| Message ID | 20200828162005.1.I4f67f494c4f759b0e5c7f487e040dfdcf16e0876@changeid (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 160c174ff6972bb56bf48ac3335297889839e1f1 |
| Headers | show |
| Series | ASoC: rt5682: Prefer async probe | expand |
On Sat, Aug 29, 2020 at 7:20 AM Douglas Anderson <dianders@chromium.org> wrote: > > The probe of rt5682 is pretty slow. A quick measurement shows that it > takes ~650 ms on at least one board. There's no reason to block all > other drivers waiting for this probe to finish. Set the flag to allow > other drivers to probe while we're probing. > > Signed-off-by: Douglas Anderson <dianders@chromium.org> > --- > NOTE: I haven't done any analysis of the driver to see _why_ it's so > slow, only that I have measured it to be slow. Someone could > certainly take the time to profile / optimize it, but in any case it > still won't hurt to be async. Hi Doug, thank you for the fix. There are multiple usleep in the probe of rt5682 driver. The major one is a 300 ms sleep after the regulator turns on. There are other sleeps for several tens of ms. > > > This is a very safe flag to turn on since: > > 1. It's not like our probe order was defined by anything anyway. When > we probe is at the whim of when our i2c controller probes and that can > be any time. > > 2. If some other driver needs us then they have to handle the fact > that we might not have probed yet anyway. Agree. soc-core already handled this by returning -EPROBE_DEFER when a component is not found. So the machine driver can probe again. Even in the current behavior, we already see machine driver probe again when the codec driver is not ready, so I think adding this async flag will not affect the machine driver. > > > 3. There may be other drivers probing at the same time as us anyway > because _they_ used async probe. > > While I won't say that it's impossible to tickle a bug by turning on > async probe, I would assert that in almost all cases the bug was > already there and needed to be fixed anyway. > > sound/soc/codecs/rt5682-i2c.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/sound/soc/codecs/rt5682-i2c.c b/sound/soc/codecs/rt5682-i2c.c > index 85aba311bdc8..6b4e0eb30c89 100644 > --- a/sound/soc/codecs/rt5682-i2c.c > +++ b/sound/soc/codecs/rt5682-i2c.c > @@ -294,6 +294,7 @@ static struct i2c_driver rt5682_i2c_driver = { > .name = "rt5682", > .of_match_table = rt5682_of_match, > .acpi_match_table = rt5682_acpi_match, > + .probe_type = PROBE_PREFER_ASYNCHRONOUS, One thing I am wondering is that there has not been any usage in codec driver for this. I think every codec driver can use this, and take the benefit of a possible faster boot time ? > > }, > .probe = rt5682_i2c_probe, > .shutdown = rt5682_i2c_shutdown, > -- > 2.28.0.402.g5ffc5be6b7-goog >
Hi, On Sun, Aug 30, 2020 at 4:05 AM Cheng-yi Chiang <cychiang@chromium.org> wrote: > > > @@ -294,6 +294,7 @@ static struct i2c_driver rt5682_i2c_driver = { > > .name = "rt5682", > > .of_match_table = rt5682_of_match, > > .acpi_match_table = rt5682_acpi_match, > > + .probe_type = PROBE_PREFER_ASYNCHRONOUS, > > > One thing I am wondering is that there has not been any usage in codec > driver for this. > I think every codec driver can use this, and take the benefit of a > possible faster boot time ? One possibility is that they are all enabled as modules instead of builtin to the kernel so nobody ever thought to do it. Modules are always probed asynchronously, so this flag is basically a no-op there (and, in fact, for anything that can be built as a module we have even more certainty that async probe is safe). In the case of the Chrome OS 5.4 tree it's possible this driver should be moved to a module. However, even if we do that my patch is still fine and would be helpful if anyone has a reason to build this driver in. Similar patches could likely be made to other codecs and would similarly speed up boots in cases where codecs were builtin. -Doug
On Fri, 28 Aug 2020 16:20:27 -0700, Douglas Anderson wrote: > The probe of rt5682 is pretty slow. A quick measurement shows that it > takes ~650 ms on at least one board. There's no reason to block all > other drivers waiting for this probe to finish. Set the flag to allow > other drivers to probe while we're probing. Applied to https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound.git for-next Thanks! [1/1] ASoC: rt5682: Prefer async probe commit: 160c174ff6972bb56bf48ac3335297889839e1f1 All being well this means that it will be integrated into the linux-next tree (usually sometime in the next 24 hours) and sent to Linus during the next merge window (or sooner if it is a bug fix), however if problems are discovered then the patch may be dropped or reverted. You may get further e-mails resulting from automated or manual testing and review of the tree, please engage with people reporting problems and send followup patches addressing any issues that are reported if needed. If any updates are required or you are submitting further changes they should be sent as incremental updates against current git, existing patches will not be replaced. Please add any relevant lists and maintainers to the CCs when replying to this mail. Thanks, Mark
diff --git a/sound/soc/codecs/rt5682-i2c.c b/sound/soc/codecs/rt5682-i2c.c index 85aba311bdc8..6b4e0eb30c89 100644 --- a/sound/soc/codecs/rt5682-i2c.c +++ b/sound/soc/codecs/rt5682-i2c.c @@ -294,6 +294,7 @@ static struct i2c_driver rt5682_i2c_driver = { .name = "rt5682", .of_match_table = rt5682_of_match, .acpi_match_table = rt5682_acpi_match, + .probe_type = PROBE_PREFER_ASYNCHRONOUS, }, .probe = rt5682_i2c_probe, .shutdown = rt5682_i2c_shutdown,
The probe of rt5682 is pretty slow. A quick measurement shows that it takes ~650 ms on at least one board. There's no reason to block all other drivers waiting for this probe to finish. Set the flag to allow other drivers to probe while we're probing. Signed-off-by: Douglas Anderson <dianders@chromium.org> --- NOTE: I haven't done any analysis of the driver to see _why_ it's so slow, only that I have measured it to be slow. Someone could certainly take the time to profile / optimize it, but in any case it still won't hurt to be async. This is a very safe flag to turn on since: 1. It's not like our probe order was defined by anything anyway. When we probe is at the whim of when our i2c controller probes and that can be any time. 2. If some other driver needs us then they have to handle the fact that we might not have probed yet anyway. 3. There may be other drivers probing at the same time as us anyway because _they_ used async probe. While I won't say that it's impossible to tickle a bug by turning on async probe, I would assert that in almost all cases the bug was already there and needed to be fixed anyway. sound/soc/codecs/rt5682-i2c.c | 1 + 1 file changed, 1 insertion(+)