| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| *** | Hamoondancer has left | [01:13] |
| ........................ (idle for 1h55mn) | ||
| Ghalnuls has left | [03:08] | |
| .... (idle for 16mn) | ||
| paulk-leonov has quit IRC (Ping timeout: 240 seconds) | [03:24] | |
| ayaka has quit IRC (Ping timeout: 244 seconds) | [03:31] | |
| Imphuh has left | [03:39] | |
| ............. (idle for 1h2mn) | ||
| indy has quit IRC (Quit: ZNC - http://znc.sourceforge.net) | [04:41] | |
| ................. (idle for 1h22mn) | ||
| lexano has quit IRC (Ping timeout: 268 seconds) | [06:03] | |
| ............... (idle for 1h13mn) | ||
| hverkuil | mchehab: please pick up https://patchwork.linuxtv.org/patch/53375/ It fixes a smatch warning and it is quite old (Dec 7) so I don't know why it hasn't been merged yet. | [07:16] |
| .... (idle for 17mn) | ||
| wens | ezequielg: noob question, why do we need separate nodes for encoding vs decoding? | [07:33] |
| ..... (idle for 22mn) | ||
| hverkuil | kbingham: pinchartl: regarding this smatch warning for vsp1:
/home/hans/work/build/media-git/drivers/media/platform/vsp1/vsp1_drm.c: /home/hans/work/build/media-git/drivers/ media/platform/vsp1/vsp1_drm.c:336 vsp1_du_pipeline_setup_brx() error: we previously assumed 'pipe->brx' could b e null (see line 244) This warning is correct, and line 336 can do a NULL dereference: format.pad = pipe->brx->source_pad; I suspect it should be brx->source_pad. Actually, looking at the big 'if' statement above that line I see that at the end it sets pipe->brx = brx. So using brx instead of pipe->brx in the code below the if seems to be the right thing to do. So the warning is actually not right, but the code is too complex for smatch to understand it. I'll make a patch to fix this, I really want smatch to stop generating warnings. | [07:55] |
| ........... (idle for 51mn) | ||
| kbingham | hverkuil, Hi, I recall Mauro found an issue with smatch in this area producing a false positive. | [08:51] |
| hverkuil | yes, but that patch is outdated. | [08:52] |
| ...... (idle for 27mn) | ||
| gtucker | mchehab, hverkuil: hello, could you please take a look at the kernelci test report email update and let me know if that's OK to be turned on in production?
"Re: Test results (v4l2) for media/master - v4.20-rc5-281-gd2b4387f3bdf" | [09:19] |
| pinchartl | hverkuil: I've discussed this with Dan Carpenter. if I recall correctly it's not an issue of code complexity
but it's about smatch not knowing that vsp1->bru is guaranteed to be non-null I don't remember all the details though I'll review your patch | [09:32] |
| hverkuil | gtucker: can you post a newer version? You made a change to address Mauro's concern (not showing what the test was for), and it helps if we see that version of the output. And as mentioned at FOSDEM, the order of the tests is reversed, and that definitely needs to be fixed. | [09:34] |
| pinchartl | the one for the vsp1 may be fine
I don't like the one for the omap3isp though | [09:34] |
| gtucker | hverkuil: ok I'll fix the test case order and send another update, thanks | [09:40] |
| hverkuil | gtucker: much appreciated! | [09:41] |
| ..................... (idle for 1h43mn) | ||
| *** | gmikeg has left | [11:24] |
| ........... (idle for 52mn) | ||
| ageis has quit IRC (Quit: exit(1); echo 'https://cointel.pro' > /dev/null; x-www-browser 'https://twitter.com/ageis') | [12:16] | |
| hverkuil | ezequielg: ping | [12:28] |
| ......... (idle for 42mn) | ||
| koike: ping | [13:10] | |
| ....... (idle for 34mn) | ||
| ezequielg | hverkuil: pong | [13:44] |
| hverkuil | ezequielg: brb | [13:44] |
| ezequielg | i'm here
wens: i guess you don't absolutely need them. but it's really messy to implement. and also, i don't think there is any harm in the split. | [13:45] |
| hverkuil | ezequielg: I must be missing something, but why couldn't you use the regular mem2mem v4l2_m2m_register_media_controller() function for the rockchip driver? | [13:51] |
| ezequielg | right.
because it supports only one set of pad->proc->pad unless we extend that, to N. | [13:52] |
| hverkuil | In your cover letter I see only one input and one output pad for the decoder entity (and ditto for the encoder).
Is the cover letter wrong? | [13:54] |
| ezequielg | no, it's correct. | [13:55] |
| wens | ezequielg: I guess my question would be, what's bad in having just one video node for both encoding and decoding? | [13:56] |
| ezequielg | there is one set for the decoder, and one set for the encoder. both related to the same mem2mem. | [13:56] |
| hverkuil | vicodec does that as well, and it just uses v4l2_m2m_register_media_controller(). | [13:56] |
| ezequielg | but it uses two mem2mem.
the decoder and the encoder are independent in vicodec. they don't get queued via the same mem2mem device. wens: well, consider the ambiguity created by having two pads, and the role of the pad depends on the configuration the user does. the codec API I think would allow this case. but, at least when I tried to implement it, things got much more complex by having to support encoded formats on both ends. | [13:56] |
| *** | svarbanov has quit IRC (Ping timeout: 240 seconds) | [14:10] |
| hverkuil | ezequielg: ah, now I see it. It was confusing because it is only in patch 8 that the decoder is hooked up in the mc.
So without reading patch 8 as well, patch 6 just looks weird. | [14:11] |
| ezequielg | the code could use some cleaning i guess.
does it? lemme check. ah. right. | [14:11] |
| wens | ezequielg: I see. yes that would involve a lot of prodding from userspace | [14:22] |
| ezequielg | wens: right. the userspace enumeration gets funkier. | [14:24] |
| .... (idle for 19mn) | ||
| hverkuil | koike: ping | [14:43] |
| ...... (idle for 26mn) | ||
| koike | hverkuil: pong | [15:09] |
| hverkuil | koike: can you take a quick look at https://patchwork.linuxtv.org/patch/54223/? This vimc patch is also pending, but that's not urgent: https://patchwork.linuxtv.org/patch/53902/ | [15:12] |
| *** | codyps has quit IRC (Quit: quit) | [15:12] |
| koike | hverkuil: sure
hverkuil: could you also take a look on this please https://patchwork.linuxtv.org/patch/54091/ ? | [15:13] |
| hverkuil | That's already in a pending pull request.
"Under Review" means - at least for my workflow - that it is part of a pull request that hasn't been merged yet (pending Mauro's final code review) | [15:17] |
| koike | hverkuil: ah ok, thanks, I need to learn to read the state field in patchwork | [15:23] |
| .................... (idle for 1h35mn) | ||
| ndufresne | mchehab, good news, https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/issues/562#note_114583 | [16:58] |
| mchehab | good!
does it require approval or was already merged? mchehab knows about nothing about gst committing rules | [17:06] |
| ..... (idle for 23mn) | ||
| *** | macmaN has quit IRC (Quit: ZNC - http://znc.sourceforge.net) | [17:30] |
| mchehab | hverkuil: there's now a new warning on vimc:
drivers/media/platform/vimc/vimc-sensor.c: In function 'vimc_sen_process_frame': drivers/media/platform/vimc/vimc-sensor.c:208:15: warning: variable 'frame_size' set but not used [-Wunused-but-set-variable] unsigned int frame_size; ^~~~~~~~~~ either something was dropped by mistake or the code require some cleanup | [17:31] |
| ..... (idle for 20mn) | ||
| hverkuil | why didn't I get that warning? Do you add additional compiler flags or something? | [17:53] |
| ah, it was a sparse error. Still doesn't explain why I missed it... | [18:04] | |
| ............................ (idle for 2h17mn) | ||
| ezequielg | ndufresne: that's great! | [20:21] |
| ndufresne | I just need to make the patches and push it | [20:24] |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |