When debugging the new dpaa2 driver, it is very handy to be able to change the software portal handoff time without recompiling the kernel.
This commit makes the handoff time a sysctl tunable, so it can be changed at runtime.
Details
Details
- Reviewers
dsl - Commits
- rG7e34176e0326: dpaa2: Make software portal holdoff time a sysctl tunable
Diff Detail
Diff Detail
- Repository
- rG FreeBSD src repository
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| sys/dev/dpaa2/dpaa2_io.c | ||
|---|---|---|
| 84 | 1000 is too small taking into account DMA tag per buffer (https://codeberg.org/dsl/freebsd-src/commit/2e198edf2e98bf7c077de9a21853d23c74e61701). I'm going to have the IRQ holdoff at 5000 by default after merging those changes, i.e. 5ms instead of 120us. 10000 would do, I guess. | |
| 229 | Can be "static void", I think. There's no place in the function where actual error is returned. | |
| sys/dev/dpaa2/dpaa2_io.h | ||
| 104 | I'd name it as "irq_holdoff" with /* ms */ given in the following comment. Could you also add /* sysctl(9) */ comment above? It'd clear doubts about this field. | |
| sys/dev/dpaa2/dpaa2_swp.c | ||
| 188 | It might be a good idea to check both "sc" and "sc->swp_desc". | |
| 301 | I'd put it as "from the device context". | |
| sys/dev/dpaa2/dpaa2_swp.c | ||
|---|---|---|
| 188 | swp_desc is not a pointer, but an embedded struct, so it cannot be NULL. | |
| sys/dev/dpaa2/dpaa2_swp.c | ||
|---|---|---|
| 188 | ah, yes, that's right, sorry for the noise | |