-
Notifications
You must be signed in to change notification settings - Fork 253
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Panic in tile_restoration_state on encoding YUV422 sample #1608
Comments
Is there a known-working commit? |
Yes, doing bisect now. |
First bad commit is 5271249 |
nice |
Disable the pretty backtrace, looks like there is a bug there :/ |
The backtrace seems to work on the release build at least. |
I have the same crash trying to encode a frame sized 6820x4320 (4:2:0). I tested the revision prior to the one that broke for this issue, and it seemed to work. My computer cannot handle it and froze trying to encode it, but since it did not crash I think it's likely that the same commit broke both. |
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
retest with master please? #1633 very possibly fixed this. If it did not, I'd like to know. |
@xiphmont Yes, this is fixed now. Thanks! @YaLTeR The reason you get illegal instruction is because there's a bug in rustc with opt-level >= 2 and debug assertions enabled. The compiler optimizes away the code that should handle the panic, and you get an illegal instruction. It's hit me a number of times while debugging my code. We could go back to opt-level = 1, but that's pretty slow, unfortunately. |
This patch set attempts to both improve cooperation between LRF and tiling setup, but also eliminate several illegal configurations. The patch strategy is to consistently eliminate evaluation of 'stretch' superblocks in RDO of LRUs for the loop restoration filter. This primarily affects the case where a tile consists of primarily/exclusively SBs that belong to an LRU stretched from some other tile.
I have the following test file: yuv422p8_input.tar.gz which is a 3-frame clip in YUV422. I have tested the same clip in YUV420 and YUV444 successfully. This YUV422 version crashes rav1e with the following backtrace:
The text was updated successfully, but these errors were encountered: