| Message ID | 20230228-topic-qos-v7-9-815606092fff@linaro.org (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | The great interconnecification fixation | expand |
On 08/03/2023 21:40, Konrad Dybcio wrote: > The interconnect driver is (or soon will be) vital to many other > devices, as it's not a given that the bootloader will set up enough > bandwidth for us or that the values we come into are reasonable. > > Promote the driver to core_initcall to ensure the consumers (i.e. > most "meaningful" parts of the SoC) can probe without deferrals. > > Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org> > --- > drivers/interconnect/qcom/msm8996.c | 12 +++++++++++- > 1 file changed, 11 insertions(+), 1 deletion(-) > > diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c > index 347fe59ec293..1eb51ed18b0b 100644 > --- a/drivers/interconnect/qcom/msm8996.c > +++ b/drivers/interconnect/qcom/msm8996.c > @@ -2109,7 +2109,17 @@ static struct platform_driver qnoc_driver = { > .sync_state = icc_sync_state, > } > }; > -module_platform_driver(qnoc_driver); > +static int __init qnoc_driver_init(void) > +{ > + return platform_driver_register(&qnoc_driver); > +} > +core_initcall(qnoc_driver_init); > + > +static void __exit qnoc_driver_exit(void) > +{ > + platform_driver_unregister(&qnoc_driver); > +} > +module_exit(qnoc_driver_exit); > > MODULE_AUTHOR("Yassine Oudjana <y.oudjana@protonmail.com>"); > MODULE_DESCRIPTION("Qualcomm MSM8996 NoC driver"); > Its probably the right-thing-to-do to have interconnects probe earlier but, then, why restrict that to 8996 only ? --- bod
On 10.03.2023 15:23, Bryan O'Donoghue wrote: > On 08/03/2023 21:40, Konrad Dybcio wrote: >> The interconnect driver is (or soon will be) vital to many other >> devices, as it's not a given that the bootloader will set up enough >> bandwidth for us or that the values we come into are reasonable. >> >> Promote the driver to core_initcall to ensure the consumers (i.e. >> most "meaningful" parts of the SoC) can probe without deferrals. >> >> Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org> >> --- >> drivers/interconnect/qcom/msm8996.c | 12 +++++++++++- >> 1 file changed, 11 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c >> index 347fe59ec293..1eb51ed18b0b 100644 >> --- a/drivers/interconnect/qcom/msm8996.c >> +++ b/drivers/interconnect/qcom/msm8996.c >> @@ -2109,7 +2109,17 @@ static struct platform_driver qnoc_driver = { >> .sync_state = icc_sync_state, >> } >> }; >> -module_platform_driver(qnoc_driver); >> +static int __init qnoc_driver_init(void) >> +{ >> + return platform_driver_register(&qnoc_driver); >> +} >> +core_initcall(qnoc_driver_init); >> + >> +static void __exit qnoc_driver_exit(void) >> +{ >> + platform_driver_unregister(&qnoc_driver); >> +} >> +module_exit(qnoc_driver_exit); >> MODULE_AUTHOR("Yassine Oudjana <y.oudjana@protonmail.com>"); >> MODULE_DESCRIPTION("Qualcomm MSM8996 NoC driver"); >> > > Its probably the right-thing-to-do to have interconnects probe earlier but, then, why restrict that to 8996 only ? To be honest with you, this one caught my attention and it was the one I tested things on.. But yeah, they should all be probing ASAP. Konrad > > --- > bod
diff --git a/drivers/interconnect/qcom/msm8996.c b/drivers/interconnect/qcom/msm8996.c index 347fe59ec293..1eb51ed18b0b 100644 --- a/drivers/interconnect/qcom/msm8996.c +++ b/drivers/interconnect/qcom/msm8996.c @@ -2109,7 +2109,17 @@ static struct platform_driver qnoc_driver = { .sync_state = icc_sync_state, } }; -module_platform_driver(qnoc_driver); +static int __init qnoc_driver_init(void) +{ + return platform_driver_register(&qnoc_driver); +} +core_initcall(qnoc_driver_init); + +static void __exit qnoc_driver_exit(void) +{ + platform_driver_unregister(&qnoc_driver); +} +module_exit(qnoc_driver_exit); MODULE_AUTHOR("Yassine Oudjana <y.oudjana@protonmail.com>"); MODULE_DESCRIPTION("Qualcomm MSM8996 NoC driver");
The interconnect driver is (or soon will be) vital to many other devices, as it's not a given that the bootloader will set up enough bandwidth for us or that the values we come into are reasonable. Promote the driver to core_initcall to ensure the consumers (i.e. most "meaningful" parts of the SoC) can probe without deferrals. Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org> --- drivers/interconnect/qcom/msm8996.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)