
and if you use the program, he is your master." Richard Stallman |