| Message ID | 20230830-smccc_export-v1-1-6ecc7661bc94@somainline.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | firmware: smccc: Export both soc_id functions | expand |
On 2023-08-30 14:23, Martin Botka wrote: > arm_smccc_get_soc_id_version and arm_smccc_get_soc_id_revision > need to be exported so they can be used by modules. > Currently sun50i cpu freq driver is planning to use these functions. > What for? I'm absolutely not keen on randomly exporting synbols until we see what this is for. Thanks, M.
The patches using this are still WIP but the basic point is that H616 has 2 die revisions (And those have couple of other versions themselves) that require different uV and some frequencies are blacklisted on those versions. The driver is using these functions to read the revision (Not the version. It just looked like the logical thing to export both and Andre confirmed. We can ofc drop the version symbol export if need be) and reads trough the OPP table and selects the supported frequencies and uV for the chip its running on :) I will try to get the cpufreq patches in order in the upcoming days and I will Cc you in those as well + post a link to them in this thread :) Cheers, Martin On Wed, Aug 30 2023 at 03:05:09 PM +01:00:00, Sudeep Holla <sudeep.holla@arm.com> wrote: > > > On Wed, Aug 30, 2023 at 2:24 PM Martin Botka > <martin.botka@somainline.org> wrote: >> arm_smccc_get_soc_id_version and arm_smccc_get_soc_id_revision >> need to be exported so they can be used by modules. >> Currently sun50i cpu freq driver is planning to use these functions. > > I am not against exporting these but please post along with the > patches using these > exported symbols so that no one complains about you exporting the > symbols mentioning > some patches that may or may not appear in the future. > > -- > Regards, > Sudeep
On Wed, Aug 30, 2023 at 04:19:23PM +0200, Martin Botka wrote: > The patches using this are still WIP but the basic point is that H616 has 2 > die revisions (And those have > couple of other versions themselves) that require different uV and some > frequencies are blacklisted on those > versions. > > The driver is using these functions to read the revision (Not the version. > It just looked like the logical > thing to export both and Andre confirmed. We can ofc drop the version symbol > export if need be) and reads > trough the OPP table and selects the supported frequencies and uV for the > chip its running on :) > > I will try to get the cpufreq patches in order in the upcoming days and I > will Cc you in those as well + > post a link to them in this thread :) > For simplicity, just make this patch part of the series that adds cpufreq support on this platform/soc. Cross referencing won't give any more confidence that the user is also getting merged at the same time as this patch.
Got it. Cheers, Martin On Wed, Aug 30 2023 at 04:43:12 PM +01:00:00, Sudeep Holla <sudeep.holla@arm.com> wrote: > On Wed, Aug 30, 2023 at 04:19:23PM +0200, Martin Botka wrote: >> The patches using this are still WIP but the basic point is that >> H616 has 2 >> die revisions (And those have >> couple of other versions themselves) that require different uV and >> some >> frequencies are blacklisted on those >> versions. >> >> The driver is using these functions to read the revision (Not the >> version. >> It just looked like the logical >> thing to export both and Andre confirmed. We can ofc drop the >> version symbol >> export if need be) and reads >> trough the OPP table and selects the supported frequencies and uV >> for the >> chip its running on :) >> >> I will try to get the cpufreq patches in order in the upcoming days >> and I >> will Cc you in those as well + >> post a link to them in this thread :) >> > > For simplicity, just make this patch part of the series that adds > cpufreq > support on this platform/soc. Cross referencing won't give any more > confidence > that the user is also getting merged at the same time as this patch. > > -- > Regards, > Sudeep
diff --git a/drivers/firmware/smccc/smccc.c b/drivers/firmware/smccc/smccc.c index db818f9dcb8e..b4224da46988 100644 --- a/drivers/firmware/smccc/smccc.c +++ b/drivers/firmware/smccc/smccc.c @@ -64,11 +64,13 @@ s32 arm_smccc_get_soc_id_version(void) { return smccc_soc_id_version; } +EXPORT_SYMBOL_GPL(arm_smccc_get_soc_id_version); s32 arm_smccc_get_soc_id_revision(void) { return smccc_soc_id_revision; } +EXPORT_SYMBOL_GPL(arm_smccc_get_soc_id_revision); static int __init smccc_devices_init(void) {
arm_smccc_get_soc_id_version and arm_smccc_get_soc_id_revision need to be exported so they can be used by modules. Currently sun50i cpu freq driver is planning to use these functions. Signed-off-by: Martin Botka <martin.botka@somainline.org> --- drivers/firmware/smccc/smccc.c | 2 ++ 1 file changed, 2 insertions(+) --- base-commit: 706a741595047797872e669b3101429ab8d378ef change-id: 20230830-smccc_export-aa68d8a25539 Best regards,