On Wed, Jun 29, 2022 at 10:16:37AM -0400, Stefan Berger wrote: > > > On 6/28/22 22:45, Jarkko Sakkinen wrote: > > On Tue, Jun 28, 2022 at 09:29:48AM -0400, Stefan Berger wrote: > > > > > > > > > On 6/27/22 18:43, Rob Herring wrote: > > > > On Thu, Jun 16, 2022 at 11:41:28AM -0400, Stefan Berger wrote: > > > > > Simplify tpm_read_log_of() by moving Openfirmware-specific code into > > > > > the Openfirmware driver to make the code reusable. Call the new > > > > > > > > There is no such 'Openfirmware driver'. > > > > > > > > > of_tpm_get_sml_parameters() function from the TPM Openfirmware driver. > > > > > > > > > > Signed-off-by: Stefan Berger <stefanb@xxxxxxxxxxxxx> > > > > > Cc: Jarkko Sakkinen <jarkko@xxxxxxxxxx> > > > > > Cc: Jason Gunthorpe <jgg@xxxxxxxx> > > > > > Cc: Rob Herring <robh+dt@xxxxxxxxxx> > > > > > Cc: Frank Rowand <frowand.list@xxxxxxxxx> > > > > > --- > > > > > drivers/char/tpm/eventlog/of.c | 31 +++++-------------------------- > > > > > drivers/of/Makefile | 2 +- > > > > > drivers/of/device_node.c | 27 +++++++++++++++++++++++++++ > > > > > > > > Humm, definitely the wrong direction. Generally, code for specific > > > > bindings does not go in drivers/of/. There used to be some, but we've > > > > moved it to the appropriate subsystems. kexec was an exception to not > > > > have 2 copies of the same code in arch/. > > > > > > The function I am moving here is called by the TPM subsystem and also now by > > > of/kexec.c. The latter is compiled under the following conditions: > > > > > > ifdef CONFIG_KEXEC_FILE > > > ifdef CONFIG_OF_FLATTREE > > > obj-y += kexec.o > > > endif > > > endif > > > > > > The code that current calls it is compiled under the following conditions: > > > > > > tpm-$(CONFIG_OF) += eventlog/of.o > > > > > > To make it available to both I could keep it in the TPM subsystem like this: > > > > > > obj-$(CONFIG_OF) = tpm_of.o > > > > > > > > > Jarrko, if you read this, any comment? > > > > > > > > > Stefan > > > > Why can't you convert of_tpm_get_sml_parameters() to inline function? > > I can do that and put it into include/linux/tpm.h. The only concern would > have been the size of the function. It is somewhat insignificant amount (dozens of bytes at most). BR, Jarkko
