1 |
unsubscribe |
2 |
|
3 |
On Wed, Mar 21, 2012 at 5:03 AM, <gentoo-amd64+help@l.g.o> wrote: |
4 |
> Topics (messages 13110 through 13125): |
5 |
> |
6 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
7 |
> 13110 - Frank Peters <frank.peters@×××××××.net> |
8 |
> |
9 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
10 |
> 13111 - Mark Knecht <markknecht@×××××.com> |
11 |
> |
12 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
13 |
> 13112 - Frank Peters <frank.peters@×××××××.net> |
14 |
> |
15 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
16 |
> 13113 - Cheng Renquan <crquan@×××××.com> |
17 |
> |
18 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
19 |
> 13114 - Mark Knecht <markknecht@×××××.com> |
20 |
> |
21 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
22 |
> 13115 - Mark Knecht <markknecht@×××××.com> |
23 |
> |
24 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
25 |
> 13116 - Frank Peters <frank.peters@×××××××.net> |
26 |
> |
27 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
28 |
> 13117 - Mark Knecht <markknecht@×××××.com> |
29 |
> |
30 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
31 |
> 13118 - Paul Hartman <paul.hartman+gentoo@×××××.com> |
32 |
> |
33 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
34 |
> 13119 - Barry Schwartz <chemoelectric@×××××××××××××.org> |
35 |
> |
36 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
37 |
> 13120 - Mark Knecht <markknecht@×××××.com> |
38 |
> |
39 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
40 |
> 13121 - Barry Schwartz <chemoelectric@×××××××××××××.org> |
41 |
> |
42 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
43 |
> 13122 - Frank Peters <frank.peters@×××××××.net> |
44 |
> |
45 |
> [gentoo-amd64] Kernel-3.3.0 and Nvidia-drivers |
46 |
> 13123 - Barry Schwartz <chemoelectric@×××××××××××××.org> |
47 |
> |
48 |
> [gentoo-amd64] Re: Kernel-3.3.0 and Nvidia-drivers |
49 |
> 13124 - Duncan <1i5t5.duncan@×××.net> |
50 |
> 13125 - Duncan <1i5t5.duncan@×××.net> |
51 |
> |
52 |
> |
53 |
> |
54 |
> Linux-3.3.0 is released and, as is my usual habit, I downloaded and compiled |
55 |
> the plain vanilla source. |
56 |
> |
57 |
> After rebooting to the command console, and before starting X, I needed |
58 |
> to re-install the nvidia-driver module for the new kernel. Doing |
59 |
> "emerge nvidia-drivers" gave me the following error (the entire build log |
60 |
> is attached): |
61 |
> |
62 |
>>>> Source prepared. |
63 |
>>>> Configuring source in /tmp/portage-acc/tmp/portage/x11-drivers/nvidia-drivers-295.20-r1/work ... |
64 |
>>>> Source configured. |
65 |
>>>> Compiling source in /tmp/portage-acc/tmp/portage/x11-drivers/nvidia-drivers-295.20-r1/work ... |
66 |
> Preparing nvidia module |
67 |
> make -j9 HOSTCC=x86_64-pc-linux-gnu-gcc CROSS_COMPILE=x86_64-pc-linux-gnu- 'LDFLAGS=-m elf_x86_64' IGNORE_CC_MISMATCH=yes V=1 SYSSRC=/usr/src/linux SYSOUT=/lib/modules/3.3.0/build CC=x86_64-pc-linux-gnu-gcc clean module |
68 |
> |
69 |
> *** Unable to determine the target kernel version. *** |
70 |
> |
71 |
> make: *** [select_makefile] Error 1 |
72 |
> emake failed |
73 |
> ERROR: x11-drivers/nvidia-drivers-295.20-r1 failed (compile phase): |
74 |
> Unable to emake HOSTCC=x86_64-pc-linux-gnu-gcc CROSS_COMPILE=x86_64-pc-linux-gnu- LDFLAGS=-m elf_x86_64 IGNORE_CC_MISMATCH=yes V=1 SYSSRC=/usr/src/linux SYSOUT=/lib/modules/3.3.0/build CC=x86_64-pc-linux-gnu-gcc clean module |
75 |
> |
76 |
> Call stack: |
77 |
> ebuild.sh, line 85: Called src_compile |
78 |
> environment, line 3711: Called linux-mod_src_compile |
79 |
> environment, line 2668: Called die |
80 |
> The specific snippet of code: |
81 |
> eval "emake HOSTCC=\"$(tc-getBUILD_CC)\" |
82 |
> |
83 |
> |
84 |
> The first part of the build log (see attached file) indicates that the |
85 |
> kernel source was correctly found (at /usr/src/linux), but for some |
86 |
> reason the make process fails. |
87 |
> |
88 |
> Could this be due to some changes in the kernel-3.3.0 tree? |
89 |
> |
90 |
> Frank Peters |
91 |
> |
92 |
> |
93 |
> On Mon, Mar 19, 2012 at 9:12 AM, Frank Peters <frank.peters@×××××××.net> wrote: |
94 |
>> Linux-3.3.0 is released and, as is my usual habit, I downloaded and compiled |
95 |
>> the plain vanilla source. |
96 |
>> |
97 |
> <SNIP> |
98 |
>> The first part of the build log (see attached file) indicates that the |
99 |
>> kernel source was correctly found (at /usr/src/linux), but for some |
100 |
>> reason the make process fails. |
101 |
>> |
102 |
>> Could this be due to some changes in the kernel-3.3.0 tree? |
103 |
>> |
104 |
>> Frank Peters |
105 |
>> |
106 |
> |
107 |
> Possibly this from Google will help? |
108 |
> |
109 |
> Good luck & post back, |
110 |
> Mark |
111 |
> |
112 |
> On Mon, 19 Mar 2012 09:35:34 -0700 |
113 |
> Mark Knecht <markknecht@×××××.com> wrote: |
114 |
> |
115 |
>> |
116 |
>> Possibly this from Google will help? |
117 |
>> |
118 |
> |
119 |
> Did you intend to include a link? If so, it must have gotten |
120 |
> lost somewhere. |
121 |
> |
122 |
> Frank Peters |
123 |
> |
124 |
> |
125 |
> I have switched to nouveau driver, for over 1 year, |
126 |
> using with gnome3 interface, no performance issue, and quite stable, |
127 |
> |
128 |
> enjoying every mainline kernel 3.0, 3.1, 3.2, 3.3, no building issue |
129 |
> |
130 |
> |
131 |
> I think it's time for everyone to give up nvidia-drivers, the blob of binary |
132 |
> |
133 |
> On Mon, Mar 19, 2012 at 9:12 AM, Frank Peters <frank.peters@×××××××.net> wrote: |
134 |
>> Linux-3.3.0 is released and, as is my usual habit, I downloaded and compiled |
135 |
>> the plain vanilla source. |
136 |
>> |
137 |
>> After rebooting to the command console, and before starting X, I needed |
138 |
>> to re-install the nvidia-driver module for the new kernel. Doing |
139 |
>> "emerge nvidia-drivers" gave me the following error (the entire build log |
140 |
>> is attached): |
141 |
>> |
142 |
>>>>> Source prepared. |
143 |
>>>>> Configuring source in /tmp/portage-acc/tmp/portage/x11-drivers/nvidia-drivers-295.20-r1/work ... |
144 |
>>>>> Source configured. |
145 |
>>>>> Compiling source in /tmp/portage-acc/tmp/portage/x11-drivers/nvidia-drivers-295.20-r1/work ... |
146 |
>> Preparing nvidia module |
147 |
>> make -j9 HOSTCC=x86_64-pc-linux-gnu-gcc CROSS_COMPILE=x86_64-pc-linux-gnu- 'LDFLAGS=-m elf_x86_64' IGNORE_CC_MISMATCH=yes V=1 SYSSRC=/usr/src/linux SYSOUT=/lib/modules/3.3.0/build CC=x86_64-pc-linux-gnu-gcc clean module |
148 |
>> |
149 |
>> *** Unable to determine the target kernel version. *** |
150 |
> |
151 |
> On Mon, Mar 19, 2012 at 12:05 PM, Frank Peters <frank.peters@×××××××.net> wrote: |
152 |
>> On Mon, 19 Mar 2012 09:35:34 -0700 |
153 |
>> Mark Knecht <markknecht@×××××.com> wrote: |
154 |
>> |
155 |
>>> |
156 |
>>> Possibly this from Google will help? |
157 |
>>> |
158 |
>> |
159 |
>> Did you intend to include a link? If so, it must have gotten |
160 |
>> lost somewhere. |
161 |
>> |
162 |
>> Frank Peters |
163 |
>> |
164 |
>> |
165 |
> |
166 |
> Gadzooks, yes I did. Sorry! |
167 |
> |
168 |
> http://weltall.heliohost.org/wordpress/2012/01/20/linux-kernel-3-3-rc1-and-nvidia-drivers/ |
169 |
> |
170 |
> To be clear, I have no idea if it will solve your exact problem. It |
171 |
> didn't for all. |
172 |
> |
173 |
> My experience with nvidia is pretty good if you give them a couple of |
174 |
> days, but very spotty on the first or second day a new kernel minor |
175 |
> rev is released. Nominally I use this page: |
176 |
> |
177 |
> http://www.nvidia.com/Download/indexsg.aspx?lang=en-us |
178 |
> |
179 |
> and then search out my specific card to determine what _exact_ driver |
180 |
> they want me to run. It almost always requires i run ~amd64. |
181 |
> |
182 |
> HTH, |
183 |
> Mark |
184 |
> |
185 |
> On Mon, Mar 19, 2012 at 12:14 PM, Cheng Renquan <crquan@×××××.com> wrote: |
186 |
>> I have switched to nouveau driver, for over 1 year, |
187 |
>> using with gnome3 interface, no performance issue, and quite stable, |
188 |
>> |
189 |
>> enjoying every mainline kernel 3.0, 3.1, 3.2, 3.3, no building issue |
190 |
>> |
191 |
>> |
192 |
>> I think it's time for everyone to give up nvidia-drivers, the blob of binary |
193 |
>> |
194 |
> |
195 |
> And how would one use CUDA if we gave up nvidia-drivers? Does nouveau |
196 |
> support all the CUDA stuff? |
197 |
> |
198 |
> Cheers, |
199 |
> Mark |
200 |
> |
201 |
> On Mon, 19 Mar 2012 09:35:34 -0700 |
202 |
> Mark Knecht <markknecht@×××××.com> wrote: |
203 |
> |
204 |
>> |
205 |
>> Possibly this from Google will help? |
206 |
>> |
207 |
> |
208 |
> OK, I think I've found the URL that you may have intended: |
209 |
> |
210 |
> http://weltall.heliohost.org/wordpress/2012/01/20/linux-kernel-3-3-rc1-and-nvidia-drivers/ |
211 |
> |
212 |
> I followed the second fix that is described and now the nvidia-drivers |
213 |
> emerge and work nicely with linux kernel-3.3. |
214 |
> |
215 |
> Thanks for the hint. |
216 |
> |
217 |
> There are changes to the kernel-3.3 tree that will have to addressed |
218 |
> by the gentoo nvidia-drivers or other users will hit this problem |
219 |
> as well. |
220 |
> |
221 |
> Frank Peters |
222 |
> |
223 |
> |
224 |
> On Mon, Mar 19, 2012 at 12:30 PM, Frank Peters <frank.peters@×××××××.net> wrote: |
225 |
>> On Mon, 19 Mar 2012 09:35:34 -0700 |
226 |
>> Mark Knecht <markknecht@×××××.com> wrote: |
227 |
>> |
228 |
>>> |
229 |
>>> Possibly this from Google will help? |
230 |
>>> |
231 |
>> |
232 |
>> OK, I think I've found the URL that you may have intended: |
233 |
>> |
234 |
>> http://weltall.heliohost.org/wordpress/2012/01/20/linux-kernel-3-3-rc1-and-nvidia-drivers/ |
235 |
>> |
236 |
>> I followed the second fix that is described and now the nvidia-drivers |
237 |
>> emerge and work nicely with linux kernel-3.3. |
238 |
>> |
239 |
>> Thanks for the hint. |
240 |
>> |
241 |
>> There are changes to the kernel-3.3 tree that will have to addressed |
242 |
>> by the gentoo nvidia-drivers or other users will hit this problem |
243 |
>> as well. |
244 |
>> |
245 |
>> Frank Peters |
246 |
>> |
247 |
>> |
248 |
> |
249 |
> Glad you found it. That is the link I just sent along and good to know |
250 |
> it helped a bit. |
251 |
> |
252 |
> As I said, on kernel revision number changes I ALWAYS wait a minimum |
253 |
> of 3-5 days before trying to upgrade. nvidia always seems to need at |
254 |
> least a few days to catch up with a new release, or that's been my |
255 |
> experience in the past anyway. |
256 |
> |
257 |
> Cheers, |
258 |
> Mark |
259 |
> |
260 |
> On Mon, Mar 19, 2012 at 3:08 PM, Mark Knecht <markknecht@×××××.com> wrote: |
261 |
>> |
262 |
>> As I said, on kernel revision number changes I ALWAYS wait a minimum |
263 |
>> of 3-5 days before trying to upgrade. nvidia always seems to need at |
264 |
>> least a few days to catch up with a new release, or that's been my |
265 |
>> experience in the past anyway. |
266 |
> |
267 |
> Same here, nvidia-drivers and vmware-modules are why I usually wait |
268 |
> until the .1 release on a new kernel series. :) |
269 |
> |
270 |
> Mark Knecht <markknecht@×××××.com> skribis: |
271 |
>> And how would one use CUDA if we gave up nvidia-drivers? Does nouveau |
272 |
>> support all the CUDA stuff? |
273 |
> |
274 |
> I think they may have reverse engineered CUDA, but whether it is |
275 |
> implemented I dont know, and I wouldnt want to be the one to test |
276 |
> it, either. :) |
277 |
> |
278 |
> Im donating processor time with boinc and thats good enough reason |
279 |
> to use the binary driver if it gives me more processor to donate. |
280 |
> |
281 |
> (Waiting before installing anything new is painful, I know, because |
282 |
> with Gentoo you always feel like you want it to be doing something, or |
283 |
> at least thats my experience. :) ) |
284 |
> |
285 |
> On Mon, Mar 19, 2012 at 2:28 PM, Barry Schwartz |
286 |
> <chemoelectric@×××××××××××××.org> wrote: |
287 |
>> Mark Knecht <markknecht@×××××.com> skribis: |
288 |
>>> And how would one use CUDA if we gave up nvidia-drivers? Does nouveau |
289 |
>>> support all the CUDA stuff? |
290 |
>> |
291 |
>> I think they may have reverse engineered CUDA, but whether it is |
292 |
>> implemented I dont know, and I wouldnt want to be the one to test |
293 |
>> it, either. :) |
294 |
>> |
295 |
>> Im donating processor time with boinc and thats good enough reason |
296 |
>> to use the binary driver if it gives me more processor to donate. |
297 |
>> |
298 |
>> (Waiting before installing anything new is painful, I know, because |
299 |
>> with Gentoo you always feel like you want it to be doing something, or |
300 |
>> at least thats my experience. :) ) |
301 |
>> |
302 |
> |
303 |
> I have no problem with you or anyone else running nouveau. None at all. |
304 |
> |
305 |
> I do have a problem with someone saying 'it's time for everyone to |
306 |
> give up nvidia-drivers' without demonstrating they know the full |
307 |
> impact of what they are suggesting. Had Cheng Renquan suggest 'Maybe |
308 |
> people who don't need anything other than basic X capabilities should |
309 |
> consider running nouveau' I wouldn't have even chimed in. |
310 |
> |
311 |
> And thanks for running boinc. It's good of you and good for you. |
312 |
> |
313 |
> Cheers, |
314 |
> Mark |
315 |
> |
316 |
> Mark Knecht <markknecht@×××××.com> skribis: |
317 |
>> I do have a problem with someone saying 'it's time for everyone to |
318 |
>> give up nvidia-drivers' without demonstrating they know the full |
319 |
>> impact of what they are suggesting. Had Cheng Renquan suggest 'Maybe |
320 |
>> people who don't need anything other than basic X capabilities should |
321 |
>> consider running nouveau' I wouldn't have even chimed in. |
322 |
> |
323 |
> Sage Notebook (sage-on-gentoo overlay) can use CUDA as well. I think |
324 |
> you don't have to worry that nvidia-drivers would go away; in the |
325 |
> worst case it would end up in a good overlay. That's unlikely IMO, and |
326 |
> would mostly likely be if there were an established alternative |
327 |
> implementation of CUDA. |
328 |
> |
329 |
> On Mon, 19 Mar 2012 14:40:34 -0700 |
330 |
> Mark Knecht <markknecht@×××××.com> wrote: |
331 |
> |
332 |
>> |
333 |
>> I have no problem with you or anyone else running nouveau. None at all. |
334 |
>> |
335 |
>> I do have a problem with someone saying 'it's time for everyone to |
336 |
>> give up nvidia-drivers' without demonstrating they know the full |
337 |
>> impact of what they are suggesting. Had Cheng Renquan suggest 'Maybe |
338 |
>> people who don't need anything other than basic X capabilities should |
339 |
>> consider running nouveau' I wouldn't have even chimed in. |
340 |
>> |
341 |
> |
342 |
> I think the main complaint is that the Nvidia drivers are closed |
343 |
> binary blobs while Nouveau is an attempt to create an entirely |
344 |
> open source driver. The spirit of free software demands that we |
345 |
> support these open initiatives. |
346 |
> |
347 |
> Actually, until recently, I was very happy using the legacy open |
348 |
> source xorg nvidia driver known simply as as "nv." It is still |
349 |
> available in Gentoo as x11-drivers/xf86-video-nv. |
350 |
> |
351 |
> Since I do not do gaming or other intensive graphic activities, |
352 |
> I could get by nicely with just nv. However, the nv driver does |
353 |
> not seem to support the latest hardware too well. On one of |
354 |
> my machines I have a card based on Nvidia GeForce 210. Although |
355 |
> the nv driver works with this card, it does not enable the Xv |
356 |
> extension and consequently the graphic response is poor. |
357 |
> |
358 |
> On older cards nv does work quite well for ordinary purposes. |
359 |
> |
360 |
> Frank Peters |
361 |
> |
362 |
> |
363 |
> Frank Peters <frank.peters@×××××××.net> skribis: |
364 |
>> I think the main complaint is that the Nvidia drivers are closed |
365 |
>> binary blobs while Nouveau is an attempt to create an entirely |
366 |
>> open source driver. |
367 |
> |
368 |
> Unfortunately the nvidia drivers are needed not only for graphics but |
369 |
> also to use the GPU for scientific calculations, including for |
370 |
> donating processor time to medical research, for instance. |
371 |
> |
372 |
> |
373 |
> Frank Peters posted on Tue, 20 Mar 2012 00:03:06 -0400 as excerpted: |
374 |
> |
375 |
>> Actually, until recently, I was very happy using the legacy open source |
376 |
>> xorg nvidia driver known simply as as "nv." It is still available in |
377 |
>> Gentoo as x11-drivers/xf86-video-nv. |
378 |
>> |
379 |
>> Since I do not do gaming or other intensive graphic activities, I could |
380 |
>> get by nicely with just nv. However, the nv driver does not seem to |
381 |
>> support the latest hardware too well. |
382 |
> |
383 |
> FWIW, the old nv driver is legacy, now, and won't be supporting new |
384 |
> hardware. AFAIK older hardware should be switching as well, for support |
385 |
> with current software, as nv won't be updated for that any longer, either. |
386 |
> |
387 |
> The same thing happened to it as happened to nvidia's proprietary network |
388 |
> driver some years ago; pretty much everyone involved, including the |
389 |
> distros that used to ship nv and nvidia itself, seems to have recognized |
390 |
> that the reverse-engineered work is now better than nvidia's basic |
391 |
> freedomware stub driver, nv, and nouveau has officially taken its place. |
392 |
> |
393 |
> -- |
394 |
> Duncan - List replies preferred. No HTML msgs. |
395 |
> "Every nonfree program has a lord, a master -- |
396 |
> and if you use the program, he is your master." Richard Stallman |
397 |
> |
398 |
> |
399 |
> Barry Schwartz posted on Tue, 20 Mar 2012 01:20:13 -0500 as excerpted: |
400 |
> |
401 |
>> Unfortunately the nvidia drivers are needed not only for graphics but |
402 |
>> also to use the GPU for scientific calculations, including for donating |
403 |
>> processor time to medical research, for instance. |
404 |
> |
405 |
> My donation ends where their disrespect for human freedom begins. (Over |
406 |
> a decade ago, back on MS, I used to run distributed.net. However, they |
407 |
> don't have a freedomware client, or at least didn't back then, and while |
408 |
> I understand the reasons why -- someone cheating and saying they did the |
409 |
> work when they didn't, just to get the stats -- that didn't change the |
410 |
> fact that I wasn't going to run servantware. But as they say, YMMV.) |
411 |
> |
412 |
> -- |
413 |
> Duncan - List replies preferred. No HTML msgs. |
414 |
> "Every nonfree program has a lord, a master -- |
415 |
> and if you use the program, he is your master." Richard Stallman |
416 |
> |
417 |
> |