Messages in this thread | | | Date | Tue, 28 Oct 2014 08:01:04 +0200 | From | Jarkko Sakkinen <> | Subject | Re: [tpmdd-devel] [PATCH v3 0/3] prepare for TPM 2.0 |
| |
On Mon, Oct 27, 2014 at 04:30:48PM +0000, Scot Doyle wrote: > Hi Jarkko, will you explain why the free_irq call was removed in > tpm_tis_init around line 693?
Because irq is reserved with a devm_request_irq() call.
> I'm wondering if the free_irq call in disable_irq in > https://lkml.org/lkml/2014/9/24/1179 should be removed since it's also > invoked during tpm_tis_init.
Please, read
https://www.kernel.org/doc/Documentation/driver-model/devres.txt
/Jarkko
> On Mon, 27 Oct 2014, Jarkko Sakkinen wrote: > > > Hey, > > > > Any comments about this? In v2 I fixed issues reported by Jason > > and in v3 fixed rather embarrasing one regression. Changes have > > been now also tested to work just fine with the TIS driver. > > > > /Jarkko > > > > On Sun, Oct 26, 2014 at 11:13:54AM +0200, Jarkko Sakkinen wrote: > > > This patch set fixes two race conditions in the TPM subsystem: > > > > > > * Two-phase initialization for struct tpm_chip so that device can > > > initialize fully initialize before exposing itself to the user > > > space. Also, in future TPM2 devices must be flagged before they > > > can be registered. > > > * Machines where there are two TPM devices exposed by ACPI have > > > a racy lookup for the PPI interface. This patch set fixes this > > > issues > > > > > > In addition, transmit_cmd() is renamed as tpm_transmit_cmd() and > > > made opaque so that separate command structure can be introduced > > > for TPM2. > > > > > > Comments about v1: > > > * I think this could be pulled to 3.18 because this clearly fixes > > > bugs in the current implementation. > > > > > > Changes since v1: > > > * Removed dev_dbgs from tpm-chip.c > > > * Use driver_lock for dev_mask (the problem was existing also in > > > tpm-interface.c when the code was moved). > > > * Check for bios_dir nullity when removed (the problem was existing > > > also in tpm-interface.c when the code was moved). > > > * Cleaned up the control flow in tpm_i2c_stm_st33.c and tpm_atmel.c. > > > * Cleaned up the control flow in tpm_ppi.c. > > > > > > Comments about v2: > > > * Thanks to Jason Gunthorpe for reviewing v1. > > > * I'm at T2 infosec conference right now so I wasn't able to test the > > > force paths in tpm_atmel.c and tpm_tis.c. > > > > > > Changes since v2: > > > * Regression in tpm-dev.c, must use tpm_transmit() instead of > > > tpm_transmit_cmd(). > > > > > > Jarkko Sakkinen (3): > > > tpm: merge duplicate transmit_cmd() functions > > > tpm: two-phase chip management functions > > > tpm: fix multiple race conditions in tpm_ppi.c > > > > > > drivers/char/tpm/Makefile | 2 +- > > > drivers/char/tpm/tpm-chip.c | 196 +++++++++++++++++++++++++++++++++++ > > > drivers/char/tpm/tpm-interface.c | 197 +++++------------------------------- > > > drivers/char/tpm/tpm-sysfs.c | 23 +---- > > > drivers/char/tpm/tpm.h | 30 ++++-- > > > drivers/char/tpm/tpm_atmel.c | 12 ++- > > > drivers/char/tpm/tpm_i2c_atmel.c | 33 ++---- > > > drivers/char/tpm/tpm_i2c_infineon.c | 37 ++----- > > > drivers/char/tpm/tpm_i2c_nuvoton.c | 44 +++----- > > > drivers/char/tpm/tpm_i2c_stm_st33.c | 38 +++---- > > > drivers/char/tpm/tpm_ibmvtpm.c | 17 ++-- > > > drivers/char/tpm/tpm_infineon.c | 29 +++--- > > > drivers/char/tpm/tpm_nsc.c | 14 ++- > > > drivers/char/tpm/tpm_ppi.c | 136 +++++++++++++++---------- > > > drivers/char/tpm/tpm_tis.c | 94 ++++++++--------- > > > drivers/char/tpm/xen-tpmfront.c | 14 +-- > > > 16 files changed, 467 insertions(+), 449 deletions(-) > > > create mode 100644 drivers/char/tpm/tpm-chip.c > > > > > > -- > > > 2.1.0 > > > > > > > ------------------------------------------------------------------------------ > > _______________________________________________ > > tpmdd-devel mailing list > > tpmdd-devel@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/tpmdd-devel > >
| |