| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| tfiga | mchehab: aha, good catch! thanks, I can see the patches there now.
that said, probably with the series feature now I won't have too much opportunity to use bundles anymore :) | [03:00] |
| ............................................ (idle for 3h39mn) | ||
| hverkuil | bparrot: the v4l-utils tag is just the git repo commit hash. The commit you mentioned is in there. Or am I misunderstanding the question? | [06:40] |
| ........ (idle for 35mn) | ||
| bbrezillon | hverkuil: Hi
hverkuil: can we access vb2_q->bufs[i] directly? looks like the cedrus driver has a private array containing the exact same vb2 bufs it's using vb2_find_timestamp() to get the index of the buffer in the vb2 queue, and then use this index to find the same buf in its private buf array | [07:15] |
| hverkuil | It's marked in videobuf2-core.h as for internal use only. | [07:17] |
| bbrezillon | could we add an helper returning a vb2 buf based on the index then? | [07:18] |
| hverkuil | I wouldn't access it directly, but you can make a function that returns it. | [07:18] |
| bbrezillon | *based on the timestamp
okay something like vb2_get_buffer_by_timestamp()? | [07:18] |
| hverkuil | No. vb2_find_timestamp returns the buffer index. So you put that in the vb2_get_buffer(int idx) helper.
it returns NULL if idx < 0, >= num_buffers, and otherwise the contents of bufs[idx] | [07:20] |
| bbrezillon | oh, right
I didn't notice this function sorry, you're suggesting to add a new helper called vb2_get_buffer() | [07:22] |
| hverkuil | right | [07:22] |
| bbrezillon | that adds an extra step, but okay | [07:22] |
| hverkuil | That buffer array is likely to change at some point when we get rid of the 32 buffer limit. So I rather have a static inline access function instead of directly accessing it.
The additional sanity checks are also useful to have. | [07:23] |
| bbrezillon | oh, I couldn't agree more
(about using wrappers instead of directly accessing the array) my comment was related to the 2-steps approach, while we could get the buffer using one function call I mean, the code has to iterate over all vb2 bufs in the queue anyway to test the ->timestamp field so we could have a function that returns this vb2 buf directly instead of first retrieving the index and then using another helper that returns the buffer at this index anyway, it's not a big deal | [07:25] |
| .......... (idle for 49mn) | ||
| LazyGrizzly | ping pinchartl | [08:17] |
| ............. (idle for 1h2mn) | ||
| bbrezillon | Kwiboo: can you point me to the code that uses l1/l0 modification lists to get the refs for the next frame?
my understanding was that those 2 things were unrelated, but I might have missed something | [09:19] |
| ................ (idle for 1h18mn) | ||
| hverkuil | pH5: it turns out we have i.MX8MM EVK boards in-house. So I got one and once I find the time to set it up I should be able to test the HW codec support. | [10:38] |
| ..... (idle for 20mn) | ||
| Kwiboo | bbrezillon, you are correct, I mixed up the marking and modification process :-) | [10:58] |
| ................ (idle for 1h18mn) | ||
| *** | LazyGrizzly has left | [12:16] |
| .... (idle for 16mn) | ||
| bbrezillon | hverkuil: I (or ezequielg) will try to test the rk bits
so we can have patches 1 to 7 merged and start applying other things on top (ezequielg is preparing a VP8 series) | [12:32] |
| hverkuil | I'll probably do another day of processing patches tomorrow. They keep coming in at high speed :-) | [12:37] |
| ......... (idle for 44mn) | ||
| bparrot | hverkuil, i had noticed that tag v4l-utils-1.16.4 and v4l-utils-1.16.5 were created after the "fix multiplanar handling" commit, i was expecting them to include it but they don't | [13:21] |
| hverkuil | bparrot: I don't make release tags, that's Gregor Jasny.
He every so often tags new releases. for distros | [13:22] |
| bparrot | i realized that, was just wondering what one had to do with the other. But ok
no problem | [13:23] |
| .................... (idle for 1h35mn) | ||
| *** | jernej_ has left | [14:58] |
| ........ (idle for 39mn) | ||
| benjiG has left | [15:37] | |
| ..... (idle for 22mn) | ||
| ndufresne | pH5, be aware that I'm now merging this, https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/merge_requests/132
as a side effect, it may break some use case on kernels where the profile/level enumeration is bugged for driver that don't implement these controls for decoder, it has no effect | [15:59] |
| ......... (idle for 40mn) | ||
| ezequielg | tfiga: series feature rock. the only thing i'm missing is having a URL link in each patch, that links to its series. | [16:39] |
| ............................ (idle for 2h16mn) | ||
| pH5 | hverkuil: that would be awesome, I don't think I'll get access to any i.MX8M Mini board soon.
ndufresne: thank you, I suppose this will be broken for coda up to v5.1, v5.2 will contain the level control fix. | [18:55] |
| ndufresne | pH5, yes, but we are pretty far from GStreamer 1.18 anyway
so far it's been around 12 months per release (that's twice the target, but we have no one full time doing maintenance so ...) | [18:57] |
| ............... (idle for 1h14mn) | ||
| *** | jernej_ has left | [20:12] |
| .......................... (idle for 2h5mn) | ||
| tfiga has quit IRC (Ping timeout: 258 seconds)
rellla has quit IRC (Ping timeout: 246 seconds) | [22:17] | |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |