Hardware video decode (VPU / cedrus) does not support sun9i #24

New issue

Open

opened 2026-08-27 23:22:27 +00:00 by tiagoagueda · 1 comment

tiagoagueda commented

2026-08-27 23:22:27 +00:00

Owner

cedrus.c supports sun4i, sun5i, sun7i, four sun8i variants, sun20i and three sun50i.
sun9i is absent, and the A80's Video Engine is a different generation.

This is a new variant against unfamiliar silicon, not a device-tree change. Worth attempting
only if something actually needs hardware decode on this box.

`cedrus.c` supports sun4i, sun5i, sun7i, four sun8i variants, sun20i and three sun50i. **sun9i is absent**, and the A80's Video Engine is a different generation. This is a new variant against unfamiliar silicon, not a device-tree change. Worth attempting only if something actually needs hardware decode on this box.

tiagoagueda added the

hardware

P4-later

labels

2026-08-27 23:22:27 +00:00

tiagoagueda commented

2026-08-30 14:26:49 +00:00

Author

Owner

Checked against the mirror, and the answer is bleaker than "not supported yet".

No page in the 4095-page wiki associates sun9i with the VPU, Cedrus or the video engine at
all. Not a stub, not a register guide, not a "someone should look at this". Every Cedrus/CedarX
page covers other SoC families.

The A80 page also records that Allwinner escalated their GPL position for this SoC
specifically, on top of the usual libnand/libisp violations. Combined with the vendor kernel
source being gone from every mirror (see ARCHIVE.md), there is no community reverse-engineering
base for sun9i video decode to build on, and no upstream momentum to join.

So this should stay P4-later, and the framing should be honest: this is not "wait for cedrus
to gain sun9i support", it is "nobody has ever started, and the usual starting materials do not
exist for this SoC". If hardware decode is ever actually wanted on this board, the realistic
paths are the vendor Android 4.4 build or software decode - not a driver effort.

Recording that here so the issue is not re-read later as though it were merely blocked on someone
else's roadmap.

From a deep pass over the offline linux-sunxi.org mirror (a80/linux-sunxi.org/wiki-backup, 4095 pages, snapshot 2026-08-29). linux-sunxi.org content is CC BY-SA; the pages named above are the attribution.

Checked against the mirror, and the answer is bleaker than "not supported yet". **No page in the 4095-page wiki associates sun9i with the VPU, Cedrus or the video engine at all.** Not a stub, not a register guide, not a "someone should look at this". Every Cedrus/CedarX page covers other SoC families. The **A80** page also records that Allwinner escalated their GPL position for this SoC specifically, on top of the usual `libnand`/`libisp` violations. Combined with the vendor kernel source being gone from every mirror (see `ARCHIVE.md`), there is no community reverse-engineering base for sun9i video decode to build on, and no upstream momentum to join. So this should stay **P4-later**, and the framing should be honest: this is not "wait for cedrus to gain sun9i support", it is "nobody has ever started, and the usual starting materials do not exist for this SoC". If hardware decode is ever actually wanted on this board, the realistic paths are the vendor Android 4.4 build or software decode - not a driver effort. Recording that here so the issue is not re-read later as though it were merely blocked on someone else's roadmap. --- *From a deep pass over the offline linux-sunxi.org mirror (`a80/linux-sunxi.org/wiki-backup`, 4095 pages, snapshot 2026-08-29). linux-sunxi.org content is CC BY-SA; the pages named above are the attribution.*

tiagoagueda referenced this issue from a commit

2026-08-30 14:27:46 +00:00

Deep-read the linux-sunxi mirror against the tracker (#73 #74 #75)