rpm5 using openat and possibly all the linux *at function family
These two thread on oss-security discuss possible use on openat - and other *at function family - in RPM5.
http://
http://
The second is related to the first but don't mention openat, it is cited only for this here.
As discussed in the threads there would probably not benefit greatly from a point of view of security for rpm5 in using openat(2) and similar *at function family, preferably portably, instead of open and similar IO function. However, if the implementation can be done portably for multiple platforms well i personally don't see too much problem. But the portability is the number one concern.
Blueprint information
- Status:
- Not started
- Approver:
- Jeff Johnson
- Priority:
- Low
- Drafter:
- devzero2000
- Direction:
- Approved
- Assignee:
- Jeff Johnson
- Definition:
- Discussion
- Series goal:
- Accepted for 5.4
- Implementation:
-
Not started
- Milestone target:
- None
- Started by
- Completed by
Related branches
Related bugs
Sprints
Whiteboard
There are very few occurrences where open -> openat needs
to be coded because rpmio already wraps all system calls.
Its easier to use open at than to analyze the threats where
races might exist imho: as stated in the summary, "portability"
is the main concern.
