On 24-11-04 14:58:38, Keith Busch wrote: > On Tue, Nov 05, 2024 at 06:53:32AM +0900, Krzysztof Wilczy´nski wrote: > > Would you have anything against if we put this new bus reset sysfs object > > access behind the following test? > > > > if (!capable(CAP_SYS_ADMIN)) > > return -EPERM; > > > > This is irregardless of what the permissions on the sysfs objects from the > > DAC point of view are set to. > > > > Checking CAP_SYS_ADMIN capability, to improve our default security stance, > > on a number of important sysfs objects (e.g., reset, remove, etc.) we have > > was something I discussed in the past with Bjorn, but never got around to > > sending a patch to add this check. > > > > Thoughts? > > Sure, I'm okay that. We are using DEVICE_ATTR_WO file attribute which > says should make it writable only by an admin, but totally fine with > adding this explicit check here too. Thank you! Depending on whether Bjorn will have any feedback to might prompt a new version of the patches to be sent, if there won't be one, then I will add this extra check directly on the branch. Krzysztof
