| Message ID | 153486702979.13066.16900998092976336647.stgit@warthog.procyon.org.uk (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [01/23] TPM: Add new TPMs to the tail of the list to prevent inadvertent change of dev | expand |
On Tue, Aug 21, 2018 at 04:57:09PM +0100, David Howells wrote: > Provide a platform driver for the user emulator driver. This seems to be > necessary to stop tpm_chip_find_get() from blowing up because it assumes > unconditionally that any device will have a driver attached: > > if (try_module_get(pos->dev->driver->owner)) { > > However, this doesn't then work right because if I remove the TPM device and > re-add it, the tpm ID isn't recycled (ie, /dev/tpm0 becomes unavailable and > the new TPM is /dev/tpm1). > > Signed-off-by: David Howells <dhowells@redhat.com> Again, a duplicate. /Jarkko
diff --git a/drivers/char/tpm/tpm_user_emul.c b/drivers/char/tpm/tpm_user_emul.c index b96350592bca..e5b13358c71e 100644 --- a/drivers/char/tpm/tpm_user_emul.c +++ b/drivers/char/tpm/tpm_user_emul.c @@ -656,17 +656,39 @@ static struct miscdevice tpm_user_dev = { .fops = &tpm_user_fops, }; +static struct platform_driver tpm_user_drv = { + .driver = { + .name = "tpm_user", + .owner = THIS_MODULE, + /* .pm = &tpm_user_pm, -- do we need pm since there's no h/w? */ + }, +}; + /* * Initialise a device */ static __init int tpm_user_mod_init(void) { - return misc_register(&tpm_user_dev); + int ret; + + ret = platform_driver_register(&tpm_user_drv); + if (ret < 0) + return ret; + + ret = misc_register(&tpm_user_dev); + if (ret < 0) + goto error_dev; + return 0; + +error_dev: + platform_driver_unregister(&tpm_user_drv); + return ret; } device_initcall(tpm_user_mod_init); static __exit void tpm_user_mod_exit(void) { misc_deregister(&tpm_user_dev); + platform_driver_unregister(&tpm_user_drv); } module_exit(tpm_user_mod_exit);
Provide a platform driver for the user emulator driver. This seems to be necessary to stop tpm_chip_find_get() from blowing up because it assumes unconditionally that any device will have a driver attached: if (try_module_get(pos->dev->driver->owner)) { However, this doesn't then work right because if I remove the TPM device and re-add it, the tpm ID isn't recycled (ie, /dev/tpm0 becomes unavailable and the new TPM is /dev/tpm1). Signed-off-by: David Howells <dhowells@redhat.com> --- drivers/char/tpm/tpm_user_emul.c | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-)